[go: up one dir, main page]

WO2020215914A1 - 信息处理方法、信息处理装置和信息处理系统 - Google Patents

信息处理方法、信息处理装置和信息处理系统 Download PDF

Info

Publication number
WO2020215914A1
WO2020215914A1 PCT/CN2020/078742 CN2020078742W WO2020215914A1 WO 2020215914 A1 WO2020215914 A1 WO 2020215914A1 CN 2020078742 W CN2020078742 W CN 2020078742W WO 2020215914 A1 WO2020215914 A1 WO 2020215914A1
Authority
WO
WIPO (PCT)
Prior art keywords
information
user
customer service
reply
intention
Prior art date
Application number
PCT/CN2020/078742
Other languages
English (en)
French (fr)
Inventor
王海
夏波
李茂贵
李涛
郜焱磊
Original Assignee
北京京东尚科信息技术有限公司
北京京东世纪贸易有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京京东尚科信息技术有限公司, 北京京东世纪贸易有限公司 filed Critical 北京京东尚科信息技术有限公司
Priority to US17/600,508 priority Critical patent/US20220129908A1/en
Priority to EP20795968.5A priority patent/EP3961423A4/en
Publication of WO2020215914A1 publication Critical patent/WO2020215914A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/33—Querying
    • G06F16/332—Query formulation
    • G06F16/3329—Natural language query formulation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/01—Customer relationship services
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/01—Customer relationship services
    • G06Q30/015—Providing customer assistance, e.g. assisting a customer within a business location or via helpdesk
    • G06Q30/016—After-sales
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/33—Querying
    • G06F16/332—Query formulation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0281—Customer communication at a business location, e.g. providing product or service information, consulting

Definitions

  • the present disclosure relates to the field of Internet technology, and more specifically, to an information processing method, an information processing device, and an information processing system.
  • the inventor found that there are at least the following problems in the prior art: Since the business of e-commerce involves pre-sales (such as product attribute consultation) and after-sales (such as return and exchange, etc.) and other services, The business processes involved are complex, and it is impossible to enable the customer service to master various businesses through simple training, resulting in the inability to effectively improve the service quality of the customer service.
  • the present disclosure provides an information processing method, an information processing device, and an information processing system that help improve the service quality of customer service.
  • One aspect of the present disclosure provides an information processing method, which is suitable for a server, and the server is respectively connected to a client and a customer service.
  • the method may include the following operations. First, receiving a first message from the client Information, in response to the received first information, perform semantic understanding of the first information to obtain the user's intention, and then determine the business process corresponding to the user's intention, and then, from the first information and the The context information of the first information acquires entity information, and the entity information is related to the business process corresponding to the user's intention, and then, the reply information of the first information is determined based on the business process corresponding to the user's intention and the entity information And sent to the customer service terminal.
  • the response information given combines the first information and the context information. Entity information, the response information given is more in line with the real needs of users, during which there is no need for customer service to search the knowledge base, thereby improving response efficiency and reducing training and service costs.
  • the determining and outputting the reply information of the first information may include the following operations: constructing a node tree for each business process, the root node of the node tree corresponds to a user intention, and then determining The type of the root node in the node tree takes the root node as the current node, and the type of the root node includes one of the following: answer type, rule type, sub-process type, and common reply type.
  • the type of the node is an answer type or a common response type: if the type of the current node is a rule type or a sub-process type, the child node is determined based on at least one of the entity information, the rule, and the sub-process, and the child node is taken as The current node determines the type of the current node, and then obtains the reply information of the current node and outputs the reply information of the current node to the customer service terminal. It should be noted that the process of constructing the node tree may be pre-built, that is, it is not necessary to perform the construction of the node tree every time the reply information is determined. Through the above operations, the reply information corresponding to the entity information provided by the user can be automatically obtained, which is more in line with the actual needs of the user.
  • the obtaining the reply information of the current node and outputting the reply information of the current node to the customer service terminal may include the following operations: On the one hand, if the type of the current node is the answer type, then Obtain and output reply information from the first answer database based on the identifier of the current node. On the other hand, if the type of the current node is a common reply type, the first information is vectorized to obtain a vectorized First information; use the vectorized first information to perform matching in the second answer database to obtain and output at least one answer information.
  • the method may further include the following operations: first, after receiving the first information sent from the client, acquiring historical data of the user of the first information, and then acquiring the user’s The user characteristics of the historical data, then, based on the user characteristics, the user’s prediction intention and/or prediction skills are obtained, and then based on at least one of the user characteristics, the prediction intention and the prediction skills, and the customer service characteristics
  • the current service customer service is determined among multiple customer services, and then the first information is sent to the current service customer service.
  • the user's historical data of the first information includes at least one of the following: last consultation content, recent order, abnormal order, and whether it is a sensitive customer.
  • last consultation content e.g., a service requested by the user's intention
  • abnormal order e.g., a sensitive order
  • whether it is a sensitive customer e.g., a sensitive customer
  • the customer service terminal communicates and connects with a server terminal and a client terminal.
  • the method may include the following operations. First, receive a message sent by the server terminal.
  • the first information and reply information, the reply information is generated by the server based on user intentions, entity information, and business processes, the business processes corresponding to the user intentions of the first information, and the entity information
  • the first information and the context information of the first information are obtained and are related to the business process.
  • the first information and the reply information are displayed, and then, a customer service operation is received, and a response to the customer service operation , Send the reply information or the information input by the customer service to the client.
  • the customer service terminal can reply to the first information sent by the client based on the reply information provided by the server, and the reply information is that the server obtains the corresponding business process by identifying the user's intention, and then based on the entity in the context Information and business process response information, therefore, can assist customer service to complete the response in different business scenarios, without the need for the customer service to understand the complex business, and the response information given is more in line with the real needs of the user, during which there is no need for the customer service to search the knowledge base. Thereby improving the efficiency of response and reducing training and service costs.
  • the method may further include the following operations: after outputting the reply information or the information input by the customer service, at least one of the first information, the reply information, or the information input by the customer service One is sent to the server, so that the server can analyze the customer service characteristics of the current service customer service. This can facilitate automatic acquisition of customer service features and be more objective.
  • the device may include a first receiving module, an intent acquisition module, a business process determining module, an entity information acquiring module, and a reply information determining module, wherein the first receiving module
  • the module is used to receive the first information from the client
  • the intention acquisition module is used to respond to the received first information, perform semantic understanding of the first information to obtain the user's intention
  • the business process is determined
  • the module is used to determine the business process corresponding to the user's intention
  • the entity information obtaining module is used to obtain the entity information from the first information and the context information of the first information
  • the entity information corresponds to the user's intention
  • the response information determination module is used to determine the response information of the first information based on the business process corresponding to the user's intention and the entity information and send it to the customer service terminal.
  • the reply information determination module may include: a node type determination submodule, a recursive submodule, and an information acquisition submodule, wherein the node type determination submodule is used to determine the root node in the node tree. Type and use the root node as the current node.
  • the root node type includes one of the following: answer type, rule type, sub-process type, and common reply type.
  • the recursive sub-module is used to repeat the following steps until the current node
  • the type is an answer type or a common answer type: if the type of the current node is a rule type or a sub-process type, the child node is determined based on at least one of the entity information, rule, and sub-process, and the child node is taken as the current node , The type of the current node is determined, and the information acquisition submodule is used to acquire the reply information of the current node and output the reply information of the current node to the customer service terminal.
  • the reply information determining module may further include a node tree construction sub-module for constructing a node tree for each business process, and the root node of the node tree corresponds to a user intention.
  • the information acquisition submodule may include: a first acquisition unit and a second acquisition unit, wherein the first acquisition unit is configured to, if the type of the current node is an answer type, base The identifier of the current node obtains and outputs reply information from the first answer database, and the second obtaining unit is configured to vectorize the first information if the type of the current node is a common reply type to obtain a vector Use the vectorized first information to perform matching in the second answer database to obtain and output at least one reply information.
  • the device may further include a historical data acquisition module, a user feature acquisition module, a prediction module, a matching module, and a routing module, wherein the historical data acquisition module is used to receive the first sent from the client After one piece of information, the historical data of the user of the first information is obtained, the user characteristic obtaining module is used to obtain the user characteristic of the historical data of the user, and the prediction module is used to obtain a prediction of the user based on the user characteristic Intentions and/or prediction skills, the matching module is configured to determine the current service customer service from multiple customer services based on at least one of the user characteristics, the predicted intention and the prediction skills, and the customer service characteristics, the routing module Used to send the first information to the current service customer service.
  • the historical data acquisition module is used to receive the first sent from the client After one piece of information, the historical data of the user of the first information is obtained, the user characteristic obtaining module is used to obtain the user characteristic of the historical data of the user, and the prediction module is used to obtain a prediction of the user based on the user
  • the user's historical data of the first information includes at least one of the following: last consultation content, recent order, abnormal order, and whether it is a sensitive customer.
  • the device may include a second receiving module, a display module, an operation receiving module, and an information output module, wherein the second receiving module is used to receive the server The first information and reply information sent, the reply information is generated by the server based on the user's intention, entity information, and business process, the business process corresponds to the user's intention of the first information, the entity information Obtained from the first information and the context information of the first information, and related to the business process, the display module is used to display the first information and the reply information, and the operation receiving module uses Upon receiving the customer service operation, the information output module is configured to send the reply information or the information input by the customer service to the client in response to the customer service operation.
  • the device may further include a customer service information acquisition module configured to combine the first information and the reply information after outputting the reply information or the information input by the customer service Or at least one of the information input by the customer service is sent to the server, so that the server can analyze the customer service characteristics of the current customer service.
  • a customer service information acquisition module configured to combine the first information and the reply information after outputting the reply information or the information input by the customer service Or at least one of the information input by the customer service is sent to the server, so that the server can analyze the customer service characteristics of the current customer service.
  • the system may include a dialogue management module, an intent recognition module, a business model module, and an answer module, wherein the dialogue management module is used to obtain first information, the The context information and entity information of the first information, and the response information outputting the first information, the intention recognition module is configured to obtain user intentions based at least on the first information, and the business model module is configured to obtain user intentions based at least on the first information.
  • the information and the user's intention determine a business process, and the answer module is used to determine the answer information of the first information based on the business process and the entity information.
  • the system may further include an intelligent scheduling module, and the scheduling only module is used to determine the matching relationship between the intention and the skill based on the user's historical consultation data through the user's intention and the content of the customer service service, and The matching relationship determines a customer service suitable for the user.
  • the system may further include a routing module, which is used to establish a connection between the client and the customer service, and deliver information between the server, the client and the customer service.
  • a routing module which is used to establish a connection between the client and the customer service, and deliver information between the server, the client and the customer service.
  • Another aspect of the present disclosure provides a computer-readable storage medium storing computer-executable instructions, which are used to implement the above-mentioned method when executed.
  • Another aspect of the present disclosure provides a computer program that includes computer-executable instructions, which are used to implement the above-mentioned method when executed.
  • FIG. 1A schematically shows an application scenario of an information processing method, an information processing device, and an information processing system according to an embodiment of the present disclosure
  • FIG. 1B schematically shows a schematic diagram of a system architecture according to an embodiment of the present disclosure
  • FIG. 2A schematically shows a flowchart of an information processing method used on a server side according to an embodiment of the present disclosure
  • FIG. 2B schematically shows an information flow diagram of an information processing method used on the server side according to an embodiment of the present disclosure
  • FIG. 3A schematically shows a flowchart of determining and outputting reply information of the first information according to an embodiment of the present disclosure
  • Fig. 3B schematically shows a flowchart of obtaining reply information of the current node according to an embodiment of the present disclosure
  • Fig. 4 schematically shows a flowchart of an information processing method used on the server side according to another embodiment of the present disclosure
  • FIG. 5A schematically shows a flowchart of an information processing method for a customer service terminal according to an embodiment of the present disclosure
  • FIG. 5B schematically shows a flowchart of an information processing method for a customer service terminal according to another embodiment of the present disclosure
  • Fig. 6A schematically shows a block diagram of an information processing apparatus according to an embodiment of the present disclosure
  • Fig. 6B schematically shows a block diagram of an information processing apparatus according to another embodiment of the present disclosure.
  • Fig. 7A schematically shows a block diagram of an information processing system according to an embodiment of the present disclosure
  • Fig. 7B schematically shows a schematic diagram of an information flow of an information processing system according to an embodiment of the present disclosure
  • FIG. 7C schematically shows a logic diagram of a dialog management module according to an embodiment of the present disclosure
  • FIG. 7D schematically shows a logic diagram of an intention recognition module according to an embodiment of the present disclosure
  • FIG. 7E schematically shows a logical schematic diagram of a business model module according to an embodiment of the present disclosure
  • Fig. 7F schematically shows a logic diagram of an answer module according to an embodiment of the present disclosure
  • FIG. 7G schematically shows a logic diagram of an intelligent scheduling module according to an embodiment of the present disclosure.
  • Fig. 8 schematically shows a block diagram of a computer system suitable for implementing information processing according to an embodiment of the present disclosure.
  • At least one of the “systems” shall include but not limited to systems having A alone, B alone, C alone, A and B, A and C, B and C, and/or systems having A, B, C, etc. ).
  • At least one of the “systems” shall include but not limited to systems having A alone, B alone, C alone, A and B, A and C, B and C, and/or systems having A, B, C, etc. ).
  • the robot supported by the server needs to understand the information sent by the customer and give the reply information, and finally send the reply information to the client through the customer service or the robot.
  • the robot As an auxiliary type of robot, all reply information is sent to the client after a manual click.
  • the robot mainly realizes auxiliary responses in the process.
  • the mainstream customer service response assistant tools are mainly divided into two ways, the first is the recommendation of answer words, and the second is the official word search.
  • verbal recommendation the customer's intention is recognized during the process of speaking, and the answer is matched from the answer database through the machine learning model, recommended to the customer service, and sent to the customer after the customer clicks.
  • the language search a knowledge base is established, and the customer service understands what the customer says, and then the customer service searches the knowledge base for related words, clicks and sends it to the customer.
  • the answers recommended by the existing words are static answers, and the process is fixed and cannot be flexibly configured.
  • the customer asks: When will my XX order arrive?
  • the recommended answer at this time may be: Hello, I'm very sorry, your order is still being delivered, please wait patiently.
  • the customer is not clearly informed of the answer they want, but simple reassurance, which cannot really meet the customer's needs, and reduces the usability of the system.
  • the answer to the existing verbal search is a fixed answer, which requires manual understanding of complex business. For example, the customer asks: Can I return this product? At this time, the customer service may need to understand that the customer is thinking of returning the goods, and search the knowledge base: return process. In complex scenarios, there may be the return process when the order status is not received, and the return process when the order status is received. It is more complicated and requires repeated confirmation by customer service and constant search, which greatly reduces the reception timeliness.
  • the embodiments of the present disclosure provide an information processing method, an information processing device, and an information processing system.
  • the method includes a business process determination process and a response information determination process.
  • determining the business process first determine the user's intention of the received first information, and then determine the business process corresponding to the user's intention.
  • enter the response information determination process enter the response information determination process, and determine the response information of the first information based on the business process and the entity information included in the first information.
  • the embodiments of the present disclosure can dynamically allocate customer service through the service skill tags and resources of the customer service, and configure the process and answers through a small number of professionals who understand the specific business, so that standardized answers can be recommended to the customer service according to the business.
  • the answer is sent to the client.
  • FIG. 1A schematically shows an application scenario of an information processing method, an information processing apparatus, and an information processing system according to an embodiment of the present disclosure.
  • FIG. 1A is only an example of application scenarios in which the embodiments of the present disclosure can be applied to help those skilled in the art understand the technical content of the present disclosure, but it does not mean that the embodiments of the present disclosure cannot be used for other applications.
  • the server side can perform semantic understanding of the question sent by the user to obtain the user's intention, such as electronic equipment maintenance, determine the corresponding business process, and then determine the reply information based on the entity information related to the business process,
  • the entity information can be extracted from the context, so that the answer that the user wants can be obtained based on the business process and entity information: call the maintenance staff at 186xxxxxxxx, and push it to the customer service, and send it to the user when the customer service confirms that there is no problem .
  • FIG. 1B schematically shows a schematic diagram of a system architecture according to an embodiment of the present disclosure.
  • the system architecture 100 may include terminal devices 101, 102, 103, a network 104, and a server 105.
  • the network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105.
  • the network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables.
  • the user can use the terminal devices 101, 102, 103 to interact with the server 105 through the network 104 to receive or send messages and so on.
  • Various communication client applications may be installed on the terminal devices 101, 102, 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only examples).
  • the terminal devices 101, 102, 103 may be various electronic devices with a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, Internet TVs, and so on.
  • the server 105 may be a server that provides various services, for example, a background management server (only an example) that replies to information sent by users using the terminal devices 101, 102, and 103.
  • the background management server may analyze and process the received user request and other data, and feed back the processing result (for example, webpage, information, or data obtained or generated according to the user request) to the terminal device.
  • terminal devices 101, 102, 103 can be used as clients or customer service terminals, and the auxiliary robots allocated by the server can be used on the customer service terminals.
  • terminal devices are merely illustrative. According to implementation needs, there can be any number of terminal devices, networks and servers.
  • Fig. 2A schematically shows a flowchart of an information processing method used on a server side according to an embodiment of the present disclosure. This method is applicable to the server side, and the server side communicates with the client and the customer service side respectively.
  • the method includes operations S201 to S209.
  • the first information may be consultation information, small chat information, etc. sent by the user to the customer service.
  • the user may ask many questions, and the customer service or robot will also give multiple responses.
  • the multiple issues mentioned above may involve multiple intents. For example, a customer may ask first in a conversation: why my goods have not arrived yet, the corresponding user intent may be: to remind the order. Then the customer asked again: Not yet, let me change a product. The user's intention at this time may be: exchange goods.
  • the first information may be semantically understood based on a pre-trained semantic procedural model to obtain user intentions.
  • the semantic understanding model includes but is not limited to: multi-class neural network, deep neural network, volume and neural network, adversarial neural network, long and short-term memory neural network, etc.
  • the first information sent by the customer may be recognized as multiple user intentions. For example, the customer sent the information: "What about the invoice”.
  • the format of returning user intentions can be as follows: [ ⁇ make up invoice, 0.9999 ⁇ , ⁇ modify invoice, 0.1999 ⁇ , ⁇ do not require invoice, 0.0099 ⁇ ], multiple user intentions will be returned, and the number is the probability of this user intention.
  • a threshold is set, such as 0.8. When the probability of the user's intention is greater than 0.8, the user's intention is returned.
  • business processes can configure processes and answers for professionals who understand specific business processes, where each business process can correspond to a user's intention.
  • entity information is obtained from the first information and the context information of the first information, where the entity information is related to the business process corresponding to the user's intention.
  • the exchange business Since a business process usually requires the support of multiple entity information, for example, the exchange business requires the user's receiving address, recipient, recipient contact information, receiving time and other physical information to ensure the smooth progress of the business
  • these entity information may exist in the previous chat records, or the customer service may need to guide the customer to obtain it. Therefore, it needs to be obtained from the first information and the context of the first information Information acquisition entity information.
  • the entity information required by the business process can be sent from the server to the customer service to prompt the customer service.
  • the reply information of the first information is determined based on the business process corresponding to the user's intention and the entity information and sent to the customer service terminal.
  • the server may send accurate reply information to the customer service end, or it may be multiple possible reply information sent by the server to the customer service end, and then the customer service selects the best reply information from the multiple possible reply information. The appropriate reply information is then sent to the client.
  • Fig. 2B schematically shows an information flow diagram of a server-side information processing method according to an embodiment of the present disclosure.
  • the client initiates a session and establishes a connection with the customer service client.
  • the routing module is responsible for the information interaction between the three ends: the dialogue message is delivered to the customer service end for display.
  • the session management module recognizes the intent of the first message sent by the client, and if the user's intent is recognized, it sends an event message to the customer service terminal, highlighting the user's intention recognized by the customer service.
  • the customer service clicks the first message or receives the user's intention the auxiliary robot pops up.
  • the auxiliary robot obtains the reply information according to the first information, the context information of the first information and the rules corresponding to the user's intention (such as historical information and information given by the customer service guide user), and returns the reply information to the customer service terminal, and the customer service enters the reply information or enters it by himself
  • the information is sent to the client.
  • Fig. 3A schematically shows a flowchart of determining and outputting reply information of the first information according to an embodiment of the present disclosure.
  • the determining and outputting the reply information of the first information may include operations S301 to S305.
  • the type of the root node in the node tree corresponding to the user's intention is determined and the root node is taken as the current node.
  • the type of the root node includes one of the following: answer type, rule type, sub-process type, and common reply type.
  • the node tree may be constructed based on the business process, for example, the node tree is constructed according to the business process in advance, or the node tree is constructed according to the business process when the reply information of the first message needs to be determined, or When it is found that the business process involved does not have a corresponding node tree, the node tree is constructed according to the business process.
  • the root node of the node tree corresponds to a user's intention, that is, the corresponding node tree is determined based on the user's intention, and the type of the root node of the node tree is determined.
  • the correspondence between the user's intention and the business process is confirmed in advance.
  • the auxiliary robot assigned to a certain customer service only involves the business process of mobile phones and small chats, and will not answer computer business process questions.
  • the business process corresponding to the user's intention that the customer service can handle may only be: mobile phone consultation, mobile phone return, mobile phone replacement, etc.
  • the business process required by the customer can be confirmed, and the node tree corresponding to the business process can be obtained.
  • the node tree is a configuration. After entering a certain node tree, the node tree is executed.
  • the following steps are repeated until the type of the current node is an answer type or a common response type: if the type of the current node is a rule type or a sub-process type, at least based on the entity information, rules, and sub-processes One way to determine the child node, and the child node as the current node; to determine the type of the current node.
  • each user's intention has a unique value in the node tree, which is used to identify the root node and determine whether it is necessary to enter the root node of the new node tree (for example, the client sends a new message, whether the new message has new User intention), if yes, enter the root node of the node tree corresponding to the new user intention, if not, locate the child node of the node tree corresponding to the user intention of the last conversation, and determine the node type of the child node
  • node Types are divided into the following categories: answer type, rule type, sub-process type (such as general rule type, reusable), and common answer type (such as Frequently Asked Questions (FAQ) type).
  • operation S305 obtain the reply information of the current node and output the reply information of the current node to the customer service terminal.
  • the final child node is the answer type or the common reply type.
  • the reply information corresponding to the child node of the answer type or the polite reply term corresponding to the common reply type can be called as the first Reply to the message.
  • the node tree corresponding to the small chat will be executed later. For example, the user sends a message: Have you eaten? The server will give a reply message: It's only xx, I haven't eaten yet. For another example, if it is the user's intention to return a product from a mobile phone, the server searches for the response information in the answer database based on the identifier of the node of the node tree corresponding to the user's intention of the mobile phone to return the product. Tian has no reason to return. Among them, the reply information in the answer database can be manually configured.
  • reply information can be combined with entity information or not, but the entity information during the conversation will be passed down to the answer layer.
  • the server side may use the entity [shipping address]. If the server side needs to use the entity [Receiving Address] in the subsequent replacement business process, at this time, the server side can send a message: Please confirm your receiving address is xxxx, if yes, click [Yes ], if not, click [No], which can effectively improve communication efficiency.
  • Fig. 3B schematically shows a flow chart of obtaining reply information of the current node according to an embodiment of the present disclosure.
  • the obtaining the reply information of the current node and outputting the reply information of the current node to the customer service terminal may include operations S311 to S313.
  • the answer information is obtained from the first answer database based on the identifier of the current node and output.
  • the answer service model For example, directly call the answer service model. Specifically, it is determined whether the type of the current node is the answer type or the FAQ type through the context of the first information. If it is the answer type, the final answer information is obtained from the cache according to the node identifier, which is returned by the child node of the node tree corresponding to the service type. Among them, the reply information is stored in the persistent answer layer after manual editing and review, and each node identifier corresponds to a reply message.
  • Fig. 4 schematically shows a flowchart of an information processing method used on the server side according to another embodiment of the present disclosure.
  • the method may further include operations S401 to S409.
  • the user of each client usually has a user ID. After the user uses the user ID to log in to the shopping application, and uses the online customer service tool of the shopping application for question consultation, the user's historical data can be obtained through the user ID.
  • the historical data of the user of the first information includes at least one of the following: last consultation content, recent order, abnormal order, whether it is a sensitive customer, and so on.
  • the user's prediction intention and/or prediction skills are acquired based on the user characteristics.
  • the communication between the user and the customer service may be related to the mobile phone, for example, mobile phone return, mobile phone replacement, mobile phone repair, etc.
  • the user's tolerance is low. For example, he has complained many times. Therefore, the user needs a more proficient and patient customer service to provide services.
  • a current service customer service is determined from multiple customer services based on at least one of the user characteristics, the prediction intention, and the prediction skills, and the customer service characteristics.
  • At least one of the user characteristics, the prediction intention, and the prediction skills is matched with the customer service characteristics corresponding to each customer service stored in the database to obtain a suitable customer service.
  • customer service features can be data stored in the server.
  • Customer service features are mainly composed of customer service portraits, for example, customer service’s historical reception user resolution rate, historical reception user emotion change curve, historical satisfaction, and skill group tags that are good at answering, etc. .
  • the matching may include a preliminary screening process and a screening process, for example, automatically constructing a skill intention model based on the intention and skill pair in advance.
  • the intention and skill pair is obtained based on the user's intention and the service content of the customer service, and the user's intention is obtained from the historical consultation information that the customer service has served the user.
  • preliminary screening is performed based on the user characteristics, customer service characteristics, and the skill intention model of the corresponding user based on the first information to obtain at least one pair of matching options.
  • the at least one set of pairing options is screened again to obtain a current customer service.
  • the characteristic algorithm model may determine the current service customer service based on the currently idle customer service, the matching degree ranking of the at least one set of pairing options, and the like.
  • the first information is sent to the current service customer service.
  • user characteristics are mainly composed of user portraits, including historical consultation intentions, basic personal information, and so on.
  • the server can select some consultation-related features from the user’s historical database, such as the last consultation content, recent orders, abnormal orders, whether sensitive customers, etc., and then predict the following information through the algorithm model of skills and intentions: First determine the customer Questions and intentions that may be consulted, and then determine what skills the customer service may need to respond to the problem, filter out multiple groups of qualified reception customers, and then use the characteristics of the customer service characteristics, user characteristics, prediction intent and prediction skills, etc.
  • the model determines the current service customer service, and then establishes a network link between the customer and the customer service. After the link is established, the customer and the customer service can send information.
  • Fig. 5A schematically shows a flowchart of an information processing method for a customer service terminal according to an embodiment of the present disclosure.
  • the customer service terminal communicates with the server terminal and the client terminal respectively.
  • the method may include operations S501 to S507. It should be noted that, regarding the process of the server determining the reply information of the first information sent by the user through the client, reference may be made to the content disclosed above, which will not be repeated here.
  • first information and reply information sent by the server are received.
  • the reply information is generated by the server based on user intentions, entity information, and business processes.
  • the entity information is obtained from the first information and the context information of the first information, and is related to the business process.
  • the detected user intention may be marked by the server and sent to the customer service terminal, for example, the user intention is displayed in a highlighted manner.
  • the user's intention after the customer service clicks on the first message, and then, after the customer service clicks on a certain user's intention, the reply information corresponding to the user's intention is displayed.
  • the customer service operation may be response information sent by the server, or information input by the customer service, or an information request sent by the server.
  • the information processing method provided by the present disclosure configures business processes and reply information through a small number of professionals who understand specific businesses, and realizes standardized answers to business recommendations, which can effectively reduce training costs, thereby realizing reasonable distribution of traffic, unification of processes, and Standardization of response and improvement of service efficiency.
  • Fig. 5B schematically shows a flowchart of an information processing method for a customer service terminal according to another embodiment of the present disclosure.
  • the method may further include operation S509.
  • the customer service features are mainly composed of customer service portraits and are extracted based on the historical service information of the customer service.
  • customer service s historical reception user resolution rate, historical reception user emotion change curve, historical satisfaction, and skill set tags that are good at answering.
  • the server-side analysis obtains the customer service characteristics of the current service customer service
  • the customer service characteristics can be stored in the customer service personal information database, so that the server can call the customer service characteristics.
  • the information processing method provided in the present disclosure obtains customer service characteristics by analyzing the historical service information of customer service, actively mining intent and skill tag pairing, and automatically modeling the user and customer service, so that it is convenient for the server to receive the first information sent by the client
  • customer service characteristics by analyzing the historical service information of customer service, actively mining intent and skill tag pairing, and automatically modeling the user and customer service, so that it is convenient for the server to receive the first information sent by the client
  • rules and algorithm predictions different users can be assigned to different customer services to ensure the timeliness and satisfaction of reception.
  • Fig. 6A schematically shows a block diagram of an information processing apparatus according to an embodiment of the present disclosure.
  • the information processing device 610 may include a first receiving module 611, an intention acquisition module 613, a business process determination module 615, an entity information acquisition module 617, and a reply information determination module 619.
  • the first receiving module 611 is configured to receive first information from the client.
  • the intention obtaining module 613 is configured to perform a semantic understanding of the first information in response to the received first information to obtain the user's intention.
  • the business process determining module 615 is used to determine the business process corresponding to the user's intention.
  • the entity information obtaining module 617 is configured to obtain entity information from the first information and context information of the first information, where the entity information is related to the business process corresponding to the user's intention.
  • the reply information determining module 619 is configured to determine the reply information of the first information based on the business process corresponding to the user's intention and the entity information and send it to the customer service terminal.
  • the reply information determining module 619 may include: a node type determining submodule, a recursive submodule, and an information acquisition submodule.
  • the node type determining submodule is used to determine the type of the root node in the node tree and use the root node as the current node.
  • the type of the root node includes one of the following: answer type, rule type, sub-process type, and common Reply type.
  • the recursive sub-module is used to repeat the following steps until the type of the current node is an answer type or a common reply type: if the type of the current node is a rule type or a sub-process type, based on the entity information, rules, and sub-process types At least one of the steps in the process determines the child node, and uses the child node as the current node to determine the type of the current node.
  • the information acquisition submodule is used to acquire the reply information of the current node and output the reply information of the current node to Kefuduan.
  • the reply information determining module 619 may further include a node tree construction sub-module, which is used to construct a node tree for each business process, and the root node of the node tree corresponds to a node tree.
  • a node tree construction sub-module which is used to construct a node tree for each business process, and the root node of the node tree corresponds to a node tree.
  • the information acquisition sub-module may include: a first acquisition unit and a second acquisition unit.
  • the first obtaining unit is configured to obtain and output reply information from a first answer database based on the identifier of the current node if the type of the current node is an answer type.
  • the second acquiring unit is configured to, if the type of the current node is a common reply type, vectorize the first information to obtain the vectorized first information, and use the vectorized first information in the first Second, perform matching in the answer database to obtain and output at least one answer message.
  • the device 610 may further include a historical data acquisition module 621, a user characteristic acquisition module 623, a prediction module 625, a matching module 627, and a routing module 629.
  • the historical data obtaining module 621 is configured to obtain the historical data of the user of the first information after receiving the first information sent from the client.
  • the user characteristic acquisition module 623 is configured to acquire the user characteristic of the historical data of the user.
  • the prediction module 625 is configured to obtain the user's prediction intention and/or prediction skills based on the user characteristics.
  • the matching module 627 is configured to determine a current service customer service from multiple customer services based on at least one of the user characteristics, the prediction intention, and the prediction skills, and customer service characteristics.
  • the routing module 629 is configured to send the first information to the current service customer service.
  • the user's historical data of the first information includes at least one of the following: last consultation content, recent order, abnormal order, and whether it is a sensitive customer.
  • Fig. 6B schematically shows a block diagram of an information processing apparatus according to another embodiment of the present disclosure.
  • the device 630 may include a second receiving module 631, a display module 633, an operation receiving module 635, and an information output module 637.
  • the second receiving module 631 is configured to receive the first information and reply information sent by the server, and the reply information is generated by the server based on user intentions, entity information, and business processes.
  • the entity information is obtained from the first information and the context information of the first information, and is related to the business process.
  • the display module 633 is used to display the first information and the reply information.
  • the operation receiving module 635 is used for receiving customer service operations.
  • the information output module 637 is configured to send the reply information or the information input by the customer service to the client in response to the customer service operation.
  • the device 630 may further include a customer service information acquisition module 639, which is configured to output the reply information or the information input by the customer service, and then combine the first information and the reply At least one of the information or the information input by the customer service is sent to the server, so that the server can analyze the customer service characteristics of the current customer service.
  • a customer service information acquisition module 639 which is configured to output the reply information or the information input by the customer service, and then combine the first information and the reply At least one of the information or the information input by the customer service is sent to the server, so that the server can analyze the customer service characteristics of the current customer service.
  • the operations performed by the devices 610 and 630 shown in FIGS. 6A to 6B may refer to the corresponding content in the information processing method described above.
  • Fig. 7A schematically shows a block diagram of an information processing system according to an embodiment of the present disclosure.
  • the system 700 may include a dialogue management module 710, an intention recognition module 730, a business model module 750, and an answer module 770.
  • the dialog management module 710 is configured to obtain first information, context information and entity information of the first information, and output reply information of the first information.
  • Fig. 7B schematically shows a logic diagram of a dialog management module according to an embodiment of the present disclosure.
  • the customer or the customer service speaks (single word), and the request (Query) is initiated.
  • the information includes the chat content, the entity information such as the inquiry order invoice, and other input information.
  • the input parameters are parsed, and the pre-logical processing is performed, for example, to determine whether the request is legal, the user's black and white list, and so on.
  • the context can be divided into three types, for example, robot context, dialogue context, and request context, which are mainly information storage and transmission. Then, update the context information.
  • the intent recognition process is to extract the entity information required by the intent form through rules and models, such as address, order number, mobile phone number, etc.
  • the intent form is a task form.
  • the ⁇ entity ⁇ of the ⁇ intention form ⁇ of [re-invoicing] may include: ⁇ company name ⁇ company tax number ⁇ invoice receiving address ⁇ , etc., which are formulated according to specific businesses.
  • the reply message is routed to the client.
  • the intention recognition module 730 is configured to obtain user intentions based on at least the first information.
  • Fig. 7C schematically shows a logic diagram of an intention recognition module according to an embodiment of the present disclosure.
  • data cleaning is performed, including but not limited to: word segmentation, stop word filtering, etc., and then an intent recognition model is requested. If the intent recognition model is not loaded, the intent recognition model needs to be loaded through the loader, and based on the loaded intent recognition model, it is determined whether there is user intent. If it is, the user intent set is returned according to the preset threshold. Then, by sorting the set of user intentions, the final user intentions are filtered out. If the user's intention has been identified or there is no user's intention, it is considered that the sentence may contain other entity information, and the entity information is extracted through the entity model.
  • reply information can also be returned to the conversation management system.
  • the feedback information is used to automatically optimize the intention recognition model.
  • the business model module 750 is configured to determine a business process based at least on the first information and the user intention.
  • Fig. 7D schematically shows a logical schematic diagram of a business model module according to an embodiment of the present disclosure.
  • each intent has a unique value in the node tree, which is used to identify the node, where the intent node is also called the root node. It should be noted that each time the first message sent by the client is received, it is necessary to determine whether to enter a new intent node (that is, whether to enter the root node of the node tree corresponding to the new business process), and if so, enter The new root node, if not, locate the child node of the previous node tree of the first information.
  • the node types are divided into the following categories: answer type, rule type, sub-process type (general rule type, reusable), FAQ type. Then, if it is an answer type, call the answer service model directly. If it is a rule or sub-process type, the node that hits is judged according to the rule, and then the node ID is returned. This process is a recursive process. If it is not the above three types, it is a FAQ type, and the answer service model is directly called. FAQ is generally a polite reply or a recommended answer.
  • the answer module 770 is configured to determine the answer information of the first information based on the business process and the entity information.
  • Fig. 7E schematically shows a logic diagram of an answer module according to an embodiment of the present disclosure.
  • the answer type or FAQ type is judged by the context of the reply message. If it is the answer type, the final answer information is obtained from the cache according to the node ID, and the node ID is returned by the child node of the business model. Among them, the reply information is stored in the persistent answer layer after manual editing and review, and each node ID corresponds to an answer. If it is not an answer type, it is a FAQ type.
  • the content of the first information is vectorized, and the FAQ model is used to make (question-answer pair) text similarity judgments, and TOP N pieces of candidate answer information that are greater than the preset threshold are retrieved, and then the answer model is used to perform the candidate answer data Reorder, filter out one or more reply messages according to the threshold, and route them to the customer service terminal.
  • system 700 may further include an intelligent scheduling module 790, which is used to establish a connection between the client and the customer service terminal.
  • an intelligent scheduling module 790 which is used to establish a connection between the client and the customer service terminal.
  • Fig. 7F schematically shows a logical schematic diagram of an intelligent scheduling module according to an embodiment of the present disclosure.
  • the skill intention model is used to predict the questions that the user may consult and predict the user's intention. According to the problem and predict the customer's intention, what skills may be required to screen out multiple groups of qualified reception customers. Then, the current service customer service is determined by the feature model based on customer service characteristics, user characteristics, predicted user intentions, and required skills. In this way, a network link between the user and the customer service can be established. After the link is established, the user and the customer service can send information.
  • Fig. 7G schematically shows a schematic diagram of an information flow of an information processing system according to an embodiment of the present disclosure.
  • the information processing system may include a front-end interaction part, a routing and intelligent wiring part, and a server determining reply information part.
  • the front-end interaction part includes the client, customer service and auxiliary robots.
  • the routing and intelligent wiring part includes the intelligent scheduling module and the routing module.
  • the server determines the reply information part including the intention recognition module, the dialogue management module, the business model module and the answer module.
  • the routing module delivers messages between the client and the customer service, and can identify the user's intention contained in the message delivered to the customer service, such as highlighting. In addition, the routing module establishes a link between the client and the customer service based on the customer service assigned by the intelligent scheduling module.
  • the routing module receives the first information, it sends the first information and the context information of the first information to the dialogue management module, and the dialogue management module calls the intention recognition module to obtain the user's intention based at least on the first information, and sends the user's intention to the service
  • the identification module the business identification module determines the node tree of the business process corresponding to the user's intention based on rule logic and so on.
  • the answer module is used to determine the answer information of the first information based on the node tree and the entity information.
  • any number of modules, submodules, units, and subunits, or at least part of the functions of any number of them, may be implemented in one module. Any one or more of the modules, sub-modules, units, and sub-units according to the embodiments of the present disclosure may be split into multiple modules for implementation.
  • any one or more of the modules, sub-modules, units, and sub-units according to the embodiments of the present disclosure may be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), System-on-chip, system-on-substrate, system-on-package, application-specific integrated circuit (ASIC), or hardware or firmware in any other reasonable way that integrates or encapsulates the circuit, or can be implemented by software, hardware, and firmware. Any one of these implementations or an appropriate combination of any of them can be implemented.
  • FPGA field programmable gate array
  • PLA programmable logic array
  • ASIC application-specific integrated circuit
  • any one of these implementations or an appropriate combination of any of them can be implemented.
  • one or more of the modules, sub-modules, units, and sub-units according to the embodiments of the present disclosure may be at least partially implemented as a computer program module, and the computer program module may perform corresponding functions when it is executed.
  • any of the first receiving module 611, the intent acquiring module 613, the business process determining module 615, the entity information acquiring module 617, and the reply information determining module 619 can be combined into one module, or any one of them can be implemented Is split into multiple modules. Or, at least part of the functions of one or more of these modules may be combined with at least part of the functions of other modules and implemented in one module.
  • At least one of the first receiving module 611, the intention acquiring module 613, the business process determining module 615, the entity information acquiring module 617, and the reply information determining module 619 may be at least partially implemented as a hardware circuit, for example Field Programmable Gate Array (FPGA), Programmable Logic Array (PLA), System on Chip, System on Substrate, System on Package, Application Specific Integrated Circuit (ASIC), or any other that can integrate or package the circuit It is implemented in a reasonable way such as hardware or firmware, or implemented in any one of the three implementation methods of software, hardware, and firmware, or an appropriate combination of any of them.
  • FPGA Field Programmable Gate Array
  • PDA Programmable Logic Array
  • ASIC Application Specific Integrated Circuit
  • At least one of the first receiving module 611, the intention acquiring module 613, the business process determining module 615, the entity information acquiring module 617, and the reply information determining module 619 may be at least partially implemented as a computer program module, when the computer program module When being run, the corresponding function can be executed.
  • FIG. 8 schematically shows a block diagram of a computer system suitable for implementing information processing according to an embodiment of the present disclosure.
  • the computer system shown in FIG. 8 is only an example, and should not bring any limitation to the function and scope of use of the embodiments of the present disclosure.
  • a computer system 800 includes a processor 801, which can be loaded into a random access memory (RAM) 803 according to a program stored in a read only memory (ROM) 802 or from a storage part 808 The program executes various appropriate actions and processing.
  • the processor 801 may include, for example, a general-purpose microprocessor (for example, a CPU), an instruction set processor and/or a related chipset and/or a special purpose microprocessor (for example, an application specific integrated circuit (ASIC)), and so on.
  • the processor 801 may also include on-board memory for caching purposes.
  • the processor 801 may include a single processing unit or multiple processing units for performing different actions of a method flow according to an embodiment of the present disclosure.
  • the processor 801, the ROM 802, and the RAM 803 are connected to each other through a bus 804.
  • the processor 801 executes various operations of the method flow according to the embodiments of the present disclosure by executing programs in the ROM 802 and/or RAM 803. It should be noted that the program may also be stored in one or more memories other than ROM 802 and RAM 803.
  • the processor 801 may also execute various operations of the method flow according to the embodiments of the present disclosure by executing programs stored in the one or more memories.
  • the system 800 may further include an input/output (I/O) interface 805, and the input/output (I/O) interface 805 is also connected to the bus 804.
  • the system 800 may also include one or more of the following components connected to the I/O interface 805: an input part 806 including a keyboard, a mouse, etc.; including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker
  • the output part 807 including the hard disk and the like; the storage part 808 including the hard disk and the like; and the communication part 809 including the network interface card such as a LAN card, a modem, and the like.
  • the communication section 809 performs communication processing via a network such as the Internet.
  • the driver 810 is also connected to the I/O interface 805 as needed.
  • a removable medium 811 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 810 as needed, so that the computer program read from it is installed into the storage section 808 as needed.
  • the method flow according to the embodiment of the present disclosure may be implemented as a computer software program.
  • the embodiments of the present disclosure include a computer program product, which includes a computer program carried on a computer-readable storage medium, and the computer program contains program code for executing the method shown in the flowchart.
  • the computer program may be downloaded and installed from the network through the communication part 809, and/or installed from the removable medium 811.
  • the above-mentioned functions defined in the system of the embodiment of the present disclosure are executed.
  • the above-described systems, devices, devices, modules, units, etc. may be implemented by computer program modules.
  • the present disclosure also provides a computer-readable storage medium.
  • the computer-readable storage medium may be included in the device/device/system described in the above embodiment; or it may exist alone without being assembled into the device/ In the device/system.
  • the aforementioned computer-readable storage medium carries one or more programs, and when the aforementioned one or more programs are executed, the method according to the embodiments of the present disclosure is implemented.
  • the computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, may include but not limited to: portable computer disk, hard disk, random access memory (RAM), read-only memory (ROM) , Erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above.
  • a computer-readable storage medium may be any tangible medium that contains or stores a program, and the program may be used by or in combination with an instruction execution system, apparatus, or device.
  • a computer-readable storage medium may include one or more memories other than the ROM 802 and/or RAM 803 and/or ROM 802 and RAM 803 described above.
  • each block in the flowchart or block diagram may represent a module, program segment, or part of code, and the above-mentioned module, program segment, or part of code contains one or more for realizing the specified logical function Executable instructions.
  • the functions marked in the block may also occur in a different order from the order marked in the drawings. For example, two blocks shown in succession can actually be executed substantially in parallel, and they can sometimes be executed in the reverse order, depending on the functions involved.
  • each block in the block diagram or flowchart, and the combination of blocks in the block diagram or flowchart can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be It is realized by a combination of dedicated hardware and computer instructions.

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Strategic Management (AREA)
  • Finance (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • General Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Game Theory and Decision Science (AREA)
  • Mathematical Physics (AREA)
  • General Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Artificial Intelligence (AREA)
  • Human Computer Interaction (AREA)
  • Information Transfer Between Computers (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

一种信息处理方法、信息处理装置和信息处理系统，该信息处理方法包括：接收来自所述客户端的第一信息；响应于接收到的所述第一信息，对所述第一信息进行语义理解，得到用户意图；确定所述用户意图对应的业务流程；从所述第一信息以及所述第一信息的上下文信息获取实体信息，所述实体信息与所述用户意图对应的业务流程相关；基于所述用户意图对应的业务流程以及所述实体信息确定所述第一信息的答复信息并发送至所述客服端。

Description

信息处理方法、信息处理装置和信息处理系统
本申请要求于2019-04-23递交的、申请号为201910328210.6的中国专利申请的优先权，其内容一并在此作为参考。
技术领域
本公开涉及互联网技术领域，更具体地，涉及一种信息处理方法、信息处理装置和信息处理系统。
背景技术
随着人工智能、通信和计算机技术的快速发展，机器人被越来越多地应用于工农业生产和日常生活等诸多领域。为了提升服务质量，电商平台的客服会通过聊天工具帮助客户解决所遇到的问题。
在实现本公开构思的过程中，发明人发现现有技术中至少存在如下问题：由于电商的业务涉及到售前(如商品属性的咨询)售后(如退换货等)等多种业务，所涉及的业务的流程复杂，无法通过简单的培训使客服掌握各种业务，导致无法有效地提升客服的服务质量。
发明内容
有鉴于此，本公开提供了一种有助于提升客服的服务质量的信息处理方法、信息处理装置和信息处理系统。
本公开的一个方面提供了一种信息处理方法，适用于服务器端，所述服务器端分别与客户端和客服端通信连接，所述方法可以包括如下操作，首先，接收来自所述客户端的第一信息，响应于接收到的所述第一信息，对所述第一信息进行语义理解，得到用户意图，然后，确定所述用户意图对应的业务流程，接着，从所述第一信息以及所述第一信息的上下文信息获取实体信息，所述实体信息与所述用户意图对应的业务流程相关，然后，基于所述用户意图对应的业务流程以及所述实体信息确定所述第一信息的答复信息并发送至所述客服端。
根据本公开的实施例，通过识别用户意图，进入到不同的业务场景，辅助客服完成应答，而不需要客服了解复杂的业务，且给出的答复信息中结合了第一信息和上下文信息中 的实体信息，给出的答复信息更符合用户真实需求，期间无需客服搜索知识库，从而提高应答的效率，减少培训、服务等成本。
根据本公开的实施例，所述确定并输出所述第一信息的答复信息可以包括如下操作：针对每个业务流程构建一个节点树，所述节点树的根节点对应一个用户意图，然后，确定所述节点树中根节点的类型并将根节点作为当前节点，所述根节点的类型包括以下之一：答案类型、规则类型、子流程类型和常见答复类型，接着，重复执行以下步骤，直至当前节点的类型为答案类型或常见答复类型：如果所述当前节点的类型为规则类型或子流程类型，则基于所述实体信息、规则、子流程中至少一种确定子节点，并将子节点作为当前节点，确定当前节点的类型，然后，获取所述当前节点的答复信息并将所述当前节点的答复信息输出至客服端。需要说明的是，构建节点树的过程可以是预先构建的，即，不用每次确定答复信息时都进行一次节点树的构建过程。通过以上操作，即可实现自动获取与用户提供的实体信息对应的答复信息，更符合用户实际需求。
根据本公开的实施例，所述获取所述当前节点的答复信息并将所述当前节点的答复信息输出至客服端可以包括如下操作：一方面，如果所述当前节点的类型为答案类型，则基于所述当前节点的标识从第一答案库中获取并输出答复信息，另一方面，如果所述当前节点的类型为常见答复类型，则对所述第一信息进行向量化，得到向量化的第一信息；利用所述向量化的第一信息在第二答案库中进行匹配，得到并输出至少一个答复信息。
根据本公开的实施例，所述方法还可以包括如下操作：首先，接收来自所述客户端发送的第一信息之后，获取所述第一信息的用户的历史数据，然后，获取所述用户的历史数据的用户特征，接着，基于所述用户特征获取用户的预测意图和/或预测技能，然后，基于所述用户特征、所述预测意图和所述预测技能中至少一种，以及客服特征从多个客服中确定当前服务客服，接着，将所述第一信息发送给所述当前服务客服。通过以上操作可以基于用户的历史数据和客服特征确定适合该用户的客服，有助于提升客服的服务质量。
根据本公开的实施例，所述第一信息的用户的历史数据至少包括以下一种：上一次咨询内容、最近订单、异常订单、是否敏感顾客。这样可以有效提取出用户意图对应的业务流程所需的实体信息等。
本公开的另一个方面提供了一种信息处理方法，适用于客服端，所述客服端分别与服务器端和客户端通信连接，所述方法可以包括如下操作，首先，接收所述服务器端发送的第一信息和答复信息，所述答复信息为所述服务器端基于用户意图、实体信息以及业务流程生成的，所述业务流程与所述第一信息的用户意图相对应，所述实体信息从所述第一信 息以及所述第一信息的上下文信息获取的，且与所述业务流程相关，然后，展示所述第一信息和所述答复信息，接着，接收客服操作，响应于所述客服操作，将所述答复信息或客服输入的信息发送给所述客户端。
根据本公开的实施例，客服端可以基于服务器端提供的答复信息对客户端发送的第一信息进行答复，且答复信息是服务器端通过识别用户意图得到对应的业务流程，再基于上下文中的实体信息和业务流程得到的答复信息，因此，可以在不同的业务场景辅助客服完成应答，而不需要客服了解复杂的业务，且给出的答复信息更符合用户真实需求，期间无需客服搜索知识库，从而提高应答的效率，减少培训、服务等成本。
根据本公开的实施例，所述方法还可以包括如下操作，在输出所述答复信息或客服输入的信息之后，将所述第一信息、所述答复信息或所述客服输入的信息中的至少一个发送给所述服务器端，以便所述服务器端分析当前服务客服的客服特征。这样可以便于实现自动化获取客服特征，且更加客观。
本公开的另一个方面提供了一种信息处理装置，所述装置可以包括第一接收模块、意图获取模块、业务流程确定模块、实体信息获取模块和答复信息确定模块，其中，所述第一接收模块用于接收来自所述客户端的第一信息，所述意图获取模块用于响应于接收到的所述第一信息，对所述第一信息进行语义理解，得到用户意图，所述业务流程确定模块用于确定所述用户意图对应的业务流程，所述实体信息获取模块用于从所述第一信息以及所述第一信息的上下文信息获取实体信息，所述实体信息与所述用户意图对应的业务流程相关，所述答复信息确定模块用于基于所述用户意图对应的业务流程以及所述实体信息确定所述第一信息的答复信息并发送至所述客服端。
根据本公开的实施例，所述答复信息确定模块可以包括：节点类型确定子模块、递归子模块和信息获取子模块，其中，所述节点类型确定子模块用于确定所述节点树中根节点的类型并将根节点作为当前节点，所述根节点的类型包括以下之一：答案类型、规则类型、子流程类型和常见答复类型，所述递归子模块用于重复执行以下步骤，直至当前节点的类型为答案类型或常见答复类型：如果所述当前节点的类型为规则类型或子流程类型，则基于所述实体信息、规则、子流程中至少一种确定子节点，并将子节点作为当前节点，确定当前节点的类型，所述信息获取子模块用于获取所述当前节点的答复信息并将所述当前节点的答复信息输出至客服端。
根据本公开的实施例，所述答复信息确定模块还可以包括节点树构建子模块，该节点树构建子模块用于针对每个业务流程构建一个节点树，所述节点树的根节点对应一个用户 意图。
根据本公开的实施例，所述信息获取子模块可以包括：第一获取单元和第二获取单元，其中，所述第一获取单元用于如果所述当前节点的类型为答案类型，则基于所述当前节点的标识从第一答案库中获取并输出答复信息，所述第二获取单元用于如果所述当前节点的类型为常见答复类型，则对所述第一信息进行向量化，得到向量化的第一信息，利用所述向量化的第一信息在第二答案库中进行匹配，得到并输出至少一个答复信息。
根据本公开的实施例，所述装置还可以包括历史数据获取模块、用户特征获取模块、预测模块、匹配模块和路由模块，其中，该历史数据获取模块用于接收来自所述客户端发送的第一信息之后，获取所述第一信息的用户的历史数据，所述用户特征获取模块用于获取所述用户的历史数据的用户特征，所述预测模块用于基于所述用户特征获取用户的预测意图和/或预测技能，所述匹配模块用于基于所述用户特征、所述预测意图和所述预测技能中至少一种，以及客服特征从多个客服中确定当前服务客服，所述路由模块用于将所述第一信息发送给所述当前服务客服。
根据本公开的实施例，所述第一信息的用户的历史数据至少包括以下一种：上一次咨询内容、最近订单、异常订单、是否敏感顾客。
本公开的另一个方面提供了一种信息处理装置，所述装置可以包括第二接收模块、展示模块、操作接收模块和信息输出模块，其中，所述第二接收模块用于接收所述服务器端发送的第一信息和答复信息，所述答复信息为所述服务器端基于用户意图、实体信息以及业务流程生成的，所述业务流程与所述第一信息的用户意图相对应，所述实体信息从所述第一信息以及所述第一信息的上下文信息获取的，且与所述业务流程相关，所述展示模块用于展示所述第一信息和所述答复信息，所述操作接收模块用于接收客服操作，所述信息输出模块用于响应于所述客服操作，将所述答复信息或客服输入的信息发送给所述客户端。
根据本公开的实施例，所述装置还可以包括客服信息获取模块，所述客服信息获取模块用于在输出所述答复信息或客服输入的信息之后，将所述第一信息、所述答复信息或所述客服输入的信息中的至少一个发送给所述服务器端，以便所述服务器端分析当前服务客服的客服特征。
本公开的另一个方面提供了一种信息处理系统，所述系统可以包括对话管理模块、意图识别模块、业务模型模块和答案模块，其中，所述对话管理模块用于获取第一信息、所述第一信息的上下文信息和实体信息，以及输出第一信息的答复信息，所述意图识别模块用于至少基于所述第一信息获取用户意图，所述业务模型模块用于至少基于所述第一信息 和所述用户意图确定业务流程，所述答案模块用于基于所述业务流程和所述实体信息确定所述第一信息的答复信息。
根据本公开的实施例，所述系统还可以包括智能调度模块，所述只能调度模块用于基于用户的历史咨询数据通过用户意图与客服服务的内容来确定意图与技能的匹配关系，并基于所述匹配关系确定该用户适合的客服。
根据本公开的实施例，所述系统还可以包括路由模块，所述路由模块用于建立客户端和客服端的连接，以及在服务器端、客户端和客服端之间投递信息。
本公开的另一个方面提供了一种计算机系统，包括一个或多个处理器，以及存储装置，其中所述存储装置，用于存储可执行指令，所述可执行指令在被所述处理器执行时，实现如上所述的方法。
本公开的另一方面提供了一种计算机可读存储介质，存储有计算机可执行指令，所述指令在被执行时用于实现如上所述的方法。
本公开的另一方面提供了一种计算机程序，所述计算机程序包括计算机可执行指令，所述指令在被执行时用于实现如上所述的方法。
附图说明
通过以下参照附图对本公开实施例的描述，本公开的上述以及其他目的、特征和优点将更为清楚，在附图中：
图1A示意性示出了根据本公开实施例的信息处理方法、信息处理装置和信息处理系统的应用场景；
图1B示意性示出了根据本公开实施例的系统架构示意图；
图2A示意性示出了根据本公开实施例的用于服务器端的信息处理方法的流程图；
图2B示意性示出了根据本公开实施例的用于服务器端的信息处理方法的信息流向图；
图3A示意性示出了根据本公开实施例的确定并输出所述第一信息的答复信息的流程图；
图3B示意性示出了根据本公开实施例的获取所述当前节点的答复信息的流程图；
图4示意性示出了根据本公开另一实施例的用于服务器端的信息处理方法的流程图；
图5A示意性示出了根据本公开实施例的用于客服端的信息处理方法的流程图；
图5B示意性示出了根据本公开另一实施例的用于客服端的信息处理方法的流程图；
图6A示意性示出了根据本公开实施例的信息处理装置的框图；
图6B示意性示出了根据本公开另一实施例的信息处理装置的框图；
图7A示意性示出了根据本公开实施例的信息处理系统的框图；
图7B示意性示出了根据本公开实施例的信息处理系统的信息流的示意图；
图7C示意性示出了根据本公开实施例的对话管理模块的逻辑示意图；
图7D示意性示出了根据本公开实施例的意图识别模块的逻辑示意图；
图7E示意性示出了根据本公开实施例的业务模型模块的逻辑示意图；
图7F示意性示出了根据本公开实施例的答案模块的逻辑示意图；
图7G示意性示出了根据本公开实施例的智能调度模块的逻辑示意图；以及
图8示意性示出了根据本公开实施例的适于实现信息处理的计算机系统的框图。
具体实施方式
以下，将参照附图来描述本公开的实施例。但是应该理解，这些描述只是示例性的，而并非要限制本公开的范围。在下面的详细描述中，为便于解释，阐述了许多具体的细节以提供对本公开实施例的全面理解。然而，明显地，一个或多个实施例在没有这些具体细节的情况下也可以被实施。此外，在以下说明中，省略了对公知结构和技术的描述，以避免不必要地混淆本公开的概念。
在此使用的术语仅仅是为了描述具体实施例，而并非意在限制本公开。在此使用的术语“包括”、“包含”等表明了所述特征、步骤、操作和/或部件的存在，但是并不排除存在或添加一个或多个其他特征、步骤、操作或部件。
在此使用的所有术语(包括技术和科学术语)具有本领域技术人员通常所理解的含义，除非另外定义。应注意，这里使用的术语应解释为具有与本说明书的上下文相一致的含义，而不应以理想化或过于刻板的方式来解释。
在使用类似于“A、B和C等中至少一个”这样的表述的情况下，一般来说应该按照本领域技术人员通常理解该表述的含义来予以解释(例如，“具有A、B和C中至少一个的系统”应包括但不限于单独具有A、单独具有B、单独具有C、具有A和B、具有A和C、具有B和C、和/或具有A、B、C的系统等)。在使用类似于“A、B或C等中至少一个”这样的表述的情况下，一般来说应该按照本领域技术人员通常理解该表述的含义来予以解释(例如，“具有A、B或C中至少一个的系统”应包括但不限于单独具有A、单独具有B、单独具有C、具有A和B、具有A和C、具有B和C、和/或具有A、B、C的系统等)。
为了满足给客服推荐答案的流程，需要由服务器端提供支持的机器人理解客户发送的信息并给出答复信息，最终通过客服或机器人给客户端发送答复信息。作为辅助类型的机器人，所有的答复信息都是由人工点击之后发送给客户端，机器人在过程中主要是实现辅助化的应答。
目前主流的客服应答辅助工具主要分为两种方式，第一种是答案话术推荐，第二种是官方话术搜索。其中，关于话术推荐，顾客说话过程中识别顾客的意图，并通过机器学习模型，从答案库中匹配到响应的答案，推荐给客服，客服点击之后发送给顾客。关于话术搜索，建立一个知识库，客服理解顾客说的内容，再由客服到知识库中搜索相关的话术，点击之后发送给顾客。
现有话术推荐的答案都为静态答案，流程固定无法灵活的配置。如顾客问：我的某某订单什么时候到货？这时候推荐的答案可能是：您好，非常抱歉，您的订单还在配送途中，请耐心等待。此时没有明确告知顾客想要的答案，而是进行简单的安抚，无法正真的满足顾客的需求，降低了系统的可用性。
现有话术搜索的答案为固定答案，需要人工理解复杂的业务，如客户问：我这个商品能退货么？这时候客服可能需要理解到顾客是想到退货，并且到知识库搜索：退货流程。在复杂场景下，则可能出现，订单状态为未收货时的退货流程、订单状态为已收货时的退货流程等。较为复杂，需要客服去反复的确认，并不停的搜索，大大的降低了接待时效。
本公开的实施例提供了一种信息处理方法、信息处理装置和信息处理系统。该方法包括业务流程确定过程和答复信息确定过程。在业务流程确定过程中，首先确定接收的第一信息的用户意图，接着确定该用户意图对应的业务流程。在完成业务流程确定过程之后，进入答复信息确定过程，基于所述业务流程以及所述第一信息中包括的实体信息确定所述第一信息的答复信息。
此外，本公开的实施例可以通过客服的服务技能标签和资源动态分配客服，并通过少量的了解具体业务的专业人员来配置流程和答案，实现根据业务给客服推荐标准化答案，客服按步骤点击将答案发送给客户端，期间不需要客服了解过多的业务，能有效减少培训成本，从而实现流量的合理分配、流程的统一化、应答的标准化。
图1A示意性示出了根据本公开实施例的信息处理方法、信息处理装置和信息处理系统的应用场景。需要注意的是，图1A所示仅为可以应用本公开实施例的应用场景的示例，以帮助本领域技术人员理解本公开的技术内容，但并不意味着本公开实施例不可以用于其他设备、系统、环境或场景。
如图1A所示，客服人员在对用户提供咨询服务时，会面对各种各样的用户和问题，涉及多种业务流程，客服人员无法在接受少量的培训时应对各种各样的用户的问题。图1A中用户买的xxx电视机无法遥控，咨询客服人员如何处理，这可能涉及到电器的维修业务流程、退货业务流程或换货业务流程，同时还要考虑是否在包换期、保修期等，不同品牌的电视机的处理流程可能不一样，因此，需要客服人员比较了解相应的业务流程才能更好的引导用户采取合适的应对策略，例如，客服人员需要在知识库中查询xxx品牌电视机的相关业务流程，造成用户需要较长的等待时间等。本公开提供的信息处理方法，可以由服务器端对用户发送的问题进行语义理解，得到用户意图，如电子设备维修，确定对应的业务流程，然后基于与该业务流程相关的实体信息确定答复信息，该实体信息可以是从上下文中提取出来的，这样就可以基于业务流程和实体信息得到用户想要的答案：拨打维修人员电话186xxxxxxxx，并推送给客服，当客服确认没有问题时即可发送给用户。
图1B示意性示出了根据本公开实施例的系统架构示意图。
如图1B所示，根据该实施例的系统架构100可以包括终端设备101、102、103，网络104和服务器105。网络104用以在终端设备101、102、103和服务器105之间提供通信链路的介质。网络104可以包括各种连接类型，例如有线、无线通信链路或者光纤电缆等等。
用户可以使用终端设备101、102、103通过网络104与服务器105交互，以接收或发送消息等。终端设备101、102、103上可以安装有各种通讯客户端应用，例如购物类应用、网页浏览器应用、搜索类应用、即时通信工具、邮箱客户端、社交平台软件等(仅为示例)。
终端设备101、102、103可以是具有显示屏并且支持网页浏览的各种电子设备，包括但不限于智能手机、平板电脑、膝上型便携计算机、台式计算机和互联网电视等等。
服务器105可以是提供各种服务的服务器，例如对用户利用终端设备101、102、103发送的信息进行答复的后台管理服务器(仅为示例)。后台管理服务器可以对接收到的用户请求等数据进行分析等处理，并将处理结果(例如根据用户请求获取或生成的网页、信息、或数据等)反馈给终端设备。
需要说明的是，终端设备101、102、103可以作为客户端，也可以作为客服端，客服端上可以使用服务器端分配的辅助机器人。
应该理解，终端设备、网络和服务器的数目仅仅是示意性的。根据实现需要，可以具有任意数目的终端设备、网络和服务器。
图2A示意性示出了根据本公开实施例的用于服务器端的信息处理方法的流程图。该 方法适用于服务器端，所述服务器端分别与客户端和客服端通信连接。
如图2A所示，该方法包括操作S201～操作S209。
在操作S201，接收来自所述客户端的第一信息。
在本实施例中，所述第一信息可以为用户发送给客服的咨询信息、闲聊信息等。
在操作S203，响应于接收到的所述第一信息，对所述第一信息进行语义理解，得到用户意图。
由于对话的结构是这样的：在一个会话内，用户可能会问很多个问题，客服或机器人也会给出多个答复信息。上述多个问题可能会涉及到多个意图。例如，客户可能在一个会话里先提问：我的货怎么还没到，对应的用户意图可能是：催单。接着客户又提问：还不到，我换一个商品吧。此时的用户意图可能是：换货。
在一个具体实施例中，可以基于预先训练好的语义理解模型(semantic procedural model)对所述第一信息进行语义理解，得到用户意图。该语意理解模型包括但不限于：多分类神经网络、深度神经网络、卷及神经、对抗神经网络、长短时记忆神经网络等。需要说明的是，客户发送的第一信息可能被识别成多个用户意图，例如，客户发送信息：“发票呢”。语义理解后返回用户意图的格式可以如：[{补发票，0.9999}，{修改发票，0.1999}，{不要发票，0.0099}]，会返回多个用户意图，数字为是这个用户意图的概率。一般会设一个阈值，比如0.8，当这个用户意图的概率大于0.8的时候返回该用户意图。
在操作S205，确定所述用户意图对应的业务流程。
其中，业务流程可以为了解具体业务流程的专业人员来配置流程和答案，其中，每个业务流程可以对应一个用户意图。
在操作S207，从所述第一信息以及所述第一信息的上下文信息获取实体信息，所述实体信息与所述用户意图对应的业务流程相关。
由于一个业务流程通常需要多个实体信息的支持，例如，换货业务，需要用户的收件地址、收件人、收件人联系方式、收件时间等实体信息，才能保证该业务的顺利进行，在客户与客服的沟通过程中，这些实体信息可能在之前的聊天记录中存在，也可能需要客服对客户进行引导才能得到，因此，需要从所述第一信息以及所述第一信息的上下文信息获取实体信息。业务流程所需的实体信息可以由服务器端发送给客服端以对客服进行提示。
在操作S209，基于所述用户意图对应的业务流程以及所述实体信息确定所述第一信息的答复信息并发送至所述客服端。
在本实施例中，服务器端可能给客服端发送的是精确的答复信息，也可能是服务器端 给客服端发送的多个可能的答复信息，再由客服从多个可能的答复信息中选取最合适的答复信息，然后发送给客户端。
图2B示意性示出了根据本公开实施例的用于服务器端的信息处理方法的信息流向图。
如图2B所示，客户端发起会话，与客服端建立连接。路由模块负责3端的信息交互：将对话消息投递到客服端显示。会话管理模块对客户端发送的第一信息进行意图识别，如果识别出用户意图，给客服端发送一条事件消息，高亮提示客服识别到的用户意图。客服点击第一信息或在接收到用户意图时，弹出辅助机器人。辅助机器人根据第一信息、第一信息的上下文信息和用户意图对应的规则(如历史信息和客服引导用户给出的信息)得到答复信息，返回答复信息给客服端，客服将答复信息或自行输入的信息发送给客户端。
图3A示意性示出了根据本公开实施例的确定并输出所述第一信息的答复信息的流程图。
如图3A所示，所述确定并输出所述第一信息的答复信息可以包括操作S301～操作S305。
在操作S301，确定用户意图对应的节点树中根节点的类型并将根节点作为当前节点，所述根节点的类型包括以下之一：答案类型、规则类型、子流程类型和常见答复类型。
在本实施例中，所述节点树可以是基于业务流程构建的，例如，预先根据业务流程构建节点树，或者，在需要确定第一信息的答复信息时根据业务流程构建节点树，或者，在发现涉及的业务流程不存在对应节点树时根据业务流程构建节点树。节点树的根节点对应一个用户意图，也就是说，基于用户意图确定对应的节点树，并确定节点树的根节点的类型。
在一个实施例中，用户意图与业务流程的对应关系是事先确认好的，比如某个客服被分配的辅助机器人只涉及手机、闲聊的业务流程，不会解答电脑的业务流程的问题。那么该客服能处理的用户意图对应的业务流程可能只有：手机咨询、手机退货、手机换货等。当确定了用户意图之后，就可以确认客户所需的业务流程，得到该业务流程对应的节点树。节点树是一个配置，当进了某个节点树之后，再执行这个节点树。
在操作S303，重复执行以下步骤，直至当前节点的类型为答案类型或常见答复类型：如果所述当前节点的类型为规则类型或子流程类型，则基于所述实体信息、规则、子流程中至少一种确定子节点，并将子节点作为当前节点；确定当前节点的类型。
例如，每个用户意图在节点树中存在一个唯一的值，用来标识根节点，判断是否需要进入新的节点树的根节点(例如，客户端发送了一条新信息，该新信息是否有新的用户意 图)，如果是，进入新的用户意图对应的节点树的根节点，如果否，则定位到上次对话的用户意图对应的节点树的子节点，判断该子节点的节点类型，节点类型分为以下几类：答案类型、规则类型、子流程类型(如通用规则类型，可复用)、常见答复类型(如常见问题解答(Frequently Asked Questions，简称FAQ)类型)。判断该子节点是否为规则类型或子流程类型，如果是，则根据规则判断命中的子节点，再返回子节点标识以便进入该子节点，此过程为递归过程，直至子节点的类型为答案类型或常见答复类型。
在操作S305，获取所述当前节点的答复信息并将所述当前节点的答复信息输出至客服端。
在本实施例中，经过操作S303，最终得到的子节点为答案类型或常见答复类型，此时，可以调用答案类型的子节点对应的答复信息或者常见答复类型对应的礼貌性回复用语作为第一信息的答复信息。
需要说明的是，在确定用户意图之后，如果是闲聊的用户意图，则后面会执行闲聊对应的节点树。如用户发送信息：吃饭了么。服务器端会给出答复信息：才xx点，我还没吃饭呢。又例如，如果是手机退货的用户意图，则服务器基于手机退货的用户意图对应的节点树的节点的标识，在答案库中查找答复信息，如回答：退货请邮寄到地址xxx，该商品支持7天无理由退货。其中，答案库中的答复信息可以是人工配置的。
此外，答复信息可以是结合实体信息的，也可以不结合实体信息，但会话过程中的实体信息是会一直向下传递到答案这层。
例如，用户在之前涉及的业务流程中给出信息：我的收货地址是xxxx。服务器端可能会用到【收货地址】这个实体。如服务器端在后续换货的业务流程中需要用到【收货地址】这个实体，此时，服务器端可以发送信息：请您在确认下您的收货地址是xxxx么，是的话点【是】，不是的话点【否】，这样可以有效提升沟通效率。
图3B示意性示出了根据本公开实施例的获取所述当前节点的答复信息的流程图。
如图3B所示，所述获取所述当前节点的答复信息并将所述当前节点的答复信息输出至客服端可以包括操作S311～操作S313。
在操作S311，如果所述当前节点的类型为答案类型，则基于所述当前节点的标识从第一答案库中获取并输出答复信息。
例如，直接调用答案服务模型。具体地，通过第一信息的上下文来判断是当前节点的类型是答案类型还是FAQ类型。如果是答案类型，则根据节点标识从缓存中获取最终的答复信息，节点标识由业务类型对应的节点树的子节点返回。其中，答复信息由人工编辑 审核后存放于持久化答案层，每个节点标识对应一个答复信息。
在操作S313如果所述当前节点的类型为常见答复类型，则对所述第一信息进行向量化，得到向量化的第一信息；利用所述向量化的第一信息在第二答案库中进行匹配，得到并输出至少一个答复信息。
在本实施例中，如果不是答案类型，则为FAQ类型。将第一信息向量化，经过与预设的问题-答案对进行文本相似度匹配，检索出TOP N条候选答复信息，对候选答复信息做重排序，按照预设阈值筛选出1条或者多条答复信息，并返回答复信息。
图4示意性示出了根据本公开另一实施例的用于服务器端的信息处理方法的流程图。
如图4所示，该方法还可以包括操作S401～操作S409。
在操作S401，接收来自所述客户端发送的第一信息之后，获取所述第一信息的用户的历史数据。
在本实施例中，每个客户端的用户通常具有用户标识，在用户使用用户标识登录购物应用后，使用购物应用的在线客服工具进行问题咨询时，就可以通过用户标识获取该用户的历史数据。
其中，所述第一信息的用户的历史数据至少包括以下一种：上一次咨询内容、最近订单、异常订单、是否敏感顾客等。
在操作S403，获取所述用户的历史数据的用户特征。
例如，用户的购买力、忍受力、投诉率、是否有异常订单等。
在操作S405，基于所述用户特征获取用户的预测意图和/或预测技能。
例如，用户近期购买过手机，则该用户与客服进行沟通就可能与手机相关，例如，手机退货、手机换货、手机维修等。又例如，用户的忍受力较低，例如，曾有过多次投诉行为，因此，该用户需要对业务更加熟练和耐心较好的客服来提供服务。
在操作S407，基于所述用户特征、所述预测意图和所述预测技能中至少一种，以及客服特征从多个客服中确定当前服务客服。
在本实施例中，将所述用户特征、所述预测意图和所述预测技能中至少一种与数据库中存储的各客服对应的客服特征进行匹配，得到相适配的客服。
其中，客服特征可以是与存储在服务器中的数据，客服特征主要由客服画像构成，例如，客服的历史接待用户解决率、历史接待用户情绪变化曲线、历史满意度、善于回答的技能组标签等。
所述进行匹配可以包括初筛过程和筛选过程，例如，预先基于意图与技能对自动构建 技能意图模型。例如，意图与技能对是基于用户意图和客服的服务内容挖掘得到的，用户意图是该客服曾经服务过用户的历史咨询信息得到的。
在接收到客户端发送的第一信息时，调用客服特征。
然后，基于该第一信息对应用户的用户特征、客服特征和所述技能意图模型进行初步筛选，得到至少一组配对选项。
然后基于特征模型，如特征算法模型，对所述至少一组配对选项进行再次筛选，得到一个当前服务客服。其中，特征算法模型可以是基于当前空闲的客服、所述至少一组配对选项的匹配度排序等来确定当前服务客服。
在操作S409，将所述第一信息发送给所述当前服务客服。
在一个具体实施例中，用户特征主要由用户画像构成，包含了历史咨询意图，个人基本信息等。服务器可以从用户的历史数据库中选择一些与咨询相关的特征，例如，上一次咨询内容、最近订单、异常订单、是否敏感顾客等，然后，通过技能与意图的算法模型预测以下信息：首先确定顾客可能咨询的问题及意图，然后，确定对应该问题可能需要具有什么技能的客服，筛选出多组符合条件的接待客服，接着，通过客服特性、用户特征、预测意图和预测技能等，由特征算法模型，确定当前服务客服，接着，建立顾客和客服之间的网络链接，链接建立之后，顾客和客服之间可以发送信息。
通过以上方法使得可以基于用户的历史数据和客服特征等确定当前合适的客服，以提升用户满意度。
图5A示意性示出了根据本公开实施例的用于客服端的信息处理方法的流程图。其中，所述客服端分别与服务器端和客户端通信连接。
如图5A所示，所述方法可以包括操作S501～操作S507。需要说明的是，关于服务器确定用户通过客户端发送的第一信息的答复信息的过程可以参考以上公开的内容，在此不再赘述。
在操作S501，接收所述服务器端发送的第一信息和答复信息，所述答复信息为所述服务器端基于用户意图、实体信息以及业务流程生成的，所述业务流程与所述第一信息的用户意图相对应，所述实体信息从所述第一信息以及所述第一信息的上下文信息获取的，且与所述业务流程相关。
需要说明的是，可以由服务器端将检测到的用户意图对第一信息进行标注后发送给客服端，例如，以高亮的方式显示用户意图。此外，也可以是在客服点击第一信息之后，才显示用户意图，然后，在客服点击某个用户意图之后，才显示该用户意图对应的答复信息。
在操作S503，展示所述第一信息和所述答复信息。
在操作S505，接收客服操作。
其中，该客服操作可以是针对服务器端发送的答复信息，或者是客服输入的信息，或者是对服务器端发送的信息请求等。
在操作S507，响应于所述客服操作，将所述答复信息或客服输入的信息发送给所述客户端。
本公开提供的信息处理方法，通过少量的了解具体业务的专业人员来配置业务流程和答复信息，而实现业务推荐标准化答案，可以有效减少培训成本，从而实现流量的合理分配、流程的统一化、应答的标准化、服务效率的提升。
图5B示意性示出了根据本公开另一实施例的用于客服端的信息处理方法的流程图。
如图5B所示，所述方法还可以包括操作S509。
在操作S509，在输出所述答复信息或客服输入的信息之后，将所述第一信息、所述答复信息或所述客服输入的信息中的至少一个发送给所述服务器端，以便所述服务器端分析当前服务客服的客服特征。
其中，所述客服特征主要由客服画像构成，基于客服的历史服务信息提取出来的。例如，客服的历史接待用户解决率、历史接待用户情绪变化曲线、历史满意度、善于回答的技能组标签等。
需要说明的是，服务器端分析得到当前服务客服的客服特征之后，可以将客服特征存储在客服的个人信息数据库中，以便于服务器调用客服特征。
本公开提供的信息处理方法，通过分析客服的历史服务信息，得到客服特征，主动挖掘意图与技能标签的配对，自动对用户和客服建模，这样便于在服务器接收到客户端发送的第一信息时，采用规则加算法预测，实现不同的用户分配给不同的客服，保证接待的时效性与满意度。
图6A示意性示出了根据本公开实施例的信息处理装置的框图。
如图6A所示，所示信息处理装置610可以包括第一接收模块611、意图获取模块613、业务流程确定模块615、实体信息获取模块617和答复信息确定模块619。
其中，所述第一接收模块611用于接收来自所述客户端的第一信息。
所述意图获取模块613用于响应于接收到的所述第一信息，对所述第一信息进行语义理解，得到用户意图。
所述业务流程确定模块615用于确定所述用户意图对应的业务流程。
所述实体信息获取模块617用于从所述第一信息以及所述第一信息的上下文信息获取实体信息，所述实体信息与所述用户意图对应的业务流程相关。
所述答复信息确定模块619用于基于所述用户意图对应的业务流程以及所述实体信息确定所述第一信息的答复信息并发送至所述客服端。
在一个实施例中，所述答复信息确定模块619可以包括：节点类型确定子模块、递归子模块和信息获取子模块。
其中，所述节点类型确定子模块用于确定所述节点树中根节点的类型并将根节点作为当前节点，所述根节点的类型包括以下之一：答案类型、规则类型、子流程类型和常见答复类型。
所述递归子模块用于重复执行以下步骤，直至当前节点的类型为答案类型或常见答复类型：如果所述当前节点的类型为规则类型或子流程类型，则基于所述实体信息、规则、子流程中至少一种确定子节点，并将子节点作为当前节点，确定当前节点的类型，所述信息获取子模块用于获取所述当前节点的答复信息并将所述当前节点的答复信息输出至客服端。
在另一个实施例中，所述答复信息确定模块619还可以包括节点树构建子模块，该节点树构建子模块用于针对每个业务流程构建一个节点树，所述节点树的根节点对应一个用户意图。
其中，所述信息获取子模块可以包括：第一获取单元和第二获取单元。
所述第一获取单元用于如果所述当前节点的类型为答案类型，则基于所述当前节点的标识从第一答案库中获取并输出答复信息。
所述第二获取单元用于如果所述当前节点的类型为常见答复类型，则对所述第一信息进行向量化，得到向量化的第一信息，利用所述向量化的第一信息在第二答案库中进行匹配，得到并输出至少一个答复信息。
在另一个实施例中，所述装置610还可以包括历史数据获取模块621、用户特征获取模块623、预测模块625、匹配模块627和路由模块629。
其中，该历史数据获取模块621用于接收来自所述客户端发送的第一信息之后，获取所述第一信息的用户的历史数据。
所述用户特征获取模块623用于获取所述用户的历史数据的用户特征。
所述预测模块625用于基于所述用户特征获取用户的预测意图和/或预测技能。
所述匹配模块627用于基于所述用户特征、所述预测意图和所述预测技能中至少一种， 以及客服特征从多个客服中确定当前服务客服。
所述路由模块629用于将所述第一信息发送给所述当前服务客服。
具体地，所述第一信息的用户的历史数据至少包括以下一种：上一次咨询内容、最近订单、异常订单、是否敏感顾客。
图6B示意性示出了根据本公开另一实施例的信息处理装置的框图。
如图6B所示，所述装置630可以包括第二接收模块631、展示模块633、操作接收模块635和信息输出模块637。
其中，所述第二接收模块631用于接收所述服务器端发送的第一信息和答复信息，所述答复信息为所述服务器端基于用户意图、实体信息以及业务流程生成的，所述业务流程与所述第一信息的用户意图相对应，所述实体信息从所述第一信息以及所述第一信息的上下文信息获取的，且与所述业务流程相关。
所述展示模块633用于展示所述第一信息和所述答复信息。
所述操作接收模块635用于接收客服操作。
所述信息输出模块637用于响应于所述客服操作，将所述答复信息或客服输入的信息发送给所述客户端。
需要说明的是，所述装置630还可以包括客服信息获取模块639，所述客服信息获取模块639用于在输出所述答复信息或客服输入的信息之后，将所述第一信息、所述答复信息或所述客服输入的信息中的至少一个发送给所述服务器端，以便所述服务器端分析当前服务客服的客服特征。
其中，图6A～图6B中所示的装置610、630执行的操作可以参考如上所述的信息处理方法中相应的内容。
图7A示意性示出了根据本公开实施例的信息处理系统的框图。
如图7A所示，所述系统700可以包括对话管理模块710、意图识别模块730、业务模型模块750和答案模块770。
所述对话管理模块710用于获取第一信息、所述第一信息的上下文信息和实体信息，以及输出第一信息的答复信息。
图7B示意性示出了根据本公开实施例的对话管理模块的逻辑示意图。
如图7B所示，首先，顾客或客服说话(单条话)，请求(Query)发起，信息包含聊天内容、咨询的订单发票等实体信息、其他入参信息。然后，解析入参，进行前置逻辑处理，例如，判断是否合法请求、用户黑白名单等。接着，查看是否已经初始化上下文，如 果否，则从存储层中获取上下文，其中，上下文可以分为3种，例如，机器人上下文、对话上下文、请求上下文，主要是信息的存放与传递。然后，更新上下文信息。接着，分析语句的意图，其中，意图分为3个层，例如，采用多分类神经网络模型来实现，最终得到用户意图，如“发票类-补开发票”。然后，判断该用户意图是否能够被理解，如果否，则推荐FAQ等兜底话术，如果是，则进行实体识别。其中，实体识别过程是通过规则和模型，提取意图表单所需实体信息，例如，地址、订单号、手机号等。意图表单是一个任务的表单，比如注册一个用户，需要输入用户名和密码，这里的{用户名}和{密码}，就是【用户注册】这个{意图表单}的{实体}，在实际的业务中，例如，【补开发票】这个{意图表单}的{实体}可能包含：{公司名字}{公司纳税号}{发票收件地址}等，这些都是根据具体的业务来制定的。接着，将答复信息路由给客户端。
所述意图识别模块730用于至少基于所述第一信息获取用户意图。
图7C示意性示出了根据本公开实施例的意图识别模块的逻辑示意图。
如图7C所示，在获取上下文之后，进行数据清洗，包括但不限于：分词、停用词过滤等，然后，请求意图识别模型。如果没有加载意图识别模型，则需要通过加载器加载该意图识别模型，并基于加载的意图识别模型判断是否有用户意图，如过有，则按预设阈值返回用户意图集合。接着，通过对用户意图集合进行排序，筛选出最终的用户意图。如果已经识别了用户意图或没有用户意图，则认为该语句可能包含其他实体信息，通过实体模型提取实体信息。
需要说明的是，还可以返回答复信息给会话管理系统，当客服点击答案后，反馈埋点信息用于自动化优化意图识别模型。
所述业务模型模块750用于至少基于所述第一信息和所述用户意图确定业务流程。
图7D示意性示出了根据本公开实施例的业务模型模块的逻辑示意图。
如图7D所示，在识别到用户意图之后，将上下文作为入参，请求业务模型模块。每个意图在节点树中存在一个唯一的值，用来标识节点，其中，意图节点也叫根节点。需要说明的是，在每次接收到客户端发送的第一信息时，需要判断是否进入新的意图节点(即是否需要进入新的业务流程对应的节点树的根节点)，如果是，则进入新的根节点，如果否，则定位到上一个第一信息的节点树的子节点。接着，根据当前节点判断节点下面的子节点类型，节点类型分为以下几类：答案类型、规则类型、子流程类型(通用规则类型，可复用)、FAQ类型。然后，如果是答案类型，则直接调用答案服务模型。如果是规则、子流程类型，则根据规则判断命中的节点，再返回节点ID，此过程为递归过程。如果不 为以上3种类型，则为FAQ类型，直接调用答案服务模型，FAQ一般为礼貌性回复用语、或兜底推荐答案。
所述答案模块770用于基于所述业务流程和所述实体信息确定所述第一信息的答复信息。
图7E示意性示出了根据本公开实施例的答案模块的逻辑示意图。
如图7E所示，通过答复信息上下文来判断是答案类型还是FAQ类型。如果是答案类型，则根据节点ID从缓存中获取最终的答复信息，节点ID由业务模型的子节点返回。其中，答复信息是由人工编辑审核后存放于持久化答案层，每个节点ID对应一个答案。如果不是答案类型，则为FAQ类型。具体地，将第一信息的内容向量化，经过FAQ模型做(问题-答案对)文本相似度判断，检索出大于预设阈值的TOP N条候选答复信息，接着通过答案模型对候选答案数据做重排序，按照阈值筛选出1条或者多条答复信息，并路由给客服端。
此外，所述系统700还可以包括智能调度模块790，该智能调度模块790用于建立客户端与客服端之间的连接。
图7F示意性示出了根据本公开实施例的智能调度模块的逻辑示意图。
如图7F所示，接收到第一信息的时候，首先需要判断该第一信息对应用户由哪个客服来接待，这时用户在进线后并未说话，或者说了一句没有实际意图的话。由于无法得到当前用户实时的用户意图，则通过用户的历史信息来预测用户意图，历史信息中用户意图与客服服务的内容来发现意图与技能的匹配关系。首先，获取客服特征，客服特征主要由客服画像构成，例如，客服的历史接待用户解决率、历史接待用户情绪变化曲线、历史满意度、善于回答的技能组标签等。并获取用户特征，用户特征主要由用户画像构成，包含了历史咨询意图，个人基本信息等。然后，从历史数据中选取至少部分与咨询相关的特征，如：上一次咨询内容、最近订单、异常订单、是否敏感顾客等。通过技能意图模型预测用户可能咨询的问题及预测用户意图，对应该问题和预测用户意图可能需要具有什么技能的客服，筛选出多组符合条件的接待客服。接着，通过客服特性、用户特征、预测用户意图、需要的技能等，由特征模型，确定当前服务客服。这样就可以建立用户和客服之间的网络链接，链接建立之后，用户和客服之间可以发送信息。
图7G示意性示出了根据本公开实施例的信息处理系统的信息流的示意图。
如图7G所示，该信息处理系统可以包括前端交互部分、路由与智能接线部分和服务器确定答复信息部分。前端交互部分包括客户端、客服端和辅助机器人，路由与智能接线 部分包括智能调度模块和路由模块，服务器确定答复信息部分包括意图识别模块、对话管理模块、业务模型模块和答案模块。
路由模块在客户端和客服端之间进行消息投递，对于投递给客服端的消息中所包含的用户意图，可以进行标识，如高亮显示等。此外，路由模块基于智能调度模块分配的客服建立客户端与客服端之间的链路。路由模块在接收到第一信息时，将第一信息以及第一信息的上下文信息发送给对话管理模块，对话管理模块调用意图识别模块至少基于第一信息获取用户意图，并将用户意图发送给业务识别模块，业务识别模块基于规则逻辑等确定用户意图对应业务流程的节点树。答案模块用于基于所述节点树和所述实体信息确定所述第一信息的答复信息。
根据本公开的实施例的模块、子模块、单元、子单元中的任意多个、或其中任意多个的至少部分功能可以在一个模块中实现。根据本公开实施例的模块、子模块、单元、子单元中的任意一个或多个可以被拆分成多个模块来实现。根据本公开实施例的模块、子模块、单元、子单元中的任意一个或多个可以至少被部分地实现为硬件电路，例如现场可编程门阵列(FPGA)、可编程逻辑阵列(PLA)、片上系统、基板上的系统、封装上的系统、专用集成电路(ASIC)，或可以通过对电路进行集成或封装的任何其他的合理方式的硬件或固件来实现，或以软件、硬件以及固件三种实现方式中任意一种或以其中任意几种的适当组合来实现。或者，根据本公开实施例的模块、子模块、单元、子单元中的一个或多个可以至少被部分地实现为计算机程序模块，当该计算机程序模块被运行时，可以执行相应的功能。
例如，第一接收模块611、意图获取模块613、业务流程确定模块615、实体信息获取模块617和答复信息确定模块619中的任意多个可以合并在一个模块中实现，或者其中的任意一个模块可以被拆分成多个模块。或者，这些模块中的一个或多个模块的至少部分功能可以与其他模块的至少部分功能相结合，并在一个模块中实现。根据本公开的实施例，第一接收模块611、意图获取模块613、业务流程确定模块615、实体信息获取模块617和答复信息确定模块619中的至少一个可以至少被部分地实现为硬件电路，例如现场可编程门阵列(FPGA)、可编程逻辑阵列(PLA)、片上系统、基板上的系统、封装上的系统、专用集成电路(ASIC)，或可以通过对电路进行集成或封装的任何其他的合理方式等硬件或固件来实现，或以软件、硬件以及固件三种实现方式中任意一种或以其中任意几种的适当组合来实现。或者，第一接收模块611、意图获取模块613、业务流程确定模块615、实体信息获取模块617和答复信息确定模块619中的至少一个可以至少被部分地实现为计算 机程序模块，当该计算机程序模块被运行时，可以执行相应的功能。
图8示意性示出了根据本公开实施例的适于实现信息处理的计算机系统的框图。图8示出的计算机系统仅仅是一个示例，不应对本公开实施例的功能和使用范围带来任何限制。
如图8所示，根据本公开实施例的计算机系统800包括处理器801，其可以根据存储在只读存储器(ROM)802中的程序或者从存储部分808加载到随机访问存储器(RAM)803中的程序而执行各种适当的动作和处理。处理器801例如可以包括通用微处理器(例如CPU)、指令集处理器和/或相关芯片组和/或专用微处理器(例如，专用集成电路(ASIC))，等等。处理器801还可以包括用于缓存用途的板载存储器。处理器801可以包括用于执行根据本公开实施例的方法流程的不同动作的单一处理单元或者是多个处理单元。
在RAM 803中，存储有系统800操作所需的各种程序和数据。处理器801、ROM 802以及RAM 803通过总线804彼此相连。处理器801通过执行ROM 802和/或RAM 803中的程序来执行根据本公开实施例的方法流程的各种操作。需要注意，所述程序也可以存储在除ROM 802和RAM 803以外的一个或多个存储器中。处理器801也可以通过执行存储在所述一个或多个存储器中的程序来执行根据本公开实施例的方法流程的各种操作。
根据本公开的实施例，系统800还可以包括输入/输出(I/O)接口805，输入/输出(I/O)接口805也连接至总线804。系统800还可以包括连接至I/O接口805的以下部件中的一项或多项：包括键盘、鼠标等的输入部分806；包括诸如阴极射线管(CRT)、液晶显示器(LCD)等以及扬声器等的输出部分807；包括硬盘等的存储部分808；以及包括诸如LAN卡、调制解调器等的网络接口卡的通信部分809。通信部分809经由诸如因特网的网络执行通信处理。驱动器810也根据需要连接至I/O接口805。可拆卸介质811，诸如磁盘、光盘、磁光盘、半导体存储器等等，根据需要安装在驱动器810上，以便于从其上读出的计算机程序根据需要被安装入存储部分808。
根据本公开的实施例，根据本公开实施例的方法流程可以被实现为计算机软件程序。例如，本公开的实施例包括一种计算机程序产品，其包括承载在计算机可读存储介质上的计算机程序，该计算机程序包含用于执行流程图所示的方法的程序代码。在这样的实施例中，该计算机程序可以通过通信部分809从网络上被下载和安装，和/或从可拆卸介质811被安装。在该计算机程序被处理器801执行时，执行本公开实施例的系统中限定的上述功能。根据本公开的实施例，上文描述的系统、设备、装置、模块、单元等可以通过计算机程序模块来实现。
本公开还提供了一种计算机可读存储介质，该计算机可读存储介质可以是上述实施例 中描述的设备/装置/系统中所包含的；也可以是单独存在，而未装配入该设备/装置/系统中。上述计算机可读存储介质承载有一个或者多个程序，当上述一个或者多个程序被执行时，实现根据本公开实施例的方法。
根据本公开的实施例，计算机可读存储介质可以是非易失性的计算机可读存储介质，例如可以包括但不限于：便携式计算机磁盘、硬盘、随机访问存储器(RAM)、只读存储器(ROM)、可擦式可编程只读存储器(EPROM或闪存)、便携式紧凑磁盘只读存储器(CD-ROM)、光存储器件、磁存储器件、或者上述的任意合适的组合。在本公开中，计算机可读存储介质可以是任何包含或存储程序的有形介质，该程序可以被指令执行系统、装置或者器件使用或者与其结合使用。例如，根据本公开的实施例，计算机可读存储介质可以包括上文描述的ROM 802和/或RAM 803和/或ROM 802和RAM 803以外的一个或多个存储器。
附图中的流程图和框图，图示了按照本公开各种实施例的系统、方法和计算机程序产品的可能实现的体系架构、功能和操作。在这点上，流程图或框图中的每个方框可以代表一个模块、程序段、或代码的一部分，上述模块、程序段、或代码的一部分包含一个或多个用于实现规定的逻辑功能的可执行指令。也应当注意，在有些作为替换的实现中，方框中所标注的功能也可以以不同于附图中所标注的顺序发生。例如，两个接连地表示的方框实际上可以基本并行地执行，它们有时也可以按相反的顺序执行，这依所涉及的功能而定。也要注意的是，框图或流程图中的每个方框、以及框图或流程图中的方框的组合，可以用执行规定的功能或操作的专用的基于硬件的系统来实现，或者可以用专用硬件与计算机指令的组合来实现。
本领域技术人员可以理解，本公开的各个实施例和/或权利要求中记载的特征可以进行多种组合和/或结合，即使这样的组合或结合没有明确记载于本公开中。特别地，在不脱离本公开精神和教导的情况下，本公开的各个实施例和/或权利要求中记载的特征可以进行多种组合和/或结合。所有这些组合和/或结合均落入本公开的范围。
以上对本公开的实施例进行了描述。但是，这些实施例仅仅是为了说明的目的，而并非为了限制本公开的范围。尽管在以上分别描述了各实施例，但是这并不意味着各个实施例中的措施不能有利地结合使用。本公开的范围由所附权利要求及其等同物限定。不脱离本公开的范围，本领域技术人员可以做出多种替代和修改，这些替代和修改都应落在本公开的范围之内。

Claims (13)

  1. 一种信息处理方法，适用于服务器端，所述服务器端分别与客户端和客服端通信连接，所述方法包括：
    接收来自所述客户端的第一信息；
    响应于接收到的所述第一信息，对所述第一信息进行语义理解，得到用户意图；
    确定所述用户意图对应的业务流程；
    从所述第一信息以及所述第一信息的上下文信息获取实体信息，所述实体信息与所述用户意图对应的业务流程相关；以及
    基于所述用户意图对应的业务流程以及所述实体信息确定所述第一信息的答复信息并发送至所述客服端。
  2. 根据权利要求1所述的方法，其中，所述确定并输出所述第一信息的答复信息包括：
    针对每个业务流程构建一个节点树，所述节点树的根节点对应一个用户意图；
    确定所述节点树中根节点的类型并将根节点作为当前节点，所述根节点的类型包括以下之一：答案类型、规则类型、子流程类型和常见答复类型；
    重复执行以下步骤，直至当前节点的类型为答案类型或常见答复类型：如果所述当前节点的类型为规则类型或子流程类型，则基于所述实体信息、规则、子流程中至少一种确定子节点，并将子节点作为当前节点；确定当前节点的类型；以及
    获取所述当前节点的答复信息并将所述当前节点的答复信息输出至客服端。
  3. 根据权利要求2所述的方法，其中，所述获取所述当前节点的答复信息并将所述当前节点的答复信息输出至客服端包括：
    如果所述当前节点的类型为答案类型，则基于所述当前节点的标识从第一答案库中获取并输出答复信息；以及
    如果所述当前节点的类型为常见答复类型，则对所述第一信息进行向量化，得到向量化的第一信息；利用所述向量化的第一信息在第二答案库中进行匹配，得到并输出至少一个答复信息。
  4. 根据权利要求1所述的方法，还包括：
    接收来自所述客户端发送的第一信息之后，获取所述第一信息的用户的历史数据；
    获取所述用户的历史数据的用户特征；
    基于所述用户特征获取用户的预测意图和/或预测技能；
    基于所述用户特征、所述预测意图和所述预测技能中至少一种，以及客服特征从多个客服中确定当前服务客服；以及
    将所述第一信息发送给所述当前服务客服。
  5. 根据权利要求4所述的方法，其中，所述第一信息的用户的历史数据至少包括以下一种：上一次咨询内容、最近订单、异常订单、是否敏感顾客。
  6. 一种信息处理方法，适用于客服端，所述客服端分别与服务器端和客户端通信连接，所述方法包括：
    接收所述服务器端发送的第一信息和答复信息，所述答复信息为所述服务器端基于用户意图、实体信息以及业务流程生成的，所述业务流程与所述第一信息的用户意图相对应，所述实体信息从所述第一信息以及所述第一信息的上下文信息获取的，且与所述业务流程相关；
    展示所述第一信息和所述答复信息；
    接收客服操作；以及
    响应于所述客服操作，将所述答复信息或客服输入的信息发送给所述客户端。
  7. 根据权利要求6所述的方法，还包括：
    在输出所述答复信息或客服输入的信息之后，将所述第一信息、所述答复信息或所述客服输入的信息中的至少一个发送给所述服务器端，以便所述服务器端分析当前服务客服的客服特征。
  8. 一种信息处理装置，所述装置包括：
    第一接收模块，用于接收来自所述客户端的第一信息；
    意图获取模块，用于响应于接收到的所述第一信息，对所述第一信息进行语义理解，得到用户意图；
    业务流程确定模块，用于确定所述用户意图对应的业务流程；
    实体信息获取模块，用于从所述第一信息以及所述第一信息的上下文信息获取实体信息，所述实体信息与所述用户意图对应的业务流程相关；以及
    答复信息确定模块，用于基于所述用户意图对应的业务流程以及所述实体信息确定所述第一信息的答复信息并发送至所述客服端。
  9. 根据权利要求8所述的装置，还包括：
    历史数据获取模块，用于接收来自所述客户端发送的第一信息之后，获取所述第一信 息的用户的历史数据；
    用户特征获取模块，用于获取所述用户的历史数据的用户特征；
    预测模块，用于基于所述用户特征获取用户的预测意图和/或预测技能；
    匹配模块，用于基于所述用户特征、所述预测意图和所述预测技能中至少一种，以及客服特征从多个客服中确定当前服务客服；以及
    路由模块，用于将所述第一信息发送给所述当前服务客服。
  10. 一种信息处理装置，所述装置包括：
    第二接收模块，用于接收所述服务器端发送的第一信息和答复信息，所述答复信息为所述服务器端基于用户意图、实体信息以及业务流程生成的，所述业务流程与所述第一信息的用户意图相对应，所述实体信息从所述第一信息以及所述第一信息的上下文信息获取的，且与所述业务流程相关；
    展示模块，用于展示所述第一信息和所述答复信息；
    操作接收模块，用于接收客服操作；以及
    信息输出模块，用于响应于所述客服操作，将所述答复信息或客服输入的信息发送给所述客户端。
  11. 一种信息处理系统，所述系统包括：
    对话管理模块，用于获取第一信息、所述第一信息的上下文信息和实体信息，以及输出第一信息的答复信息；
    意图识别模块，用于至少基于所述第一信息获取用户意图；
    业务模型模块，用于至少基于所述第一信息和所述用户意图确定业务流程；
    答案模块，用于基于所述业务流程和所述实体信息确定所述第一信息的答复信息。
  12. 一种计算机系统，包括：
    一个或多个处理器；
    存储装置，用于存储可执行指令，所述可执行指令在被所述处理器执行时，实现根据权利要求1～7中任一项所述的方法。
  13. 一种计算机可读存储介质，其上存储有可执行指令，该指令被处理器执行时实现根据权利要求1～7中任一项所述的方法。
PCT/CN2020/078742 2019-04-23 2020-03-11 信息处理方法、信息处理装置和信息处理系统 WO2020215914A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
US17/600,508 US20220129908A1 (en) 2019-04-23 2020-03-11 Method, apparatus and system of processing information
EP20795968.5A EP3961423A4 (en) 2019-04-23 2020-03-11 INFORMATION PROCESSING METHOD, INFORMATION PROCESSING DEVICE AND INFORMATION PROCESSING SYSTEM

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910328210.6 2019-04-23
CN201910328210.6A CN111831799A (zh) 2019-04-23 2019-04-23 信息处理方法、信息处理装置和信息处理系统

Publications (1)

Publication Number Publication Date
WO2020215914A1 true WO2020215914A1 (zh) 2020-10-29

Family

ID=72911470

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/078742 WO2020215914A1 (zh) 2019-04-23 2020-03-11 信息处理方法、信息处理装置和信息处理系统

Country Status (4)

Country Link
US (1) US20220129908A1 (zh)
EP (1) EP3961423A4 (zh)
CN (1) CN111831799A (zh)
WO (1) WO2020215914A1 (zh)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112347770A (zh) * 2020-11-16 2021-02-09 北京京东尚科信息技术有限公司 交互处理方法及装置
CN112380660A (zh) * 2020-11-16 2021-02-19 腾讯科技（北京）有限公司 一种网络布局图配置方法、装置、电子设备及存储介质
CN112417886A (zh) * 2020-11-20 2021-02-26 平安普惠企业管理有限公司 意图实体信息抽取方法、装置、计算机设备及存储介质
CN113609369A (zh) * 2021-08-04 2021-11-05 北京沃东天骏信息技术有限公司 信息推荐方法、电子设备及存储介质
CN113645346A (zh) * 2021-08-11 2021-11-12 中国联合网络通信集团有限公司 功能触发方法、装置、服务器及计算机可读存储介质
CN114022890A (zh) * 2021-11-15 2022-02-08 京东科技信息技术有限公司 针对图像的应答方法和装置

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111611368B (zh) * 2020-05-22 2023-07-04 北京百度网讯科技有限公司 多轮对话中公共场景对话回溯的方法和装置
US11176516B1 (en) * 2020-12-21 2021-11-16 Coupang Corp. Systems and methods for automated information collection and processing
CN113076406B (zh) * 2021-03-19 2024-08-27 联想(北京)有限公司 一种信息处理方法及装置
CN113704432A (zh) * 2021-08-31 2021-11-26 广州方舟信息科技有限公司 一种基于互联网医院的人工智能客服系统构建方法及装置
CN115033676B (zh) * 2022-06-22 2024-04-26 支付宝(杭州)信息技术有限公司 一种意图识别模型训练及用户意图识别方法及装置
CN116578692B (zh) * 2023-07-13 2023-09-15 江西微博科技有限公司 基于大数据的ai智能服务计算方法
CN119293332A (zh) * 2024-10-09 2025-01-10 上海建朗信息科技有限公司 一种基于大模型的辅助办理业务方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103309948A (zh) * 2013-05-20 2013-09-18 携程计算机技术（上海）有限公司 联络中心舆情监控分析和智能分配处理系统及方法
CN107562856A (zh) * 2017-08-28 2018-01-09 深圳追科技有限公司 一种自助式客户服务系统及方法
CN108804643A (zh) * 2018-06-05 2018-11-13 北京玄科技有限公司 多轮会话的流程配置单元、方法及智能机器人交互系统
CN109189898A (zh) * 2018-07-27 2019-01-11 北京新广视通科技有限公司 一种智能应答方法和系统

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7003090B2 (en) * 2001-08-28 2006-02-21 International Business Machines Corporation Method of establishing a communication channel to intelligent support for ebusiness applications
US10455088B2 (en) * 2015-10-21 2019-10-22 Genesys Telecommunications Laboratories, Inc. Dialogue flow optimization and personalization
AU2018273197B2 (en) * 2017-05-22 2021-08-12 Genesys Cloud Services, Inc. System and method for dynamic dialog control for contact center systems

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103309948A (zh) * 2013-05-20 2013-09-18 携程计算机技术（上海）有限公司 联络中心舆情监控分析和智能分配处理系统及方法
CN107562856A (zh) * 2017-08-28 2018-01-09 深圳追科技有限公司 一种自助式客户服务系统及方法
CN108804643A (zh) * 2018-06-05 2018-11-13 北京玄科技有限公司 多轮会话的流程配置单元、方法及智能机器人交互系统
CN109189898A (zh) * 2018-07-27 2019-01-11 北京新广视通科技有限公司 一种智能应答方法和系统

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112347770A (zh) * 2020-11-16 2021-02-09 北京京东尚科信息技术有限公司 交互处理方法及装置
CN112380660A (zh) * 2020-11-16 2021-02-19 腾讯科技（北京）有限公司 一种网络布局图配置方法、装置、电子设备及存储介质
CN112380660B (zh) * 2020-11-16 2024-05-17 腾讯科技（北京）有限公司 一种网络布局图配置方法、装置、电子设备及存储介质
CN112417886A (zh) * 2020-11-20 2021-02-26 平安普惠企业管理有限公司 意图实体信息抽取方法、装置、计算机设备及存储介质
CN113609369A (zh) * 2021-08-04 2021-11-05 北京沃东天骏信息技术有限公司 信息推荐方法、电子设备及存储介质
CN113645346A (zh) * 2021-08-11 2021-11-12 中国联合网络通信集团有限公司 功能触发方法、装置、服务器及计算机可读存储介质
CN114022890A (zh) * 2021-11-15 2022-02-08 京东科技信息技术有限公司 针对图像的应答方法和装置

Also Published As

Publication number Publication date
CN111831799A (zh) 2020-10-27
EP3961423A1 (en) 2022-03-02
US20220129908A1 (en) 2022-04-28
EP3961423A4 (en) 2022-12-21

Similar Documents

Publication Publication Date Title
WO2020215914A1 (zh) 信息处理方法、信息处理装置和信息处理系统
US10489866B2 (en) System and method for providing a social customer care system
US20200059559A1 (en) System and Method for Dialogue Tree Generation
US20200099790A1 (en) Live agent recommendation for a human-robot symbiosis conversation system
CN109190114B (zh) 用于生成回复信息的方法和装置
US20160189558A1 (en) Learning Based on Simulations of Interactions of a Customer Contact Center
US8898157B2 (en) Systems and methods for providing search relevancy in communication initiation searches
CN113810265B (zh) 用于消息插入和引导的系统和方法
US11341506B2 (en) Automated social media queuing system
US20190130413A1 (en) Service ticket assignment and collaboration
US20220366427A1 (en) Systems and methods relating to artificial intelligence long-tail growth through gig customer service leverage
US11763318B2 (en) Systems and methods relating to providing chat services to customers
CN113379229B (zh) 一种资源调度方法和装置
CA2965457C (en) Computer-implemented system and method for providing on-demand expert advice to a consumer
US11595526B2 (en) Systems and methods of contact center client assignment leveraging quantum computation
KR20190058257A (ko) 챗봇 트레이너 플랫폼 및 그 운영 방법
CN113704432A (zh) 一种基于互联网医院的人工智能客服系统构建方法及装置
CN119129646A (zh) 信息交互方法、装置、电子设备和存储介质
CN111046151A (zh) 一种消息处理方法及装置
CN111782776A (zh) 一种通过填槽实现意图识别的方法和装置
US11659089B1 (en) Systems and methods relating to managing cross-channel interactions in contact centers
US20240338710A1 (en) Real-time assistance for a customer at a point of decision through hardware and software smart indicators deterministically generated through artificial intelligence
CN119441414A (zh) 用于提供在线客户支持的系统和方法
CN115409486A (zh) 自动关单处理方法、装置、计算机系统及可读存储介质
US20140189509A1 (en) Passive interaction guide system and method

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20795968

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2020795968

Country of ref document: EP

Effective date: 20211123