CN113553412B - Question-answering processing method, question-answering processing device, electronic equipment and storage medium - Google Patents
Question-answering processing method, question-answering processing device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN113553412B CN113553412B CN202110738564.5A CN202110738564A CN113553412B CN 113553412 B CN113553412 B CN 113553412B CN 202110738564 A CN202110738564 A CN 202110738564A CN 113553412 B CN113553412 B CN 113553412B
- Authority
- CN
- China
- Prior art keywords
- word
- question
- query statement
- questions
- answer
- Prior art date
- Legal status (The legal status 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 status listed.)
- Active
Links
- 238000012545 processing Methods 0.000 title claims abstract description 39
- 238000003672 processing method Methods 0.000 title claims abstract description 24
- 238000012549 training Methods 0.000 claims abstract description 17
- 239000013598 vector Substances 0.000 claims description 112
- 230000011218 segmentation Effects 0.000 claims description 33
- 238000000034 method Methods 0.000 claims description 21
- 238000013507 mapping Methods 0.000 claims description 8
- 238000000605 extraction Methods 0.000 claims description 5
- 239000012634 fragment Substances 0.000 claims description 3
- 238000013473 artificial intelligence Methods 0.000 abstract description 8
- 238000013135 deep learning Methods 0.000 abstract description 6
- 238000003058 natural language processing Methods 0.000 abstract description 6
- 230000000694 effects Effects 0.000 abstract description 4
- 238000004590 computer program Methods 0.000 description 12
- 238000005516 engineering process Methods 0.000 description 11
- 238000010586 diagram Methods 0.000 description 9
- 238000004891 communication Methods 0.000 description 8
- 230000003993 interaction Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- 238000011176 pooling Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 229910002056 binary alloy Inorganic materials 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000009411 base construction Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 230000003924 mental process Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
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
- 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
- G06F16/3322—Query formulation using system suggestions
-
- 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
- G06F16/3331—Query processing
- G06F16/3332—Query translation
- G06F16/3334—Selection or weighting of terms from queries, including natural language queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Databases & Information Systems (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application discloses a question-answering processing method, a question-answering processing device, electronic equipment and a storage medium, relates to the technical field of computers, and particularly relates to the field of artificial intelligence such as natural language processing and deep learning. The implementation scheme is as follows: acquiring a first query statement and a historical query statement which are currently input by a user; according to the first query statement and the historical query statement, a plurality of candidate questions are obtained from a preset question-answer set; inputting each candidate problem, the first query statement and the historical query statement into a network model generated by training to obtain a first matching degree between each candidate problem and the first query statement and the historical query statement; extracting target questions from the plurality of candidate questions according to each first matching degree; and obtaining the answer corresponding to the target question from the question-answer set. Therefore, when a plurality of candidate questions are obtained from the question-answering set and the matching degree is obtained by using the network model, the history query sentences are fully considered, the question recall effect is improved, and the accuracy of answers returned by the question-answering system is further improved.
Description
Technical Field
The application relates to the technical field of computers, in particular to the field of artificial intelligence such as natural language processing, deep learning and the like, and specifically relates to a question-answering processing method, a question-answering processing device, electronic equipment and a storage medium.
Background
With the development of computer technology and internet technology, man-machine interaction is increasingly widely applied to life of people. The intelligent question-answering system is one of core technologies of man-machine interaction. When a user inputs a question to be queried, the question-answering system can recall the most similar question according to the matching degree of the questions in the question-answering set and the questions queried by the user, and then the answer corresponding to the question is displayed.
Therefore, how to improve the accuracy of recall problems is a highly desirable problem.
Disclosure of Invention
The application provides a question and answer processing method, a question and answer processing device, electronic equipment and a storage medium.
According to an aspect of the present application, there is provided a question-answering processing method, including:
acquiring a first query statement and a historical query statement which are currently input by a user;
acquiring a plurality of candidate questions from a preset question-answer set according to the first query sentence and the historical query sentence, wherein the question-answer set comprises a plurality of question-answer pairs, and each question-answer pair comprises a question and a corresponding answer;
Inputting each candidate problem, the first query statement and the historical query statement into a network model generated through training to obtain a first matching degree between each candidate problem and the first query statement and the historical query statement;
extracting target questions from the plurality of candidate questions according to each first matching degree;
and obtaining answers corresponding to the target questions from the question-answer set.
According to another aspect of the present application, there is provided a question-answering processing apparatus including:
the first acquisition module is used for acquiring a first query statement and a historical query statement which are currently input by a user;
the second acquisition module is used for acquiring a plurality of candidate questions from a preset question-answer set according to the first query statement and the historical query statement, wherein the question-answer set comprises a plurality of question-answer pairs, and each question-answer pair comprises a question and a corresponding answer;
the third obtaining module is used for inputting each candidate problem, the first query statement and the historical query statement into a network model generated by training so as to obtain a first matching degree between each candidate problem and the first query statement and the historical query statement;
The extraction module is used for extracting target questions from the plurality of candidate questions according to each first matching degree;
and the fourth acquisition module is used for acquiring answers corresponding to the target questions from the question-answer set.
According to another aspect of the present application, there is provided an electronic device including:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of the above embodiments.
According to another aspect of the present application, there is provided a non-transitory computer readable storage medium storing computer instructions for causing the computer to perform the method according to the above-described embodiments.
According to another aspect of the present application, there is provided a computer program product comprising a computer program which, when executed by a processor, implements a method according to the above embodiments.
It should be understood that the description of this section is not intended to identify key or critical features of the embodiments of the application or to delineate the scope of the application. Other features of the present application will become apparent from the description that follows.
Drawings
The drawings are for better understanding of the present solution and do not constitute a limitation of the present application. Wherein:
fig. 1 is a flow chart of a question-answering processing method provided in an embodiment of the present application;
fig. 2 is a flow chart of another question-answering processing method according to an embodiment of the present application;
FIG. 3 is a schematic process diagram of a network model according to an embodiment of the present disclosure;
fig. 4 is a flow chart of another question-answering processing method according to an embodiment of the present application;
FIG. 5 is a schematic diagram of obtaining a plurality of candidate questions according to an embodiment of the present application;
fig. 6 is a schematic structural diagram of a question-answering processing device according to an embodiment of the present application;
fig. 7 is a block diagram of an electronic device for implementing the question-answering processing method of the embodiment of the present application.
Detailed Description
Exemplary embodiments of the present application are described below in conjunction with the accompanying drawings, which include various details of the embodiments of the present application to facilitate understanding, and should be considered as merely exemplary. Accordingly, one of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present application. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
The following describes a question-answering processing method, an apparatus, an electronic device, and a storage medium according to embodiments of the present application with reference to the accompanying drawings.
Artificial intelligence is the discipline of studying certain mental processes and intelligent behaviors (e.g., learning, reasoning, thinking, planning, etc.) of a person using a computer, both in the technical field of hardware and in the technical field of software. Artificial intelligence hardware technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing, and the like; the artificial intelligence software technology comprises a computer vision technology, a voice recognition technology, a natural language processing technology, a deep learning technology, a big data processing technology, a knowledge graph technology and the like.
NLP (Natural Language Processing ) is an important direction in the fields of computer science and artificial intelligence, and the content of NLP research includes, but is not limited to, the following branch fields: text classification, information extraction, automatic abstracting, intelligent question and answer, topic recommendation, machine translation, topic word recognition, knowledge base construction, deep text representation, named entity recognition, text generation, text analysis (lexical, syntactic, grammatical, etc.), speech recognition and synthesis, and the like.
Deep learning is a new research direction in the field of machine learning. Deep learning is the inherent regularity and presentation hierarchy of learning sample data, and the information obtained during such learning is helpful in interpreting data such as text, images and sounds. Its final goal is to have the machine have analytical learning capabilities like a person, and to recognize text, image, and sound data.
Fig. 1 is a flow chart of a question-answering processing method according to an embodiment of the present application.
The question-answering processing method can be executed by the question-answering processing device, and the device can be configured into the electronic equipment, and the key information which is possibly omitted by the currently input query statement is supplemented by fully considering the dialogue history when the question is recalled, so that the question recall effect of a question-answering system is improved, and the accuracy of returned answers is further improved.
As shown in fig. 1, the question-answering processing method includes:
step 101, a first query sentence and a historical query sentence which are currently input by a user are obtained.
The question-answering processing method can be applied to a question-answering system, such as a FAQ (Frequently Asked Questions, common question-answering) system. When a user enters a query term in the question-answering system, the question-answering system may obtain the first query term currently entered by the user in order to facilitate distinguishing what is referred to as the first query term. For example, a question and answer system provided by a user at an airport enters the query sentence "XX flight Point take off".
In the application, when a user inputs a first query sentence, the user can input the first query sentence through characters or through voice, and after the question-answering system acquires the voice input by the user, the acquired voice can be identified through a voice identification function so as to acquire the first query sentence.
In the practical application of the question-answering system, the user may omit some information according to the context, and based on this, the question-answering system in the application can also acquire historical query sentences.
Generally, the closer the time interval between the historical query statement and the first query statement, the greater the relevance of the historical query statement to the first query statement. For example, the relevance between the first query statement of the current round of dialogue and the historical query statement in the penultimate round of dialogue is higher than the relevance to the historical query statement in the penultimate round of dialogue 5. Then, when the historical query statement is obtained, N historical query statements before the first query statement may be obtained, where N is a positive integer.
Step 102, obtaining a plurality of candidate questions from a preset question-answer set according to the first query statement and the historical query statement.
In the application, a question-answer set is preset in a question-answer system, wherein the question-answer set comprises a plurality of question-answer pairs, and each question-answer pair comprises a question and a corresponding answer.
Because the question-answer set contains a plurality of question-answer pairs, in order to improve the processing efficiency, after the first query sentence and the historical query sentence currently input by the user are acquired, a plurality of candidate questions can be acquired from the question-answer set according to the first query sentence and the historical query sentence.
When a plurality of candidate questions are acquired, the similarity between the first query sentence and each historical query sentence can be calculated, and the historical query sentence with higher similarity with the first query sentence is determined according to the similarity. And then, calculating the similarity between the questions in each question-answer pair in the question set and the screened historical query sentences, and extracting a plurality of candidate questions from a plurality of question-answer pairs according to the first query sentences and the historical query sentences with higher similarity. Therefore, the dialogue history is considered by extracting a plurality of candidate questions based on the historical query statement and the first query statement, and screening accuracy of the candidate questions is improved.
Step 103, inputting each candidate question, the first query statement and the historical query statement into the trained network model to obtain a first matching degree between each candidate question and the first query statement and the historical query statement.
In the application, after a plurality of candidate questions are obtained, each candidate question, the first query statement and the historical query statement can be input into a network model generated through training, and encoding and decoding processing is performed by the network model so as to obtain a first matching degree between each candidate question and the first query statement and the historical query statement.
The network model may be generated through training in a deep learning manner, and the higher the first matching degree obtained through the network model, the higher the similarity between the candidate problem and the first query statement is.
If the obtained historical query sentences are multiple, when the first matching degree corresponding to each candidate problem is obtained, the multiple historical query sentences are input into the network model. Thus, the information contained in the query statement in the history dialogue can be fully considered when the problem is recalled.
Step 104, extracting a target question from the plurality of candidate questions according to each first matching degree.
After the first matching degree corresponding to each candidate problem is obtained, the target problem can be extracted from the plurality of candidate problems according to the first matching degree corresponding to each candidate problem. The target problem may be one or a plurality of target problems.
For example, a candidate problem having the highest first matching degree may be extracted from a plurality of candidate problems as the target problem. Alternatively, a candidate problem having a first degree of matching greater than a threshold may be regarded as the target problem.
Or, the plurality of candidate questions are ranked according to the first matching degree, and the preset number of candidate questions can be used as target questions.
Step 105, obtaining the answer corresponding to the target question from the question-answer set.
Because the target questions are questions in the question-answer set, after the target questions are acquired, answers corresponding to the target questions can be acquired from the question-answer set. If the target questions are one, the answer can be returned directly, and if the target questions are a plurality of, each target question and the corresponding answer can be returned for the user to select.
In the embodiment of the application, a plurality of candidate questions are obtained from a preset question-answer set according to a first query statement and a historical query statement input by a user through obtaining the first query statement and the historical query statement, each candidate question, the first query statement and the historical query statement are input into a network model generated through training, so that a first matching degree between each candidate question and the first query statement and the first matching degree between each candidate question and the historical query statement are obtained, a target question is extracted from the plurality of candidate questions according to each first matching degree, and answers corresponding to the target questions are obtained from the question-answer set. Therefore, when a plurality of candidate questions are obtained from the question-answering set and the matching degree is obtained by using the network model, the history query sentences are fully considered, so that the question recall effect is improved, and the accuracy of answers returned by the question-answering system is further improved.
In one embodiment of the present application, the first matching degree between each candidate problem and the first query statement and the historical query statement may be obtained through the network model in the manner described in fig. 2. Fig. 2 is a flow chart of another question-answering processing method according to an embodiment of the present application.
As shown in fig. 2, the step of inputting each candidate question, the first query sentence and the historical query sentence into the training generated network model to obtain a first matching degree between each candidate question and the first query sentence and the historical query sentence includes:
in step 201, vector mapping is performed on each word in each candidate question, the first query statement and the historical query statement, so as to obtain a word vector corresponding to each word.
In the application, when vector mapping is performed on each word, a one-bit effective coding mode can be adopted, and a word steering amount model trained in advance can also be adopted to obtain the word vector of each word.
Step 202, encoding the position of each word in the sentence to obtain a first position vector corresponding to each word.
In the present application, each word may be position-coded according to the position of each word in the sentence, so as to obtain the first position vector. For example, the first query term "how weather today" may be counted from 0, where each word corresponds to 0, 1, 2, 3, 4, 5, respectively, and then each number may be converted into a binary system with the same number of bits, so as to obtain the first position vector corresponding to each word.
Step 203, obtaining a second position vector corresponding to each word according to the sentence where each word is located.
Because there are multiple sentences such as candidate questions, first query sentences, historical query sentences and the like, the position information of each word comprises the sentence where the word is located and the position of the sentence, so that in the application, the second position vector corresponding to each word can be obtained according to the sentence where the word is located. That is, the second position vector is used to represent the sentence in which the word is located, and the word second position vector in the same sentence is the same.
As an implementation manner, mapping relations between different sentence types and vectors may be predefined, for example, a candidate problem corresponding vector a, a vector B corresponding to a first query sentence currently input, and a vector C corresponding to a history query sentence are defined. If the sentence of a word is a candidate problem, determining the second position vector of the word as A, if the sentence of the word is a first query sentence, determining the second position vector of the word as B, and if the sentence of the word is a historical query sentence, determining the second position vector of the word as C.
If the historical query sentence is a plurality of, the longer the distance between the historical query sentence and the first query sentence is, the smaller the correlation between the historical query sentence and the first query sentence is. As another possible implementation manner, in the case that the sentence in which any word is located is a candidate problem, the second position vector corresponding to any word may be determined according to the candidate problem; when the sentence in which any word is located is a first query sentence, a second position vector corresponding to any word can be determined according to the first query sentence; when the sentence in which any word is located is a history query sentence, the second position vector corresponding to any word may be determined according to the distance between the history query sentence and the first query sentence. Wherein the separation distance may be a time interval length.
For example, the type value corresponding to the candidate problem is 0, the type value corresponding to the first query sentence is 1, the type value corresponding to the historical query sentence of the first-to-last dialog is 2, and the other steps are analogized, so that the type value corresponding to the sentence where each word is located can be converted into a binary system with the same number of bits, and the binary system is used as the second position vector corresponding to each word. It will be appreciated that the larger the type value, the farther apart the historical query statement is from the first query statement, and the less relevant the historical query statement is to the first query statement.
When the second position vector corresponding to each word is determined, the second position vector corresponding to the word is determined according to the interval distance between the historical query statement and the first query statement under the condition that the statement in which one word is located is the historical query statement, so that the influence of the distance between the historical query statement and the first query statement on the first matching degree is considered.
Step 204, determining a vector representation corresponding to each word according to the word vector, the first position vector and the second position vector.
In the application, the word vector, the first position vector and the second position vector of each word can be spliced to obtain the vector representation corresponding to each word. That is, the vector representation for each word includes a word vector, a first position vector, and a second position vector for each word.
Since the word vector width of each word is the same, the first position vector width of each word is the same, and the second position vector width of each word is the same, the width of the vector representation of each word is also the same.
For example, the widths of the word vector, the first position vector, and the second position vector are l1, l2, and l3, respectively, and then the vector representation of each word has a width of k=l1+l2+l3, i.e., the vector representation of each word is a matrix of 1*k.
Step 205, inputting the vector representation corresponding to each word into the training generated network model to obtain the first matching degree between each candidate question and the first query sentence and the historical query sentence.
In the application, the vector representations corresponding to the words can be input into a network model generated by training, the network model can encode the vector representations corresponding to the words to obtain a first feature vector, and the feature vector can represent the interaction relation among each candidate problem, the first query statement and the historical query statement. And then, carrying out average pooling processing on the first feature vector to obtain a second feature vector, and converting the second feature vector into the first matching degree through a full connection layer.
Fig. 3 is a schematic processing diagram of a network model according to an embodiment of the present application. In fig. 3, candidate questions ci (i=1, 2,3, …, m), a first query statement q0, and history query statements q1, q2, …, qn are input to an embedding layer, the embedding layer outputs vector representations corresponding to each word, and the vector representations of each word are input to a conversion network (for example, a two-layer converter network), the network processes the vector representations of each word, outputs a first feature vector, the first feature vector is input to an average pooling layer for average pooling processing, a second feature vector is obtained, and the second feature vector is converted by a full connection layer to obtain a first matching degree between ci and q0, q1, q2, …, qn.
For example, the vector corresponding to each word is represented as a matrix 1*k, the ci, q0, q1, q2, …, qn have h words in total, the first eigenvector is a matrix h x k, the elements on the same column are added and averaged, that is, the second eigenvector obtained by average pooling is a matrix 1*k, so that all the information of ci, q0, q1, q2, …, qn is encoded into a vector 1*k, and the current matching state is represented by a single vector.
In the embodiment of the application, the word coding and the position coding are performed on each word in each candidate question, the first query statement and the historical query statement, the vector representation corresponding to each word is determined, the vector representations corresponding to each word are input into the network model, so that the first matching degree between each candidate question and the first query statement and the first matching degree between each candidate question and the historical query statement are obtained, and therefore the target question can be selected from a plurality of candidate questions by utilizing the first matching degree corresponding to each candidate question, and the answer corresponding to the target question is obtained.
In one embodiment of the present application, when obtaining a plurality of candidate questions, the plurality of candidate questions may also be extracted from the plurality of question-answer pairs by segmenting words in the first query statement and the historical query statement. Fig. 4 is a schematic flow chart of another question-answering processing method according to an embodiment of the present application.
As shown in fig. 4, the obtaining a plurality of candidate questions from a preset question-answer set according to the first query sentence and the historical query sentence may include:
step 401, performing word segmentation processing on the first query sentence and the historical query sentence respectively to obtain a word segmentation set corresponding to the first query sentence.
In the application, word segmentation processing can be performed on the first query statement and the historical query statement respectively to obtain each word segment contained in the first query statement and each word segment contained in the historical query statement, the obtained each word segment is subjected to de-duplication processing, and each word segment obtained after processing forms a word segment set.
Step 402, extracting a plurality of candidate questions from a plurality of question-answer pairs according to a second matching degree between each first word in the word segmentation set and the questions in each question-answer pair.
For each first word in the word segmentation set, calculating a second matching degree between each first word and the question in each question-answer set, for example, a distance between a vector corresponding to the first word and a vector corresponding to the question may be used as the second matching degree, an average value of the second matching degrees between each question and each first word is calculated, the questions in the question-answer set are ranked according to the average value of the second matching degrees corresponding to each question, and a plurality of candidate questions are extracted from a plurality of question-answer sets. For example, a pre-set number of questions with a higher average of the second matching degree may be extracted as candidate questions.
For example, there are 100 question-answer pairs in the question-answer set, and each question-answer pair contains one question, that is, there are 100 questions in the question-answer set, and the first 20 questions with the highest average value of the second matching degree can be extracted from the 100 questions as candidate questions.
Because the importance of each word in the query sentence is different, when a plurality of candidate questions are extracted from a plurality of question-answer pairs according to the second matching degree between each word and the question in each question-answer pair, the weight of each first word set in the word sets can be determined, and the word with higher weight is extracted from the word sets as a target word according to the weight of each first word, and a plurality of candidate questions are extracted from the plurality of question-answer pairs based on the second matching degree between the question in each question-answer pair and the target word. Therefore, the target word is screened out according to the weight of the first word, and the target word is utilized to perform rough-ranking search to obtain a plurality of candidate problems, so that the processing efficiency can be improved.
When determining the weight of each first word segment, the weight of each first word segment can be obtained by utilizing a model generated through training. Alternatively, the weight of the first word may be determined according to the number of occurrences of each first word in each question in the question-answering set, where the greater the number of occurrences, the greater the weight.
After the target word is obtained, a third matching degree between each second word segment included in the question in each question-answer pair and each target word segment can be calculated, and the number of target word segments included in each question is determined according to the third matching degree between each second word segment included in each question and each target word segment. For example, if the third matching degree between the target word and the second word exceeds the set threshold, the target word may be considered to be included in the question.
After determining the number of target tokens included in the questions in each question-answer pair, the questions in the question-answer pairs may be ranked according to the number of target tokens included in the questions in each question-answer pair and the weight of the included target tokens, and a plurality of candidate questions may be extracted from the questions.
For example, the plurality of questions may be ranked from large to small according to the number of target tokens included in each question, and if the number of target tokens included is the same, the sum of the weights of the included target tokens is compared, and the greater the sum of the weights, the more forward in the ranking. And then selecting the problems of the preset quantity from the sorting as candidate problems.
In the method, when extracting a plurality of questions from a plurality of question-answer pairs, a plurality of candidate questions can be extracted according to the number of target word segments and the weight of the target word segments contained in each question-answer pair, and the operation is simple.
In the embodiment of the present application, when a plurality of candidate questions are obtained from a preset question-answer set, word segmentation processing may be performed on a first query sentence and a history query sentence, so as to obtain a word segmentation set corresponding to the first query sentence, and according to a second matching degree between each first word segment in the word segmentation set and a question in each question-answer set, a plurality of candidate questions may be extracted from the plurality of question-answer pairs. Therefore, by utilizing the word segmentation in the first query statement and the historical query statement, not only can a plurality of candidate questions be obtained efficiently, but also the query statement in the dialogue history is considered, and the accuracy of the obtained candidate questions is improved.
After the above method is used to obtain a plurality of candidate questions, each candidate question, the first query statement and the historical query statement may be input into a training generated network model to obtain a first matching degree between each candidate question and the first query statement and the historical query statement, and the specific process may refer to the above embodiment and will not be repeated herein. After the first matching degree corresponding to each candidate question is obtained, the target question from the plurality of candidate questions can be obtained, the answer of the target question obtained from the question-answer set can be obtained, and the answer is returned to the user.
Fig. 5 is a schematic diagram of obtaining a plurality of candidate questions according to an embodiment of the present application. In fig. 5, the current first query sentence q0 and n historical query sentences q1, q2, …, qn are subjected to word segmentation processing to obtain a plurality of segmented words, and coarse-ranking search is performed by using the segmented words to obtain coarse-ranking candidate results c1, c2, …, cm, namely m questions are extracted from a plurality of question-answer pairs. The rough-ranking search using the word segmentation here refers to extracting a plurality of questions from a plurality of question-answer pairs using the word segmentation.
In order to achieve the above embodiments, the embodiments of the present application further provide a question-answering processing device. Fig. 6 is a schematic structural diagram of a question-answering processing device according to an embodiment of the present application.
As shown in fig. 6, the question-answering processing apparatus 600 includes:
a first obtaining module 610, configured to obtain a first query term and a historical query term currently input by a user;
a second obtaining module 620, configured to obtain a plurality of candidate questions from a preset question-answer set according to the first query sentence and the historical query sentence, where the question-answer set includes a plurality of question-answer pairs, and each question-answer pair includes a question and a corresponding answer;
a third obtaining module 630, configured to input each of the candidate questions, the first query sentence, and the historical query sentence into a training generated network model, so as to obtain a first matching degree between each of the candidate questions and the first query sentence and the historical query sentence;
An extracting module 640, configured to extract a target problem from the plurality of candidate problems according to each of the first matching degrees;
and a fourth obtaining module 650, configured to obtain an answer corresponding to the target question from the question-answer set.
In one possible implementation manner of the embodiment of the present application, the third obtaining module 630 includes:
the vector mapping unit is used for carrying out vector mapping on each word in each candidate problem, the first query statement and the historical query statement so as to obtain a word vector corresponding to each word;
the first acquisition unit is used for encoding the position of each word in the sentence so as to acquire a first position vector corresponding to each word;
the second acquisition unit is used for acquiring a second position vector corresponding to each word according to the statement of each word;
the determining unit is used for determining a vector representation corresponding to each word according to the word vector, the first position vector and the second position vector;
and the third acquisition unit is used for inputting the vector representation corresponding to each word into the training generated network model so as to acquire the first matching degree between each candidate problem and the first query statement and the historical query statement.
In a possible implementation manner of the embodiment of the present application, the determining unit is configured to:
under the condition that the statement in which any word is located is the candidate problem, determining a second position vector corresponding to the any word according to the candidate problem;
determining a second position vector corresponding to any word according to the first query statement when the statement in which the word is located is the first query statement;
and under the condition that the statement in which any word is located is the historical query statement, determining a second position vector corresponding to the any word according to the interval distance between the historical query statement and the first query statement.
In one possible implementation manner of the embodiment of the present application, the second obtaining module 620 includes:
the word segmentation processing unit is used for respectively carrying out word segmentation processing on the first query statement and the historical query statement so as to obtain a word segmentation set corresponding to the first query statement;
and the extraction unit is used for extracting a plurality of candidate questions from the plurality of question-answer pairs according to the second matching degree between each first word in the word segmentation set and the questions in each question-answer pair.
In a possible implementation manner of the embodiment of the present application, the extracting unit is configured to:
Determining the weight of each first word in the word segmentation set;
extracting target word segmentation from the word segmentation set according to each weight;
and extracting the candidate questions from the question-answer pairs according to a second matching degree between the questions in the question-answer pairs and the target word.
In a possible implementation manner of the embodiment of the present application, the extracting unit is further configured to:
determining the number of target words contained in the questions in each question-answer pair according to the third matching degree between each second word contained in the questions in each question-answer pair and each target word;
and extracting the candidate questions from the question-answer pairs according to the number of target word fragments contained in the questions in each question-answer pair and the corresponding weights.
Note that, the explanation of the foregoing embodiments of the question-answering processing method is also applicable to the question-answering processing apparatus of this embodiment, and therefore will not be described herein.
In the embodiment of the application, a plurality of candidate questions are obtained from a preset question-answer set according to a first query statement and a historical query statement input by a user through obtaining the first query statement and the historical query statement, each candidate question, the first query statement and the historical query statement are input into a network model generated through training, so that a first matching degree between each candidate question and the first query statement and the first matching degree between each candidate question and the historical query statement are obtained, a target question is extracted from the plurality of candidate questions according to each first matching degree, and answers corresponding to the target questions are obtained from the question-answer set. Therefore, when a plurality of candidate questions are obtained from the question-answering set and the matching degree is obtained by using the network model, the history query sentences are fully considered, so that the question recall effect is improved, and the accuracy of answers returned by the question-answering system is further improved.
According to embodiments of the present application, there is also provided an electronic device, a readable storage medium and a computer program product.
Fig. 7 shows a schematic block diagram of an example electronic device 700 that may be used to implement embodiments of the present application. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the application described and/or claimed herein.
As shown in fig. 7, the apparatus 700 includes a computing unit 701 that can perform various appropriate actions and processes according to a computer program stored in a ROM (Read-Only Memory) 702 or a computer program loaded from a storage unit 708 into a RAM (Random Access Memory ) 703. In the RAM 703, various programs and data required for the operation of the device 700 may also be stored. The computing unit 701, the ROM 702, and the RAM 703 are connected to each other through a bus 704. An I/O (Input/Output) interface 705 is also connected to bus 704.
Various components in device 700 are connected to I/O interface 705, including: an input unit 706 such as a keyboard, a mouse, etc.; an output unit 707 such as various types of displays, speakers, and the like; a storage unit 708 such as a magnetic disk, an optical disk, or the like; and a communication unit 709 such as a network card, modem, wireless communication transceiver, etc. The communication unit 709 allows the device 700 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunication networks.
The computing unit 701 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of computing unit 701 include, but are not limited to, a CPU (Central Processing Unit ), a GPU (Graphic Processing Units, graphics processing unit), various dedicated AI (Artificial Intelligence ) computing chips, various computing units running machine learning model algorithms, a DSP (Digital Signal Processor ), and any suitable processor, controller, microcontroller, etc. The computing unit 701 performs the respective methods and processes described above, such as a question-answer processing method. For example, in some embodiments, the question-answering processing method may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as storage unit 708. In some embodiments, part or all of the computer program may be loaded and/or installed onto device 700 via ROM 702 and/or communication unit 709. When a computer program is loaded into the RAM 703 and executed by the computing unit 701, one or more steps of the question-answering processing method described above can be performed. Alternatively, in other embodiments, the computing unit 701 may be configured to perform the question-answer processing method by any other suitable means (e.g. by means of firmware).
Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuit System, FPGA (Field Programmable Gate Array ), ASIC (Application-Specific Integrated Circuit, application-specific integrated circuit), ASSP (Application Specific Standard Product, special-purpose standard product), SOC (System On Chip ), CPLD (Complex Programmable Logic Device, complex programmable logic device), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.
Program code for carrying out methods of the present application may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of this application, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, RAM, ROM, EPROM (Electrically Programmable Read-Only-Memory, erasable programmable read-Only Memory) or flash Memory, an optical fiber, a CD-ROM (Compact Disc Read-Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., CRT (Cathode-Ray Tube) or LCD (Liquid Crystal Display ) monitor) for displaying information to a user; and a keyboard and pointing device (e.g., a mouse or trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic input, speech input, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: LAN (Local Area Network ), WAN (Wide Area Network, wide area network), internet and blockchain networks.
The computer system may include a client and a server. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also called a cloud computing server or a cloud host, and is a host product in a cloud computing service system, so that the defects of high management difficulty and weak service expansibility in the traditional physical hosts and VPS service (Virtual Private Server, virtual special servers) are overcome. The server may also be a server of a distributed system or a server that incorporates a blockchain.
According to an embodiment of the present application, there is further provided a computer program product, which when executed by an instruction processor in the computer program product, performs the question-answering processing method according to the above embodiment of the present application.
It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps described in the present disclosure may be performed in parallel, sequentially, or in a different order, provided that the desired results of the technical solutions disclosed in the present application are achieved, and are not limited herein.
The above embodiments do not limit the scope of the application. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives are possible, depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application are intended to be included within the scope of the present application.
Claims (12)
1. A question-answering processing method comprises the following steps:
acquiring a first query statement and a historical query statement which are currently input by a user;
acquiring a plurality of candidate questions from a preset question-answer set according to the first query sentence and the historical query sentence, wherein the question-answer set comprises a plurality of question-answer pairs, and each question-answer pair comprises a question and a corresponding answer;
Vector mapping is carried out on each word in each candidate problem, the first query statement and the historical query statement so as to obtain a word vector corresponding to each word; encoding the position of each word in the sentence to obtain a first position vector corresponding to each word; acquiring a second position vector corresponding to each word according to the sentence in which each word is located, wherein when the sentence in which any word is located is the historical query sentence, the second position vector corresponding to any word is determined according to the interval distance between the historical query sentence and the first query sentence; determining a vector representation corresponding to each word according to the word vector, the first position vector and the second position vector; inputting vector representations corresponding to each word into a network model generated by training to obtain a first matching degree between each candidate problem and the first query statement and the historical query statement;
extracting target questions from the plurality of candidate questions according to each first matching degree;
and obtaining answers corresponding to the target questions from the question-answer set.
2. The method of claim 1, wherein the obtaining the second position vector corresponding to each word according to the sentence in which each word is located includes:
Under the condition that the statement in which any word is located is the candidate problem, determining a second position vector corresponding to the any word according to the candidate problem;
determining a second position vector corresponding to any word according to the first query statement when the statement in which the word is located is the first query statement;
and under the condition that the statement in which any word is located is the historical query statement, determining a second position vector corresponding to the any word according to the interval distance between the historical query statement and the first query statement.
3. The method of any one of claims 1-2, wherein the obtaining, from the first query statement and the historical query statement, a plurality of candidate questions from a preset question-answer set includes:
performing word segmentation processing on the first query statement and the historical query statement respectively to obtain a word segmentation set corresponding to the first query statement;
and extracting a plurality of candidate questions from the plurality of question-answer pairs according to a second matching degree between each first word in the word segmentation set and the questions in each question-answer pair.
4. The method of claim 3, wherein said extracting a plurality of candidate questions from said plurality of question-answer pairs based on a second degree of matching between each of said tokens and questions in each of said question-answer pairs comprises:
Determining the weight of each first word in the word segmentation set;
extracting target word segmentation from the word segmentation set according to each weight;
and extracting the candidate questions from the question-answer pairs according to a second matching degree between the questions in the question-answer pairs and the target word.
5. The method of claim 4, wherein after extracting the target word from the word segmentation set according to each of the weights, further comprising:
determining the number of target words contained in the questions in each question-answer pair according to the third matching degree between each second word contained in the questions in each question-answer pair and each target word;
and extracting the candidate questions from the question-answer pairs according to the number of target word fragments contained in the questions in each question-answer pair and the corresponding weights.
6. A question-answering processing apparatus comprising:
the first acquisition module is used for acquiring a first query statement and a historical query statement which are currently input by a user;
the second acquisition module is used for acquiring a plurality of candidate questions from a preset question-answer set according to the first query statement and the historical query statement, wherein the question-answer set comprises a plurality of question-answer pairs, and each question-answer pair comprises a question and a corresponding answer;
The third obtaining module is used for inputting each candidate problem, the first query statement and the historical query statement into a network model generated by training so as to obtain a first matching degree between each candidate problem and the first query statement and the historical query statement;
the extraction module is used for extracting target questions from the plurality of candidate questions according to each first matching degree;
a fourth obtaining module, configured to obtain an answer corresponding to the target question from the question-answer set;
wherein, the third acquisition module includes:
the vector mapping unit is used for carrying out vector mapping on each word in each candidate problem, the first query statement and the historical query statement so as to obtain a word vector corresponding to each word;
the first acquisition unit is used for encoding the position of each word in the sentence so as to acquire a first position vector corresponding to each word;
a second obtaining unit, configured to obtain a second position vector corresponding to each word according to a sentence where each word is located, where, when the sentence where any word is located is the historical query sentence, determining the second position vector corresponding to any word according to a distance between the historical query sentence and the first query sentence; determining a vector representation corresponding to each word according to the word vector, the first position vector and the second position vector;
The determining unit is used for determining a vector representation corresponding to each word according to the word vector, the first position vector and the second position vector;
and the third acquisition unit is used for inputting the vector representation corresponding to each word into the training generated network model so as to acquire the first matching degree between each candidate problem and the first query statement and the historical query statement.
7. The apparatus of claim 6, wherein the determining unit is configured to:
under the condition that the statement in which any word is located is the candidate problem, determining a second position vector corresponding to the any word according to the candidate problem;
determining a second position vector corresponding to any word according to the first query statement when the statement in which the word is located is the first query statement;
and under the condition that the statement in which any word is located is the historical query statement, determining a second position vector corresponding to the any word according to the interval distance between the historical query statement and the first query statement.
8. The apparatus of any of claims 6-7, wherein the second acquisition module comprises:
The word segmentation processing unit is used for respectively carrying out word segmentation processing on the first query statement and the historical query statement so as to obtain a word segmentation set corresponding to the first query statement;
and the extraction unit is used for extracting a plurality of candidate questions from the plurality of question-answer pairs according to the second matching degree between each first word in the word segmentation set and the questions in each question-answer pair.
9. The apparatus of claim 8, wherein the decimation unit is configured to:
determining the weight of each first word in the word segmentation set;
extracting target word segmentation from the word segmentation set according to each weight;
and extracting the candidate questions from the question-answer pairs according to a second matching degree between the questions in the question-answer pairs and the target word.
10. The apparatus of claim 9, wherein the decimation unit is further configured to:
determining the number of target words contained in the questions in each question-answer pair according to the third matching degree between each second word contained in the questions in each question-answer pair and each target word;
and extracting the candidate questions from the question-answer pairs according to the number of target word fragments contained in the questions in each question-answer pair and the corresponding weights.
11. An electronic device, comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-5.
12. A non-transitory computer readable storage medium storing computer instructions for causing the computer to perform the method of any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110738564.5A CN113553412B (en) | 2021-06-30 | 2021-06-30 | Question-answering processing method, question-answering processing device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110738564.5A CN113553412B (en) | 2021-06-30 | 2021-06-30 | Question-answering processing method, question-answering processing device, electronic equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113553412A CN113553412A (en) | 2021-10-26 |
| CN113553412B true CN113553412B (en) | 2023-07-25 |
Family
ID=78131163
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110738564.5A Active CN113553412B (en) | 2021-06-30 | 2021-06-30 | Question-answering processing method, question-answering processing device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113553412B (en) |
Families Citing this family (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114020774A (en) * | 2021-11-22 | 2022-02-08 | 国家电网有限公司大数据中心 | Method, device and equipment for processing multiple rounds of question-answering sentences and storage medium |
| CN114238562A (en) * | 2021-12-06 | 2022-03-25 | 广东明创软件科技有限公司 | Text processing method and device, electronic equipment and storage medium |
| CN114265925B (en) * | 2021-12-24 | 2025-07-25 | 科大讯飞(苏州)科技有限公司 | Question answering method and device, electronic equipment and storage medium |
| CN114461749B (en) * | 2022-02-15 | 2023-04-07 | 北京百度网讯科技有限公司 | Data processing method and device for conversation content, electronic equipment and medium |
| CN116795953B (en) * | 2022-03-08 | 2024-06-25 | 腾讯科技（深圳）有限公司 | Question-answer matching method and device, computer readable storage medium and computer equipment |
| CN114626387A (en) * | 2022-03-14 | 2022-06-14 | 京东科技信息技术有限公司 | Question sentence processing method and device, computer equipment and storage medium |
| CN114691957A (en) * | 2022-03-25 | 2022-07-01 | 北京百度网讯科技有限公司 | Information processing method, information processing device, electronic equipment and storage medium |
| CN117216183A (en) * | 2022-06-02 | 2023-12-12 | 吉林亿联银行股份有限公司 | Processing method and device of intelligent question-answering system, storage medium and processor |
| CN115952274B (en) * | 2023-03-10 | 2023-06-27 | 北京百度网讯科技有限公司 | Data generation method, training method and device based on deep learning model |
| CN115964471B (en) * | 2023-03-16 | 2023-06-02 | 成都安哲斯生物医药科技有限公司 | Medical data approximate query method |
| CN116992001A (en) * | 2023-08-07 | 2023-11-03 | 北京百度网讯科技有限公司 | Knowledge question and answer method, device, equipment, speaker and storage medium |
| CN117312530B (en) * | 2023-11-10 | 2024-09-03 | 北京百度网讯科技有限公司 | Questionnaire and model training method, device, equipment, medium and product |
| CN117493595B (en) * | 2023-12-19 | 2025-03-25 | 北京百度网讯科技有限公司 | Image search method, device, equipment and medium based on large model |
| CN118052282B (en) * | 2024-02-06 | 2025-02-28 | 北京月之暗面科技有限公司 | A large language reasoning system and method |
| CN119917528B (en) * | 2025-04-01 | 2025-06-24 | 卓世科技(海南)有限公司 | SQL sentence generation method and device |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110096580A (en) * | 2019-04-24 | 2019-08-06 | 北京百度网讯科技有限公司 | A kind of FAQ dialogue method, device and electronic equipment |
| CN110309283A (en) * | 2019-06-28 | 2019-10-08 | 阿里巴巴集团控股有限公司 | A kind of answer of intelligent answer determines method and device |
| CN110597966A (en) * | 2018-05-23 | 2019-12-20 | 北京国双科技有限公司 | Automatic question answering method and device |
| CN111046159A (en) * | 2019-12-13 | 2020-04-21 | 杭州翼兔网络科技有限公司 | Man-machine interaction method, device, equipment and medium |
| CN111241237A (en) * | 2019-12-31 | 2020-06-05 | 中国建设银行股份有限公司 | Intelligent question and answer data processing method and device based on operation and maintenance service |
| CN111428010A (en) * | 2019-01-10 | 2020-07-17 | 北京京东尚科信息技术有限公司 | Man-machine intelligent question and answer method and device |
| CN111858859A (en) * | 2019-04-01 | 2020-10-30 | 北京百度网讯科技有限公司 | Automatic question-answering processing method, device, computer equipment and storage medium |
| CN111898643A (en) * | 2020-07-01 | 2020-11-06 | 上海依图信息技术有限公司 | Semantic matching method and device |
| WO2021068615A1 (en) * | 2019-10-12 | 2021-04-15 | 深圳壹账通智能科技有限公司 | Method and device for acquiring question and answer data in document, computer device, and storage medium |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10275515B2 (en) * | 2017-02-21 | 2019-04-30 | International Business Machines Corporation | Question-answer pair generation |
| JP2020123131A (en) * | 2019-01-30 | 2020-08-13 | 株式会社東芝 | Dialog system, dialog method, program, and storage medium |
-
2021
- 2021-06-30 CN CN202110738564.5A patent/CN113553412B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110597966A (en) * | 2018-05-23 | 2019-12-20 | 北京国双科技有限公司 | Automatic question answering method and device |
| CN111428010A (en) * | 2019-01-10 | 2020-07-17 | 北京京东尚科信息技术有限公司 | Man-machine intelligent question and answer method and device |
| CN111858859A (en) * | 2019-04-01 | 2020-10-30 | 北京百度网讯科技有限公司 | Automatic question-answering processing method, device, computer equipment and storage medium |
| CN110096580A (en) * | 2019-04-24 | 2019-08-06 | 北京百度网讯科技有限公司 | A kind of FAQ dialogue method, device and electronic equipment |
| CN110309283A (en) * | 2019-06-28 | 2019-10-08 | 阿里巴巴集团控股有限公司 | A kind of answer of intelligent answer determines method and device |
| WO2021068615A1 (en) * | 2019-10-12 | 2021-04-15 | 深圳壹账通智能科技有限公司 | Method and device for acquiring question and answer data in document, computer device, and storage medium |
| CN111046159A (en) * | 2019-12-13 | 2020-04-21 | 杭州翼兔网络科技有限公司 | Man-machine interaction method, device, equipment and medium |
| CN111241237A (en) * | 2019-12-31 | 2020-06-05 | 中国建设银行股份有限公司 | Intelligent question and answer data processing method and device based on operation and maintenance service |
| CN111898643A (en) * | 2020-07-01 | 2020-11-06 | 上海依图信息技术有限公司 | Semantic matching method and device |
Non-Patent Citations (4)
| Title |
|---|
| A Chinese Question Answering System based on GPT;Shuai Liu等;IEEE Xplore;全文 * |
| 基于注意力和字嵌入的中文医疗问答匹配方法;陈志豪;余翔;刘子辰;邱大伟;顾本刚;;计算机应用(第06期);全文 * |
| 基于特征融合的社区问答问句相似度计算;杨海天;王健;林鸿飞;;江西师范大学学报(自然科学版)(第02期);全文 * |
| 基于问句相似度的本体问答系统;刘汉兴;刘财兴;林旭东;;广西师范大学学报(自然科学版)(第01期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113553412A (en) | 2021-10-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113553412B (en) | Question-answering processing method, question-answering processing device, electronic equipment and storage medium | |
| EP4113357A1 (en) | Method and apparatus for recognizing entity, electronic device and storage medium | |
| CN113435208B (en) | Training method and device for student model and electronic equipment | |
| CN111858843B (en) | Text classification method and device | |
| CN113392253B (en) | Visual question-answering model training and visual question-answering method, device, equipment and medium | |
| JP2022135991A (en) | Method for training cross-modal retrieval model, device, apparatus and storage medium | |
| CN111966782A (en) | Retrieval method and device for multi-turn conversations, storage medium and electronic equipment | |
| CN114417871B (en) | Model training and named entity recognition method, device, electronic equipment and medium | |
| CN114416943A (en) | Training method and device for dialogue model, electronic equipment and storage medium | |
| CN116152833B (en) | Training method of form restoration model based on image and form restoration method | |
| CN115048505A (en) | Corpus screening method and device, electronic equipment and computer readable medium | |
| CN115062718A (en) | Language model training method and device, electronic equipment and storage medium | |
| CN112906368A (en) | Industry text increment method, related device and computer program product | |
| CN116502649A (en) | Training method and device for text generation model, electronic equipment and storage medium | |
| CN112232089B (en) | Pre-training method, device and storage medium of semantic representation model | |
| CN116467461A (en) | Data processing method, device, equipment and medium applied to power distribution network | |
| CN115640520A (en) | Method, device and storage medium for pre-training cross-language cross-modal model | |
| EP4521264A1 (en) | Method of detecting text, training method, apparatus, device, medium, and program product | |
| CN113360602B (en) | Method, apparatus, device and storage medium for outputting information | |
| CN113641724A (en) | Knowledge tag mining method and device, electronic equipment and storage medium | |
| CN116244432B (en) | Pre-training method and device for language model and electronic equipment | |
| CN112507705A (en) | Position code generation method and device and electronic equipment | |
| CN114758649B (en) | Voice recognition method, device, equipment and medium | |
| CN116304385A (en) | Method and device for searching interest points of Chinese webpage and electronic equipment | |
| CN116244413B (en) | New intention determining method, apparatus and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |