CN105608221B - A kind of self-learning method and device towards question answering system - Google Patents
A kind of self-learning method and device towards question answering system Download PDFInfo
- Publication number
- CN105608221B CN105608221B CN201610015187.1A CN201610015187A CN105608221B CN 105608221 B CN105608221 B CN 105608221B CN 201610015187 A CN201610015187 A CN 201610015187A CN 105608221 B CN105608221 B CN 105608221B
- Authority
- CN
- China
- Prior art keywords
- knowledge
- information
- input information
- question answering
- record
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/903—Querying
- G06F16/9032—Query formulation
- G06F16/90332—Natural language query formulation or dialogue systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of self-learning method and device towards question answering system.This method includes：Receive the first input information；Determine that there are first question answering process in the preset time before receiving first input information, start self study flow；The output information in the first question answering process is extracted, the output information and described first is entered information as into potential storage knowledge store to knowledge base to be learned；The degree of correlation for calculating the output information and first input information, scores to the obtained degree of correlation, and the potential storage knowledge store that the degree of correlation is met preset requirement obtains knowledge record to knowledge base.The application can feed back the input of user using abundant, the accurate language material stored in knowledge base, and the role transforming between user is completed according to dialogue language material, to by the way of personalizing and user session, keep man-machine dialog procedure more smoothly.
Description
Technical field
The present invention relates to intelligent Service technical fields, specifically, being related to a kind of self-learning method towards question answering system
And device.
Background technology
Intelligent answer robot belongs to artificial intelligence and the crossing domain of natural language processing, can pass through natural language
Mode is exchanged and is answered a question with user, answers the problem of user is proposed with natural language, ensures that people rapidly and accurately obtain
Information simultaneously meets the demands such as people's company, amusement.
Existing question and answer robot mostly uses after receiving the problem of user proposes, selection and user from knowledge base
The problem of the answer information that matches fed back.However, with the new sentence emergence of new term in actual life, in question and answer
Hold included language material to enrich constantly, to promote the answer quality of question answering system, it is necessary to timely update to knowledge base.
However, the database update of existing question and answer robot is asked by the crawl of more new knowledge base manually by operating personnel
Answer language material.Not only labor intensive, and the quality of question and answer language material can not ensure at no point in the update process, eventually lead to not promoted and ask
Answer system answer quality.Therefore, there is an urgent need for a kind of self-learning methods and device that knowledge base can be automatically updated in question answering process.
Invention content
It is an object of the present invention to solve the technological deficiency that knowledge base cannot automatically update in existing question answering system.
The embodiment of the present invention provides a kind of self-learning method towards question answering system first, includes the following steps：
Receive the first input information；
Determine that there are first question answering process in the preset time before receiving first input information, start self study
Flow；
The output information in the first question answering process is extracted, the output information and described first are entered information as
Potential storage knowledge store is to knowledge base to be learned；
The degree of correlation for calculating the output information and first input information, scores to the obtained degree of correlation,
And the potential storage knowledge store that the degree of correlation met preset requirement obtains knowledge record to knowledge base.
In one embodiment, further include：
After obtaining knowledge record, the second input information is received；
When second input information includes the output information, first input information is exported；
Evaluation information is received, operation is increased and decreased to the evaluation score value of the knowledge record using the evaluation information.
In one embodiment, the output information and described first is being entered information as into potential storage knowledge store
Into the step of knowledge base to be learned：
In the first question answering process, extract successively by first input information and its be directed to it is first described defeated
Go out the potential storage knowledge of information composition, and stores to knowledge base to be learned.
In one embodiment, further include：
When the evaluation score value of stored knowledge record in knowledge base is less than preset value, by the knowledge record from knowledge
It is deleted in library.
The embodiment of the present invention also provides a kind of self study device towards question answering system, including：
First receiving unit is configured to receive the first input information；
Self study start unit is configured to determination and exists in the preset time before receiving first input information
First question answering process starts self study flow；
Extraction unit is configured to extract the output information in the first question answering process, by the output information and institute
It states first and enters information as potential storage knowledge store to knowledge base to be learned；
Storage unit is configured to calculate the degree of correlation of the output information and first input information, to acquired
The degree of correlation score, and the degree of correlation is met into the potential storage knowledge store of preset requirement to knowledge base, obtains knowledge note
Record.
In one embodiment, further include：
Second receiving unit, is configured to after obtaining knowledge record, receives the second input information；
Output unit is configured to when second input information includes the output information, exports the first input letter
Breath；
Evaluation unit is configured to receive evaluation information, using the evaluation information to the evaluation score value of the knowledge record
It is increased and decreased operation.
In one embodiment, the extraction unit is additionally operable to：
In the first question answering process, extract successively by first input information and its be directed to it is first described defeated
Go out the potential storage knowledge of information composition, and stores to knowledge base to be learned.
In one embodiment, further include：
Deleting unit is configured to when the evaluation score value of stored knowledge record in knowledge base is less than preset value, will
The knowledge record is deleted from knowledge base.
The embodiment of the present invention can extract the supplement of the Question Log during nan-machine interrogation and improve knowledge base, can rely on
Automatically request-answering system itself completes self study.Number of users is more, and the question and answer number of user and system is more, the accumulation of knowledge base
Speed is faster.
The embodiment of the present invention can also identify evaluation information input by user, to each knowledge being stored in knowledge base
Record is given a mark.The high knowledge record of user's evaluation can obtain higher scoring, and the unsatisfied knowledge record of user scores
It is relatively low.
During self study, automatically request-answering system extracts the language material supplement knowledge for meeting dialogue custom in some scene
Library, after the language material accumulation to a certain extent in knowledge base, system can utilize abundant, the accurate language stored in knowledge base
Material feeds back the input of user, and system completes the role transforming between user according to dialogue language material, to using anthropomorphic
The mode and user session of change, keep man-machine dialog procedure more smoothly.
Other features and advantages of the present invention will be illustrated in the following description, also, partly becomes from specification
It obtains it is clear that understand through the implementation of the invention.The purpose of the present invention and other advantages can be by specification, rights
Specifically noted structure is realized and is obtained in claim and attached drawing.
Description of the drawings
Attached drawing is used to provide further understanding of the present invention, and a part for constitution instruction, the reality with the present invention
It applies example and is used together to explain the present invention, be not construed as limiting the invention.In the accompanying drawings：
Fig. 1 is the step flow chart of the self-learning method of the embodiment of the present invention one；
Fig. 2 is the step flow chart of the self-learning method of the embodiment of the present invention two；
Fig. 3 is the structural schematic diagram of the self study device of the embodiment of the present invention three；
Fig. 4 is the structural schematic diagram of the self study device of the embodiment of the present invention four.
Specific implementation mode
To make the object, technical solutions and advantages of the present invention clearer, the present invention is made below in conjunction with attached drawing further
Ground is described in detail.
The embodiment of the present invention is illustrated below in conjunction with Figure of description, it should be understood that described herein preferred
Embodiment is only for the purpose of illustrating and explaining the present invention and is not intended to limit the present invention.And in the case where not colliding, the present invention
Embodiment in feature can be combined with each other.
The embodiment of the present invention provides a kind of self-learning method, and question and answer note is specifically extracted during human-computer dialogue
Address book stored realizes the automatic extension of knowledge base in knowledge base.In view of the continuity during human-computer dialogue, user may be
Evaluation information of the input to question answering process in question answering process.The embodiment of the present invention can recognize these evaluation informations, to knowledge
The knowledge record preserved in library is given a mark, so that the knowledge record for being stored in knowledge base meets dialogue custom, so that being
System study reaches the effect that preferably personalizes to the dialogue custom and the mode of thinking of user.Below by way of human-computer dialogue process
Two examples of priority illustrate.
First example：
User：You ate meal
Robot：I ate.
User：What meal that you eat
Robot：The Pizza that I eats.
Posterior first example：
User：It is good boring
Robot：You ate meal
User：I ate.
Robot：What meal that you eat
User：You are very clever！
Posterior second example：
User：It is good boring.
Robot：You ate meal
User：I ate.
Robot：Hundred Pizzas found pleasure in well are hunky-dory.
User：You are very stupid！
In above-mentioned posterior first example, " you are very clever " input by user is commenting to first question answering process
Valence.Smoothly due to interactive process ratio, the evaluation of user is the evaluation of positive commendation property, shows first question and answer note
Record meets user's needs.The Question Log " I ate, what meal that you eat " in first example is extracted, is known as one
Memorize address book stored in knowledge base, and according to commendation property evaluation increase this knowledge record score value.The embodiment of the present invention
Middle can achieve the effect that is to make automatically request-answering system study to " I ate, what meal that you eat " this dialogue custom.
During posterior first exemplary human-computer dialogue, if user's input information is " I ate ", it is
System feedback " what meal that you eat ".Conversational character between system and user changes, and the answer of system is with more personification
Change effect.And " you are very clever " input by user is carried out for " I ate, what meal that you eat " this knowledge record
Evaluation.
During posterior second exemplary human-computer dialogue, since there are not question and answer not in the dialogue in interactive process
Match, the answer of system feedback does not simultaneously meet user's needs.For " I ate, and hundred Pizzas found pleasure in well are hunky-dory ", user is defeated
Enter the evaluation of " you are very stupid " this negative derogatory sense property.
If extracted first Question Log " I ate, and hundred Pizzas found pleasure in well are hunky-dory " is stored in knowledge base
In, then the score value of this knowledge record is reduced according to the evaluation of derogatory sense property.Can achieve the effect that in the embodiment of the present invention is,
Systematic learning was to " I ate, and hundred Pizzas found pleasure in well are hunky-dory " and did not met the question and answer custom of user.Subsequent man-machine
In dialog procedure, if user's input information is " I ate ", system cannot answer that " hundred Pizzas found pleasure in well are quite well
", but need to calculate other output results.
In this way, constantly carry out rolling study during human-computer dialogue, extraction " problem-answer " is this asking of occurring in pairs
Result is answered as knowledge record.So that dialogue custom and the mode of thinking of the systematic learning to user, reach the effect that preferably personalizes
Fruit.
It will be discussed below the step flow of the self-learning method of the embodiment of the present invention.
Embodiment one
The present embodiment provides a kind of self-learning methods towards question answering system, are specifically carried during human-computer dialogue
Question Log is taken, and obtains real knowledge record after validation problem and the correlation of answer and is stored in knowledge base, it is real
The automatic extension of existing knowledge base makes question answering system study be accustomed to the dialogue of user.
For clarity, " the first input information " and " the second input information " that occurs in steps flow chart is said first
It is bright.Here, user's input information that " the first input information " expression formerly receives in sequential, and " the second input information " table
Show the user's input information received later in sequential.It should be noted that the first input information and the second input information can
To come from the same user or different users.
Wherein, input information can be there are many form, including but not limited to voice, text, gesture or positioning operation etc..
The content of input information may include puing question to, evaluating and talk in professional jargon.In the example of question answering process above, " the first input information "
Refer to " what meal that you eat " in first example, " the second input information " refers to " I in posterior first example
It ate ".
Illustrate the step flow of self-learning method referring to Fig. 1.
In step S101, the first input information of user is received.Preferably, in this step also to receive first
Input information is filtered processing, to ignore the invalid word in instruction.In addition, can also be carried out to the first input information received
Error correction.
In step s 102, it checks and whether there is dialog procedure before receiving the first input information.Specifically, by
Need to learn after completing interactive process in user, and the learning process of user again cannot with it is first
Question answering process interval it is too long, it is therefore desirable to check there are question and answer mistakes in the preset time before receiving the first input information
Journey.
For example, in first example, after receiving " what meal that you eat ", inspection is within half an hour
It is no to there is question answering process as " you ate meal, I ate ".
There are in the case of question answering process, execute follow-up step in preset time before receiving the first input information
Suddenly, start self study flow.
In step s 103, the output information in first question answering process is extracted, that is, extracts the output in question answering process and believes
The first input information for being directed to is ceased, the output information and described first are entered information as into potential storage knowledge store to waiting for
Learning knowledge library.
For example, being extracted in first question answering process according to the first input information " what meal that you eat ", the enquirement is targeted
Answer information " I ate ".Then " I ate " and " what meal that you eat " is a potential storage knowledge.
It is emphasized that since human-computer dialogue is time-continuing process, it in step s 103 can be to lasting human-computer dialogue
Language material in journey carries out rolling extraction, to ensure that complete, comprehensively study to user thinking and dialogue are accustomed to.Specifically,
In the first question answering process, extracts be made of first input information and its first output information being directed to successively
Potential storage knowledge, and store to knowledge base to be learned.
In step S104, the degree of correlation of the output information and first input information is calculated, to obtained correlation
Degree scores., can whether consistent with the theme between output information and the first input information in a preferable example, with
And syntax, the correlation degree in terms of grammer are calculated to score the degree of correlation.
Such as the relevance score between calculating " I ate " and " what meal that you eat ", there are following several ways.
With test problems " what meal that you eat " and it can answer whether " I ate " belongs to same session theme, because
Belong to " having a meal " this theme, therefore the relevance score between problem " what meal that you eat " and answer " I ate "
It is higher.
Can with test problems " what meal that you eat " and answer in " I ate " subject with the presence or absence of the first person and
The correspondence of the second person.In this illustration, the preceding primary subject " I " answered in result and the subject " you " in question information
In correspondence with each other, therefore the score value of the degree of correlation is higher.
Can with test problems " what meal that you eat " and answer " I ate " in clause with the presence or absence of interrogative sentence with
The correspondence of declarative sentence.In this illustration, preceding primary answer result " I ate " is declarative sentence, " you in question information
What meal eaten " is interrogative sentence, therefore there are the correspondences in clause, therefore the score value of the degree of correlation is higher.
In step S105, judge whether the score value of the degree of correlation meets preset condition, if meeting preset condition, holds
Row step S106, the potential storage knowledge store that the degree of correlation is met preset requirement obtain knowledge record to knowledge base.If discontented
Sufficient preset condition, thens follow the steps S107, abandons this potential storage knowledge.
It is in advance default relevance threshold, if the score value of the degree of correlation is higher than in a preferred example
Relevance threshold, then it is assumed that meet preset condition.This discriminant approach be in order to prevent system capture automatically it is potential storage know
The degree of correlation not high language material is extracted when knowledge, the language material for avoiding these degrees of correlation not high influences system and subsequently exports answer letter
The quality of breath.Therefore, the potential storage knowledge for being unsatisfactory for degree of correlation condition is excluded in step s 107.
By taking example one as an example, the output information and the first input that potential storage knowledge is included are judged in step S105
Information meets preset incidence relation, then in step s 106, by potential storage knowledge " I ate " and " what meal that you eat
" stored to knowledge base as knowledge record.
So far the process of the extraction knowledge record from question answering process is completed.From the above analysis as can be seen that nan-machine interrogation's mistake
Question Log in journey, which can supplement, improves knowledge base, can complete self study by automatically request-answering system itself.Use in system
Amount amount is bigger, and interactive number is more, and system more can be accumulated promptly, supplement knowledge base automatically.
Embodiment two
The present embodiment also provides another self-learning method, and after improving learning knowledge library automatically, system can be timely
Knowledge record for every new typing scores, and updates the score value of every record, can not will meet user session in this way
The knowledge record of custom reduces score value.During human-computer dialogue, evaluation information be typically follow closely previous question answering process it
Afterwards, self-learning method provided in this embodiment at any time knows this new typing after knowledge record is entered into knowledge base
Memorize record is scored, and can guarantee the accurate update of score value.
The step flow of the self-learning method of the present embodiment is illustrated below in conjunction with Fig. 2.The step identical as Fig. 1 is not
It repeats again.What is different from the first embodiment is that after obtaining knowledge record in step s 106, step S108 is executed, receives user
The second input information.For example, in posterior second example described above, " I ate " input by user is second defeated
Enter information.
Next, judging whether the second input information includes the output information in step S109.If it is determined that second
Input information includes the input information, thens follow the steps S110, exports the first input information；It is no to then follow the steps S101,
It is ready to enter self study process.
For example, in posterior second example, the second input information " I ate " of user just in knowledge record
Output information " I ate " it is identical.So, system carries out pair according to the mode of thinking of the user learnt and user
Words, i.e. the first input information of system output " what meal that you eat ".
Therefore, what the knowledge record that question answering system learns reflected is the question and answer custom of user, in order to subsequent
In dialog procedure, dialogue expection, the mode of thinking and the living habit of system analog subscriber and user engage in the dialogue.
Then, in step S111, the evaluation information of user is received, and utilize the evaluation information to institute in step S112
The evaluation score value for stating knowledge record is increased and decreased operation.Specifically, the evaluation information of user includes such as " you are very clever " " you
It is very stupid " etc evaluation.These evaluation informations are typically whether to meet the every-day language of user for last round of question answering process
Custom.
The detailed process to score the knowledge record having stored in knowledge base in step S112 is provided at this
One example.
The tendentiousness of first identification and evaluation information increases the credible of knowledge record when the tendentiousness of evaluation information is commendation
Spend score value；When the tendentiousness of evaluation information is derogatory sense, the confidence level score value of knowledge record is reduced.
Obviously, the tendentiousness of " you are very clever " is positive commendation in posterior first example, indicates first and knows
Memorize record meets user's needs, increases the first question and answer mistake of stored knowledge record " I ate, what meal that you eat "
Journey confidence level score value.In posterior second example, the tendentiousness of " you are very stupid " is the derogatory sense of negative sense, then reduces knowledge record
The score value of " I ate, and hundred Pizzas found pleasure in well are hunky-dory ".
It can be seen that in the question answering process between question answering system and user, system is constantly made to obtain for certain scene more
Horn of plenty, accurate language material, the knowledge record stored in system knowledge base not only conforms with the question and answer custom of user, also with user's
Living habit, session operational scenarios matching.To keep question answering process more smoothly and robot acquisition more preferably personalizes question and answer effect.
Meanwhile there is the evaluation mechanism for constantly updating iteration for certain question answering process so that robot can be extracted for the problem of user
More preferably answer, to improve user experience.
In addition, to ensure that the knowledge record in knowledge base undesirable is known preferably, being deleted in step S113
Memorize is recorded.Specifically, when the evaluation score value of stored knowledge record in knowledge base is less than preset value, the knowledge is remembered
Record is deleted from knowledge base.
In view of knowledge record during human-computer dialogue, constructed in the form of " problem-answer " is progressive rolling movement extraction
, method provided in this embodiment is by the deletion of supplement perfect, knowledge record score value the update and knowledge record of knowledge base
Operation forms whole cyclic process, and it is optimal that can keep record in knowledge base at any time.
Embodiment three
The present embodiment provides a kind of self study device towards question answering system, can extract question and answer during human-computer dialogue
Record storage realizes the automatic extension of knowledge base in knowledge base, and question answering system study is made to be accustomed to the dialogue of user.Fig. 3 is
The structural schematic diagram of the self study device.The self study device includes mainly the first receiving unit 301, self study start unit
302, extraction unit 303 and storage unit 304.
Wherein, the first receiving unit 301 is for receiving the first input information.Self study start unit 302 is configured to determine
There are first question answering process in the preset time before receiving first input information, start self study flow.
Extraction unit 303 is configured to extract the output information in the first question answering process, by the output information and institute
It states first and enters information as potential storage knowledge store to knowledge base to be learned.Extraction unit 303 can anti-locking system from
The not high language material of the degree of correlation is extracted when dynamic crawl potential storage knowledge, the language material for avoiding these degrees of correlation not high influences system
The quality of the follow-up output answer information of system.
Preferably, extraction unit 303 can also carry out rolling extraction during lasting human-computer dialogue.I.e. it is described
In first question answering process, extract successively be made of first input information and its first described output information being directed to it is potential
It is put in storage knowledge, and is stored to knowledge base to be learned.
Storage unit 304 is configured to calculate the degree of correlation of the output information and first input information, to acquired
The degree of correlation score, and the degree of correlation is met into the potential storage knowledge store of preset requirement to knowledge base, obtains knowledge note
Record.
Example IV
The present embodiment provides the another kinds self study device towards question answering system.Compared with embodiment three, the present embodiment carries
The self study device of confession at any time comments the knowledge record of this new typing after knowledge record is entered into knowledge base
Point, it can guarantee the accurate update of score value.
Fig. 4 is the structural schematic diagram of self study device provided in this embodiment.Identical structure is no longer with embodiment three
It repeats.Compared with embodiment three, the present embodiment further includes evaluation unit 305, the second receiving unit 306, output unit 307 and deletes
Except unit 308.
Second receiving unit 306 is configured to after obtaining knowledge record, receives the second input information.
Output unit 307 is configured to when second input information includes the output information, exports the first input letter
Breath.
Evaluation unit 305 is configured to receive evaluation information, using the evaluation information to the evaluation score value of the knowledge record
It is increased and decreased operation.Every knowledge record is first stored in knowledge base in this way, and constantly expands during automatic study and knows
Memorize is recorded and the score value that timely updates, and the answer information in so highest knowledge record of score is exactly optimal.
In addition, self study device shown in Fig. 4 further includes deleting unit 308, it is configured to when stored in knowledge base
When the evaluation score value of knowledge record is less than preset value, the knowledge record is deleted from knowledge base.Thus by the benefit of knowledge base
The delete operation for charging kind, knowledge record score value update and knowledge record forms whole circulate operation, can keep knowing
It is optimal to know record in library.
While it is disclosed that embodiment content as above but described only to facilitate understanding the present invention and adopting
Embodiment is not limited to the present invention.Any those skilled in the art to which this invention pertains are not departing from this
Under the premise of the disclosed spirit and scope of invention, any modification and change can be made in the implementing form and in details,
But the scope of patent protection of the present invention, still should be subject to the scope of the claims as defined in the appended claims.
Claims (8)
1. a kind of self-learning method towards question answering system, which is characterized in that include the following steps：
Receive the first input information；
Determine that there are first question answering process in the preset time before receiving first input information, start self study flow；
The output information in the first question answering process is extracted, the output information is entered information as with described first potential
Knowledge store is put in storage to knowledge base to be learned；
The degree of correlation for calculating the output information and first input information, scores to the obtained degree of correlation, and will
The degree of correlation meets the potential storage knowledge store of preset requirement to knowledge base, obtains knowledge record.
2. self-learning method as described in claim 1, which is characterized in that further include：
After obtaining knowledge record, the second input information is received；
When second input information includes the output information, first input information is exported；
Evaluation information is received, operation is increased and decreased to the evaluation score value of the knowledge record using the evaluation information.
3. self-learning method as described in claim 1, which is characterized in that believe by the output information and first input
Breath is as potential storage knowledge store into the step of knowledge base to be learned：
In the first question answering process, extracted successively by first input information and its first output letter being directed to
The potential storage knowledge constituted is ceased, and is stored to knowledge base to be learned.
4. self-learning method as claimed in claim 2 or claim 3, which is characterized in that further include：
When the evaluation score value of stored knowledge record in knowledge base is less than preset value, by the knowledge record from knowledge base
It deletes.
5. a kind of self study device towards question answering system, which is characterized in that including：
First receiving unit is configured to receive the first input information；
Self study start unit is configured to determine exist formerly in the preset time before receiving first input information
Question answering process starts self study flow；
Extraction unit is configured to extract the output information in the first question answering process, by the output information and described the
One enters information as potential storage knowledge store to knowledge base to be learned；
Storage unit is configured to calculate the degree of correlation of the output information and first input information, to obtained phase
Guan Du scores, and the potential storage knowledge store that the degree of correlation is met preset requirement obtains knowledge record to knowledge base.
6. self study device as claimed in claim 5, which is characterized in that further include：
Second receiving unit, is configured to after obtaining knowledge record, receives the second input information；
Output unit is configured to, when second input information includes the output information, export first input information；
Evaluation unit is configured to receive evaluation information, is carried out to the evaluation score value of the knowledge record using the evaluation information
Increase and decrease operation.
7. self study device as claimed in claim 5, which is characterized in that the extraction unit is additionally operable to：
In the first question answering process, extracted successively by first input information and its first output letter being directed to
The potential storage knowledge constituted is ceased, and is stored to knowledge base to be learned.
8. self study device as claimed in claims 6 or 7, which is characterized in that further include：
Deleting unit is configured to when the evaluation score value of stored knowledge record in knowledge base is less than preset value, will be described
Knowledge record is deleted from knowledge base.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610015187.1A CN105608221B (en) | 2016-01-11 | 2016-01-11 | A kind of self-learning method and device towards question answering system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610015187.1A CN105608221B (en) | 2016-01-11 | 2016-01-11 | A kind of self-learning method and device towards question answering system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105608221A CN105608221A (en) | 2016-05-25 |
| CN105608221B true CN105608221B (en) | 2018-08-21 |
Family
ID=55988160
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610015187.1A Active CN105608221B (en) | 2016-01-11 | 2016-01-11 | A kind of self-learning method and device towards question answering system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105608221B (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106202159A (en) * | 2016-06-23 | 2016-12-07 | 深圳追科技有限公司 | A kind of man-machine interaction method of customer service system |
| CN107040450B (en) * | 2016-07-20 | 2018-06-01 | 平安科技（深圳）有限公司 | Automatic reply method and device |
| CN106649706A (en) * | 2016-12-20 | 2017-05-10 | 北京云知声信息技术有限公司 | Natural language knowledge learning method and apparatus |
| CN106844627B (en) * | 2017-01-20 | 2020-06-19 | 竹间智能科技（上海）有限公司 | Online learning method and device based on dialog system |
| CN107066541A (en) * | 2017-03-13 | 2017-08-18 | 平安科技（深圳）有限公司 | The processing method and system of customer service question and answer data |
| CN108491519A (en) * | 2018-03-26 | 2018-09-04 | 上海智臻智能网络科技股份有限公司 | Man-machine interaction method and device, storage medium, terminal |
| CN108536811B (en) * | 2018-04-04 | 2020-07-17 | 上海智臻智能网络科技股份有限公司 | Voice interaction path determining method and device based on machine learning, storage medium and terminal |
| CN108984658A (en) * | 2018-06-28 | 2018-12-11 | 阿里巴巴集团控股有限公司 | A kind of intelligent answer data processing method and device |
| CN109815321B (en) * | 2018-12-26 | 2020-12-11 | 出门问问信息科技有限公司 | Question answering method, device, equipment and storage medium |
| CN110931017A (en) * | 2019-11-26 | 2020-03-27 | 国网冀北清洁能源汽车服务(北京)有限公司 | A charging interaction method for a charging pile and a charging interaction device for a charging pile |
| CN111309992B (en) * | 2020-02-19 | 2023-06-13 | 深圳市天博智科技有限公司 | Intelligent robot response method, system, robot and storage medium |
| CN111984777A (en) * | 2020-09-01 | 2020-11-24 | 中国平安财产保险股份有限公司 | Production system reporting problem processing method and device based on natural language processing |
| CN112148743A (en) * | 2020-09-18 | 2020-12-29 | 北京达佳互联信息技术有限公司 | Update method, device, device and storage medium for intelligent customer service knowledge base |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104699708A (en) * | 2013-12-09 | 2015-06-10 | 中国移动通信集团北京有限公司 | Self-learning method and device for customer service robot |
| CN104809197A (en) * | 2015-04-24 | 2015-07-29 | 同程网络科技股份有限公司 | On-line question and answer method based on intelligent robot |
| CN105068661A (en) * | 2015-09-07 | 2015-11-18 | 百度在线网络技术（北京）有限公司 | Man-machine interaction method and system based on artificial intelligence |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8738617B2 (en) * | 2010-09-28 | 2014-05-27 | International Business Machines Corporation | Providing answers to questions using multiple models to score candidate answers |
-
2016
- 2016-01-11 CN CN201610015187.1A patent/CN105608221B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104699708A (en) * | 2013-12-09 | 2015-06-10 | 中国移动通信集团北京有限公司 | Self-learning method and device for customer service robot |
| CN104809197A (en) * | 2015-04-24 | 2015-07-29 | 同程网络科技股份有限公司 | On-line question and answer method based on intelligent robot |
| CN105068661A (en) * | 2015-09-07 | 2015-11-18 | 百度在线网络技术（北京）有限公司 | Man-machine interaction method and system based on artificial intelligence |
Non-Patent Citations (2)
| Title |
|---|
| Predicting the quality of user-generated answers using co-training in community-based question answering portals;Bingquan Liu et al;《Pattern RecognitionLetters》;20150305;第29-34页 * |
| 交互式问答系统中的待改进问题自动识别方法;葛丽萍;《中国优秀硕士学位论文全文数据库信息科技辑》;20150215(第2期);第I139-158页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105608221A (en) | 2016-05-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105608221B (en) | A kind of self-learning method and device towards question answering system | |
| CN109710772B (en) | Question-answer base knowledge management system based on deep learning and implementation method thereof | |
| CN108446286B (en) | Method, device and server for generating natural language question answers | |
| US8818926B2 (en) | Method for personalizing chat bots | |
| CN105138710A (en) | Chat agent system and method | |
| CN108733650B (en) | Personalized word obtaining method and device | |
| CN110489756A (en) | Conversational human-computer interaction spoken language evaluation system | |
| CN106776926A (en) | Improve the method and system of responsibility when robot talks with | |
| US20220198337A1 (en) | Information processing system, information processing method and information processing device | |
| CN114270337A (en) | System and method for personalized and multi-modal context-aware human-machine dialog | |
| CN108763548A (en) | Collect method, apparatus, equipment and the computer readable storage medium of training data | |
| CN105677896B (en) | Exchange method and interactive system based on Active Learning | |
| CN113617036A (en) | Game dialogue processing method, device, equipment and storage medium | |
| CN108831229B (en) | Chinese automatic grading method | |
| US11587460B2 (en) | Method and system for adaptive language learning | |
| CN117252260B (en) | Interview skill training method, equipment and medium based on large language model | |
| CN112185187B (en) | Learning method and intelligent device for social language | |
| CN107240394A (en) | A kind of dynamic self-adapting speech analysis techniques for man-machine SET method and system | |
| CN108897517B (en) | Information processing method and electronic equipment | |
| Ismail et al. | Review of personalized language learning systems | |
| CN112992124A (en) | Feedback type language intervention method, system, electronic equipment and storage medium | |
| US10691894B2 (en) | Natural polite language generation system | |
| Yu et al. | The BURCHAK corpus: A challenge data set for interactive learning of visually grounded word meanings | |
| CN110413737B (en) | Synonym determination method, synonym determination device, server and readable storage medium | |
| CN118917437A (en) | Man-machine dialogue method based on AI intelligent large model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |