CN110020007A - Robot interactive control method and device, storage medium, server - Google Patents
Robot interactive control method and device, storage medium, server Download PDFInfo
- Publication number
- CN110020007A CN110020007A CN201710832363.5A CN201710832363A CN110020007A CN 110020007 A CN110020007 A CN 110020007A CN 201710832363 A CN201710832363 A CN 201710832363A CN 110020007 A CN110020007 A CN 110020007A
- Authority
- CN
- China
- Prior art keywords
- current context
- answer
- interactive control
- candidate
- robot interactive
- 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.)
- Pending
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
-
- 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
- G06F16/90335—Query processing
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
A kind of robot interactive control method and device, storage medium, server, which comprises determine current context, and the current context is separately input into multiple chat robots；Obtain the candidate answer that each chat robots are made in response to the current context；Each candidate is answered and the matching degree of the current context is ranked up, highest candidate answer of matching degree is selected and is used as answer.Based on technical solution of the present invention, shows that information is available with the high information of user interest matching degree jointly using multiple chat robots, improve the reading experience of user.
Description
Technical field
The present invention relates to technical field of information processing, and in particular to robot interactive control method and device, storage medium,
Server.
Background technique
Existing information is shown and is often presented one by one using list mode, such as microblogging, various newsfeed applications
Deng.Often by the matching degree of the issuing time of content and user interest between each information in information list
Equal feature organizations information list, therefore correlation semantically may lacked in same information list between each information
Property, it can not generate and interact with user, influence user experience.Furthermore although the algorithm based on user's portrait can be according to user's
Interest adjusts information, but the interest of user often has the change of some instantaneities, and is drawn a portrait based on user and recommend money
The algorithm of news information can not but capture this change in due course, and the adjustment of information is caused to lag behind user interest.
Summary of the invention
Present invention solves the technical problem that be how to realize show user information have semantic dependency, and with
User interest keeps matched.
In order to solve the above technical problems, the embodiment of the present invention provides a kind of robot interactive control method, robot interactive
Control method comprises determining that current context, and the current context is separately input into multiple chat robots；It obtains every
The candidate answer that a chat robots are made in response to the current context；By it is each it is candidate answer with it is described it is current on
Matching degree hereafter is ranked up, and is selected highest candidate answer of matching degree and is used as answer.
Optionally, the determining current context includes: that at least one content corpus conduct is chosen from corpus of content
The current context.
Optionally, the content corpus in the corpus of content is hot topic.
Optionally, the determining current context includes: that user is entered information as the current context.
Optionally, the determining current context includes: using chat record as the current context, the chat note
Record includes previous current context and its answer.
Optionally, it before the matching degree by each candidate answer and the current context is ranked up, also wraps
It includes: according to the history chat record and pre-set business rules of active user, whole candidate answer being filtered.
Optionally, described be filtered to whole candidate answers includes: to filter out to repeat in whole candidate answer
Or it is answered less than the candidate of matching degree preset threshold.
Optionally, candidate answer with the matching degree of the current context includes: that candidate answer is worked as with described
The semantic relevancy of front upper and lower text.
Optionally, the method also includes: output is described to answer to terminal.
In order to solve the above technical problems, the embodiment of the present invention also provides a kind of robot interactive control device, robot is handed over
Mutual control device includes: current context determining module, is adapted to determine that current context, and current context difference is defeated
Enter to multiple chat robots；It answers and obtains module, be suitable for obtaining each chat robots in response to described when front upper and lower
The candidate answer that text is made；Answer sorting module, suitable for carrying out each candidate answer with the matching degree of the current context
Sequence selects highest candidate answer of matching degree and is used as answer.
Optionally, the current context determining module chooses at least one content corpus as institute from corpus of content
State current context.
Optionally, the content corpus in the corpus of content is hot topic.
Optionally, user is entered information as the current context by the current context determining module.
Optionally, the current context determining module is using chat record as the current context, the chat note
Record includes previous current context and its answer.
Optionally, described device further include: answer filtering module, the answer sorting module by it is each it is candidate answer with
Before the matching degree of the current context is ranked up, the filtering module of answering is according to the history chat record of active user
And pre-set business rules, whole candidate answer is filtered.
Optionally, described be filtered to whole candidate answers includes: to filter out to repeat in whole candidate answer
Or it is answered less than the candidate of matching degree preset threshold.
Optionally, candidate answer with the matching degree of the current context includes: that candidate answer is worked as with described
The semantic relevancy of front upper and lower text.
Optionally, described device further include: output module is suitable for exporting described answer to terminal.
The embodiment of the invention also discloses a kind of storage mediums, are stored thereon with computer instruction, the computer instruction
Described the step of being based on robot interactive control method is executed when operation.
The embodiment of the invention also discloses a kind of server, including memory and processor, it is stored on the memory
The computer instruction that can be run on the processor, the processor execute the robot when running the computer instruction
The step of interaction control method.
Compared with prior art, the technical solution of the embodiment of the present invention has the advantages that
Technical solution of the present invention provides a kind of robot interactive control method, comprising: determines current context, and will be described
Current context is separately input into multiple chat robots；Each chat robots are obtained in response to the current context
The candidate answer made；Each candidate answer is ranked up with the matching degree of the current context, selects matching degree highest
Candidate answer as answering.The form that technical solution of the present invention imitates human chat shows information, can cause user
Browse the interest of information.On the one hand, information is shown using multiple chat robots, can avoid the occurrence of single chat
Robot lacks the case where information, is able to use family and obtains more informations；On the other hand, returning for user is showed
Answering is the information high with current context matching degree, thus front and back repeatedly answer in information have correlation, protect
Thinking continuity of the user when reading information has been demonstrate,proved, the reading experience of user can be promoted.
Further, determine that current context includes: that user is entered information as institute described in technical solution of the present invention
Current context is stated, it is possible thereby to which user is allowed to interact by way of chat with chat robots.When chat robots are shown
Information deviate user interest and/or when deviating from the current true hobby of user, user can be by being actively entered use
The input information at family changes session theme in time, quickly adjusts obtained information, makes information around user interest exhibition
It opens.
In addition, in the technical solution of the present invention, at least one content corpus can be chosen from corpus of content as institute
State current context.As a result, in the case where no user inputs, multiple chat robots still can provide relevant answer,
It thereby may be ensured that information can obtain lasting displaying.
Further, technical solution of the present invention it is described by it is each it is candidate answer with the matching degree of the current context into
Before row sequence, can also according to the history chat record and pre-set business rules of active user, to whole candidates answer into
Row filtering improves language continuity, advantageously reduces it is possible thereby to comprehensively consider the candidate semantic dependency answered with context
Information repeats displaying, improves user experience.
Detailed description of the invention
Fig. 1 is a kind of flow chart of robot interactive control method of the embodiment of the present invention；
Fig. 2 is a kind of application scenarios schematic diagram of robot interactive control method of the embodiment of the present invention；
Fig. 3 is the application scenarios schematic diagram of another kind robot interactive control method of the embodiment of the present invention；
Fig. 4 is a kind of structural schematic diagram of robot interactive control device of the embodiment of the present invention.
Specific embodiment
As stated in the background art, traditional information shows not generating with user and interact, and lacks semantic dependency, money
The adjustment update of news information often lags behind user interest, influences user experience.
In addition, present inventor is noticed after study if when use traditional chat robots to chat
When mode shows information, since the information amount of single chat robots is on the low side, user is difficult to obtain enough information
Information may influence user experience.It is mutual that traditional chat robots mainly use user's active enquirement, robot passively to answer
Flowing mode.If without user input, chat be difficult to continue, can not continuous presentation information, also will affect user experience.
Technical solution of the present invention provides a kind of robot interactive control method, comprising: determines current context, and will be described
Current context is separately input into multiple chat robots；Each chat robots are obtained in response to the current context
The candidate answer made；Each candidate answer is ranked up with the matching degree of the current context, selects matching degree highest
Candidate answer as answering.The form that technical solution of the present invention imitates human chat shows information, can cause user
Browse the interest of information.On the one hand, information is shown using multiple chat robots, can avoid the occurrence of single chat
Robot lacks the case where information, is able to use family and obtains more informations；On the other hand, returning for user is showed
Answering is the information high with current context matching degree, thus front and back repeatedly answer in information have correlation, protect
Thinking continuity of the user when reading information has been demonstrate,proved, the reading experience of user can be improved.
It is understandable to enable above-mentioned purpose of the invention, feature and beneficial effect to become apparent, with reference to the accompanying drawing to this
The specific embodiment of invention is described in detail.
Fig. 1 is a kind of flow chart of robot interactive control method of the embodiment of the present invention.
Robot interactive control method shown in FIG. 1 may comprise steps of:
S101: it determines current context, and the current context is separately input into multiple chat robots；
S102: the candidate answer that each chat robots are made in response to the current context is obtained；
S103: each candidate answer is ranked up with the matching degree of the current context, it is highest to select matching degree
Candidate answers as answer.
In specific implementation, in step s101, the determining current context may include: to choose from corpus of content
At least one content corpus is as the current context, wherein the content corpus in the corpus of content can be hot topic
Topic, such as the hot topic grabbed from news, microblogging.The corpus of content can be pre-stored within server.Or
Person, can be defeated by the chat window write-in in terminal by the current context that enters information as of user, such as user
Enter information, which is transmitted to server.It again alternatively, can be described to chat using chat record as the current context
Its record includes previous current context and its answer.
In step s 102, each chat robots make candidate answer respectively.Specifically, each chat robots
The information to match with the current context can be selected according to the content corpus of itself respectively to answer as candidate.It waits
It may include various informations, such as text, picture, video, audio etc. that choosing, which is answered,.
It may include: history chat record and the default industry according to active user before step S103 in specific implementation
Business rule is filtered whole candidate answer.Matching degree is repeated or is less than for example, filtering out in whole candidate answer
The candidate of preset threshold is answered.One non-limitative example is will to be already present in the information mistake in current chat record
It filters, exports the duplicate information of content to reduce.
In step s 103, each candidate answer is ranked up with the matching degree of the current context.The candidate
Answering to the matching degree of the current context may include: that the candidate answer is related with the semanteme of the current context
Degree.According to matching degree ranking results, highest candidate answer of matching degree can be selected and be used as answer.
In specific implementation, the robot interactive control method, which can also include: that output is described, to be answered to terminal, example
Such as, the answer can be sent to terminal, and be presented in the chat window of terminal.
Fig. 2 is a kind of application scenarios schematic diagram of the interaction control method of robot of the embodiment of the present invention.
With reference to Fig. 2, the input information 2011 of user determines that current context 2012, such as user pass through the chat in terminal
The input information 2011 of the input information 2011 of windows write access customer, the user can be text information, voice messaging etc..Or
Person can root if saving chat record (for example, containing previous current context and its answer in chat record)
Current context 2012 is determined according to the input information 2011 of the chat record and user.At this point, the chat record and user
Input information can be arranged by weight.Those skilled in the art can calculate power using well known machine learning algorithm
Weight, for example, by the holding time sequence arrangement from the new to the old of the input information 2011 of the chat record and user and respective
Weight successively decrease.Reason for doing so is that one side can comprehensively consider the continuity and correlation of all current contexts 2012
Property, and then guarantee that candidate answer that subsequent step generates has grammer continuity and semantic dependency with current context 2012；
On the other hand, weight, which is successively decreased, can guarantee that the quantity of current context 2012 meets preset requirement.
Current context 2012 is sent to chat robots set 202, for example, chat robots set 202 can wrap
Include chat robots 1, chat robots 2 ... ..., chat robots M, each chat robots are receiving current context
After 2012, the information to match with the current context 2012 can be selected according to the content corpus of itself respectively, and
It is answered the information as candidate, candidate answer that each chat robots provide can be one, be also possible to multinomial.
For example, chat robots 1 export candidate answer 1 and candidate answer 2, chat robots 2 export candidate answer 3 ... ..., chatting machine
Device people M output is candidate to answer N.Candidate answer of the output can be the informations such as text, picture, video, audio.
Preferably, answer filtering can be carried out to each candidate answer by step S203, specifically, can be according to working as
The history chat record and pre-set business rules of preceding user carries out answer filtering to all candidate answer, such as filters out the institute
It repeats in some candidate answers or is answered less than the candidate of matching degree preset threshold.Wherein, money can be set in pre-set business rules
The display form of information is interrogated, for example, successively using different forms such as text, picture, video and audios.One infinite example
It is that, if the answer in chat record has the information of a plurality of text and graphic form, text and picture can be filtered out
The candidate of form is answered, and retains video and the candidate of audio form is answered.
Later in step S204, answer sequence is carried out.Specifically, to filtered candidate answer, it can be according to institute
It states candidate answer and carries out matching degree sequence with the semantic relevancy of the current context 2012, it is highest therefrom to choose matching degree
Candidate answers.In specific implementation, following steps can be used: firstly, using the obtained term vector of training on mass text and
Sentence vector calculates answer content and current context 2012 in similarity semantically；Then, using conjunction, adverbial word, auxiliary word and
Some common words constitute co-occurrence probabilities matrix, and the matrix can use the sentence that front and back connects in corpus of content and count
It arrives；Secondly, calculating separately each candidate answer and the continuity degree of current context 2012 grammatically；Then, using logical
It crosses the dialog model that a large amount of dialogue corpus training obtain and calculates each candidate and answer and meet human conversation with current context 2012
The degree of habit；Finally, by above-mentioned three kinds of index values (namely similarity, continuity degree, meet human conversation habit journey
Degree) it is weighted and averaged, and the foundation as answer sequence, and then obtain and the highest candidate of 2012 matching degree of current context
It answers as answer.
In step S205, complete to answer output process.For example, the answer can be sent to terminal, and it is presented on
In the chat window of terminal.It should be noted that the answer can be used as chat record preservation.Further, preservation chat
It records the current context 2012 for being determined for chat sessions next time.Even if the not no input information of user as a result,
Each chat robots can also be iteratively repeated step S203 to step S205 constantly to show information, be formed in front and back
Container has the information stream of high correlation.
Alternatively, the answer can be used for determining current context next time, to trigger the generation answered next time.
It will be apparent to a skilled person that current context can be calculated using any one enforceable mode
2012 with the matching degree of the content corpus of chat robots, the embodiment of the present invention to this with no restriction.
Fig. 3 is the application scenarios schematic diagram of another kind robot interactive control method of the embodiment of the present invention.
With reference to Fig. 3, at least one content corpus can be chosen from corpus of content 3011, the content corpus of selection determines
Current context 3012.Wherein, the content corpus in the corpus of content 3011 can be hot topic, for example, from news,
The hot topic grabbed in microblogging.The corpus of content can be pre-stored within server.Alternatively, if saving chat
Record (for example, comprising previous current context 3012 and its answering in chat record), then can be according to the chat record
And the content corpus chosen determines current context 3012.At this point, the chat record and the content corpus of selection can be by
Weight is arranged.Those skilled in the art can calculate weight using well known machine learning algorithm, for example weight can be by
The sequence of the holding time of the chat record and the content corpus of selection from the new to the old arranges and respective weight is successively decreased.In this way
The reason of doing is, on the one hand can comprehensively consider the continuity and correlation of all current contexts 3012, and then after guarantee
Candidate answer that continuous step generates has grammer continuity and semantic dependency with current context 3012；On the other hand, weight
Successively decreasing can guarantee that the quantity of current context 3012 meets preset requirement.
Current context 3012 is sent to chat robots set 302, for example, chat robots set 302 can wrap
Chat robots 1 are included, chat robots 2 ... ..., chat robots M, each chat robots are in reception current context 3012
Afterwards, the information to match with the current context 3012 can be selected according to the content corpus of itself respectively and by the money
It interrogates information to answer as candidate, candidate answer that each chat robots provide can be one, be also possible to multinomial.For example,
Chat robots 1 export candidate answer 1 and candidate answer 2, and chat robots 2 export candidate answer 3 ... ..., chat robots M
Output is candidate to answer N.Candidate answer of the output can be the informations such as text, picture, video, audio.
In a non-limiting embodiment, answer filtering can be carried out to each candidate answer by step S303,
Specifically, answered to all candidate answer according to the history chat record of active user and pre-set business rules
Filter, such as filter out and repeat in whole candidate answer or answered less than the candidate of matching degree preset threshold.Wherein, industry is preset
The display form of information can be set in business rule, such as in turn using different forms such as text, picture, video and audios.
One infinite example is, can if the answer in chat record has the information of a plurality of text and graphic form
It is answered with filtering out text and the candidate of graphic form, retains video and the candidate of audio form is answered.
Later in step s 304, answer sequence is carried out.Specifically, to filtered candidate answer, it can be according to institute
It states candidate answer and carries out matching degree sequence with the semantic relevancy of the current context 3012, it is highest therefrom to choose matching degree
Candidate answers.In specific implementation, answer sequence can use following steps: firstly, being obtained using trained on mass text
Term vector and sentence vector calculate answer content and current context 3012 in similarity semantically；Then, conjunction, pair are utilized
Word, auxiliary word and some common words constitute co-occurrence probabilities matrix, and the matrix can use what front and back in corpus of content connected
Sentence counts to obtain；Secondly, calculating separately each candidate answer and the continuity degree of current context 3012 grammatically；Again
It is secondary, met using each candidate answer is calculated by the dialog model that largely training of dialogue corpus obtains with current context 3012
The degree of mankind's session dialogue habit；Finally, above-mentioned three kinds of index values (namely similarity, continuity degree, are met mankind's meeting
The degree of words dialogue habit) it is weighted and averaged, and the foundation as answer sequence, and then obtain and current context 3012
Answer is used as with highest candidate answer is spent.
In step S305, complete to answer output process.For example, the answer can be sent to terminal, and it is presented on
In the chat window of terminal.It should be noted that the answer can be used as chat record preservation, further, preservation is chatted
It records the current context 3012 for being determined for chat sessions next time.Even if the not no input information of user as a result,
Each chat robots can also be iteratively repeated step S303 to S305 constantly to show information, form context tool
There is the information stream of high correlation.
Alternatively, the answer can be used for determining current context next time, to trigger the generation answered next time.
It will be apparent to a skilled person that on the one hand, the arbitrary content corpus including hot topic
3011 informations provided can be used as the input information of determining current context 3012, and the embodiment of the present invention does not make this
Limitation；On the other hand, the content language of current context 3012 Yu chat robots can be calculated using any enforceable mode
The matching degree of material, the embodiment of the present invention to this with no restriction.
Fig. 4 is a kind of structural schematic diagram of robot interactive control device of the embodiment of the present invention.
Robot interactive control device 40 shown in Fig. 4 may include current context determining module 401, answer acquisition mould
Block 402 and answer sorting module 404.
In specific implementation, current context determining module 401 is adapted to determine that current context, and by the current context
It is separately input into multiple chat robots.The current context determining module can choose at least one from corpus of content
For content corpus as the current context, the content corpus in the corpus of content can be hot topic, such as from micro-
Hot topic that is rich, being grabbed in news.Alternatively, user can be entered information as institute by the current context determining module
State current context.Or the current context determining module can be using chat record as the current context, institute
Stating chat record includes previous current context and its answer.
It answers and obtains the time that module 402 is made suitable for obtaining each chat robots in response to the current context
Choosing is answered.In specific implementation, each chat robots can be selected with described according to the content corpus of itself when front upper and lower respectively
The information that text matches is answered as candidate.Candidate answer can be the information such as text, picture, video, audio letter
Breath.
Answer sorting module 404 is suitable for for each candidate answer being ranked up with the matching degree of the current context, chooses
Highest candidate answer of matching degree is selected to be used as answer.The working principle of answer sorting module 404 is it is to be understood that firstly, utilize
It is similar semantically to calculate answer content and current context for the training obtained term vector and sentence vector on mass text
Degree；Then, co-occurrence probabilities matrix is constituted using conjunction, adverbial word, auxiliary word and some common words, the matrix can use content
The sentence that front and back connects in source counts to obtain；Secondly, calculating separately each candidate answer and the company of current context grammatically
Coherence degree；Then, using by the largely obtained dialog model of dialogue corpus training calculate it is each it is candidate answer with it is current on
Hereafter meet the degree of mankind's session habit；Finally, by above-mentioned three kinds of index values, (similarity, meets mankind's meeting at continuity degree
Talk about the degree of habit) it is weighted and averaged, and the foundation as answer sequence, and then obtain and current context matching degree highest
Candidate answer as answering.
Further, robot interactive control device 40 can also include answering filtering module 403, in answer sequence mould
Before the matching degree of each candidate answer and the current context is ranked up by block, 403 basis of answer filtering module
The history chat record and pre-set business rules of active user is filtered whole candidate answer.Such as it filters out described complete
Portion it is candidate answer in repeat or answered less than the candidate of matching degree preset threshold, wherein it is described it is candidate answer with it is described current
The matching degree of context includes: the candidate semantic relevancy answered with the current context.
Further, robot interactive control device 40 can also include output module 405, be suitable for exporting described answer to end
End for example, the answer can be sent to terminal, and is presented in the chat window of terminal.
As a non-limiting embodiment, illustrate by taking the input information " what if I have failed in love " of user as an example below
Working principle based on robot interactive control device.
1. current context determining module 401 is by the input information " what if I have failed in love " of user as when front upper and lower
Text, and the current context is sent to all chat robots answered and obtained in module 402.
2. assuming to answer the chat robots in acquisition module 402 for user configuration includes entitled dot and small two cruelly
Robot, they have respectively replied several answers according to content corpus.Such as the answer of dot robot includes two information,
Wherein one is video content, entitled " what if fail in love? it plays games.", it is picture microblogging there are also one, text is
If " failure in love can not allow your weight to become double figures again, your this love can be Bai Tanliao ".Little Bao robot
Answer be one it is secondary walk caricature cruelly, caricature text is " pleasantly surprised after meal, to reform into unmarried ".So far, it answers and obtains module 402 1
The candidate of three chat robots is obtained altogether to answer.
3. candidate answer is sent to answer filtering module 403, the module is first by candidate's answer and active user
Nearest chat record is compared, and the answer of discovery little Bao robot occurred in nearest chat record, because
This is filtered out.Then according to intended service rule, such as 3 times or more identical informations cannot continuously occur, comparison is each
A candidate answer is simultaneously filtered.
4. assuming to answer by also two candidates after answering filtering module 403, answer sorting module 404 will be described at this time
Two candidate to answer " what if fail in love? play games " and " your weight can not be allowed to become double figures again if failed in love,
Your this love can be Bai Tanliao " and " what if I have failed in love " carries out matching degree sequence, selects the highest time of matching degree
Choosing is answered as answer.In specific implementation, the algorithm that can be used is referred in embodiment illustrated in fig. 4 to robot interactive control
The associated description of the module of device 40 processed, details are not described herein again.Assuming that after calculating, the selection " failure in love of answer sorting module 404
What if? it plays games." this video is as current best answers.
5. output module 405 exports the best answers to terminal, it is presented to the user.In specific implementation, in order to make to chat
The interaction in its machine human world continues down, the answer of this text can be saved as chat record true to current context
Cover half block 401.At this point, current context determining module 401 includes two current contexts, first is " how I have failed in love
Do? ", Article 2 be " what if failed in love? play games！".When next round chat robots are answered, can will work as front upper and lower
The chat record of text is arranged by scheduled weight, so that the information shown semantically has correlation in content, is had
Conducive to the thinking continuity for keeping user, improve user's reading experience.
6. robot interactive control device 40 is iteratively repeated the above process, select every time and current context matching degree highest
It is candidate answer as answering, ensure that thinking continuity of the user in reading, improve the reading experience of user.If user
Want to understand a new topic, then can input information, it is described to enter information as current context so as to change robot
The theme of information exchange, and then quickly change the information of displaying.
Further, the embodiment of the invention also discloses a kind of storage mediums, are stored thereon with computer instruction, the meter
The step of executing robot interactive control method in above-mentioned embodiment illustrated in fig. 1 when the instruction operation of calculation machine.Preferably, the storage
Medium may include computer readable storage medium.In specific implementation, the storage medium may include ROM, RAM, disk
Or CD etc..
Further, the embodiment of the invention also discloses a kind of server, including memory and processor, the memories
On be stored with the computer instruction that can be run on the processor, the processor executes when running the computer instruction
In above-mentioned embodiment illustrated in fig. 1 the step of robot interactive control method.
Although present disclosure is as above, present invention is not limited to this.Anyone skilled in the art are not departing from this
It in the spirit and scope of invention, can make various changes or modifications, therefore protection scope of the present invention should be with claim institute
Subject to the range of restriction.
Claims (20)
1. a kind of robot interactive control method characterized by comprising
It determines current context, and the current context is separately input into multiple chat robots；
Obtain the candidate answer that each chat robots are made in response to the current context；
Each candidate answer is ranked up with the matching degree of the current context, highest candidate answer of matching degree is selected and makees
To answer.
2. robot interactive control method according to claim 1, which is characterized in that the determining current context packet
It includes: choosing at least one content corpus from corpus of content as the current context.
3. robot interactive control method according to claim 2, which is characterized in that the content in the corpus of content
Corpus is hot topic.
4. robot interactive control method according to claim 1, which is characterized in that the determining current context packet
It includes: user is entered information as into the current context.
5. robot interactive control method according to claim 1, which is characterized in that the determining current context packet
Include: using chat record as the current context, the chat record includes previous current context and its answer.
6. robot interactive control method according to claim 1, which is characterized in that it is described by it is each it is candidate answer with
Before the matching degree of the current context is ranked up, further includes: according to the history chat record of active user and default industry
Business rule is filtered whole candidate answer.
7. robot interactive control method according to claim 6, which is characterized in that it is described to whole candidates answer into
Row filtering includes: to filter out to repeat in whole candidate answer or answer less than the candidate of matching degree preset threshold.
8. robot interactive control method according to claim 1, which is characterized in that it is described it is candidate answer with it is described current
The matching degree of context includes: the candidate semantic relevancy answered with the current context.
9. robot interactive control method according to any one of claims 1 to 8, which is characterized in that further include: output institute
Answer is stated to terminal.
10. a kind of robot interactive control device characterized by comprising
Current context determining module is adapted to determine that current context, and the current context is separately input into multiple chat
Its robot；
It answers and obtains module, returned suitable for obtaining each chat robots in response to the candidate that the current context is made
It answers；
Answer sorting module selects matching suitable for each candidate answer to be ranked up with the matching degree of the current context
It spends highest candidate answer and is used as answer.
11. robot interactive control device according to claim 10, which is characterized in that the current context determines mould
Block chooses at least one content corpus as the current context from corpus of content.
12. robot interactive control device according to claim 11, which is characterized in that interior in the corpus of content
Appearance corpus is hot topic.
13. robot interactive control device according to claim 10, which is characterized in that the current context determines mould
Block enters information as the current context for user's.
14. robot interactive control device according to claim 10, which is characterized in that the current context determines mould
For block using chat record as the current context, the chat record includes previous current context and its answer.
15. robot interactive control device according to claim 10, which is characterized in that further include: filtering module is answered,
Before the matching degree of each candidate answer and the current context is ranked up by the answer sorting module, the answer
Filtering module is filtered whole candidate answer according to the history chat record and pre-set business rules of active user.
16. robot interactive control device according to claim 15, which is characterized in that the answer filtering module is suitable for
It filters out and repeats in whole candidate answer or answered less than the candidate of matching degree preset threshold.
17. robot interactive control device according to claim 10, which is characterized in that candidate answer is worked as with described
The matching degree of front upper and lower text includes: the candidate semantic relevancy answered with the current context.
18. robot interactive control device according to any one of claims 10 to 17, which is characterized in that further include: it is defeated
Module out is suitable for exporting described answer to terminal.
19. a kind of storage medium, is stored thereon with computer instruction, which is characterized in that the computer instruction executes when running
The step of any one of claim 1 to 9 robot interactive control method.
20. a kind of server, including memory and processor, be stored on the memory to run on the processor
Computer instruction, which is characterized in that perform claim requires any one of 1 to 9 when the processor runs the computer instruction
The step of robot interactive control method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710832363.5A CN110020007A (en) | 2017-09-15 | 2017-09-15 | Robot interactive control method and device, storage medium, server |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710832363.5A CN110020007A (en) | 2017-09-15 | 2017-09-15 | Robot interactive control method and device, storage medium, server |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110020007A true CN110020007A (en) | 2019-07-16 |
Family
ID=67186311
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710832363.5A Pending CN110020007A (en) | 2017-09-15 | 2017-09-15 | Robot interactive control method and device, storage medium, server |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110020007A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110543552A (en) * | 2019-09-06 | 2019-12-06 | 网易（杭州）网络有限公司 | Conversation interaction method and device and electronic equipment |
| CN113127608A (en) * | 2019-12-31 | 2021-07-16 | 微软技术许可有限责任公司 | Plan-guided response provision |
| CN114144773A (en) * | 2019-08-01 | 2022-03-04 | 国际商业机器公司 | Adjusting conversational flow based on behavior in human-machine cognitive interactions |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080040114A1 (en) * | 2006-08-11 | 2008-02-14 | Microsoft Corporation | Reranking QA answers using language modeling |
| CN103229120A (en) * | 2010-09-28 | 2013-07-31 | 国际商业机器公司 | Providing answers to questions using hypothesis pruning |
| CN105183848A (en) * | 2015-09-07 | 2015-12-23 | 百度在线网络技术（北京）有限公司 | Human-computer chatting method and device based on artificial intelligence |
| CN107135143A (en) * | 2017-03-27 | 2017-09-05 | 厦门快商通科技股份有限公司 | Many chat robots switching systems and its dialogue method |
-
2017
- 2017-09-15 CN CN201710832363.5A patent/CN110020007A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080040114A1 (en) * | 2006-08-11 | 2008-02-14 | Microsoft Corporation | Reranking QA answers using language modeling |
| CN103229120A (en) * | 2010-09-28 | 2013-07-31 | 国际商业机器公司 | Providing answers to questions using hypothesis pruning |
| CN105183848A (en) * | 2015-09-07 | 2015-12-23 | 百度在线网络技术（北京）有限公司 | Human-computer chatting method and device based on artificial intelligence |
| CN107135143A (en) * | 2017-03-27 | 2017-09-05 | 厦门快商通科技股份有限公司 | Many chat robots switching systems and its dialogue method |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114144773A (en) * | 2019-08-01 | 2022-03-04 | 国际商业机器公司 | Adjusting conversational flow based on behavior in human-machine cognitive interactions |
| CN114144773B (en) * | 2019-08-01 | 2022-10-28 | 国际商业机器公司 | Adjusting conversational flow based on behavior in human-machine cognitive interactions |
| CN110543552A (en) * | 2019-09-06 | 2019-12-06 | 网易（杭州）网络有限公司 | Conversation interaction method and device and electronic equipment |
| CN110543552B (en) * | 2019-09-06 | 2022-06-07 | 网易（杭州）网络有限公司 | Conversation interaction method and device and electronic equipment |
| CN113127608A (en) * | 2019-12-31 | 2021-07-16 | 微软技术许可有限责任公司 | Plan-guided response provision |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hancock et al. | Learning from dialogue after deployment: Feed yourself, chatbot! | |
| CN113779378B (en) | Method, system, and medium for forming chat robot output based on user status | |
| Gunkel | Communication and artificial intelligence: Opportunities and challenges for the 21st century | |
| CN107342078B (en) | Conversation strategy optimized cold start system and method | |
| CN108763495B (en) | Interactive method, system, electronic equipment and storage medium | |
| CN110349572A (en) | A kind of voice keyword recognition method, device, terminal and server | |
| CN109829039A (en) | Intelligent chat method, device, computer equipment and storage medium | |
| US20210201167A1 (en) | Method of knowledge sharing among dialogue systems, dialogue method and device | |
| CN109791549A (en) | Machine customer interaction towards dialogue | |
| CN105930367B (en) | Intelligent chat robot control method and control device | |
| US20160019280A1 (en) | Identifying question answerers in a question asking system | |
| US11615129B2 (en) | Electronic message text classification framework selection | |
| US20220210098A1 (en) | Providing responses in an event-related session | |
| CN110020007A (en) | Robot interactive control method and device, storage medium, server | |
| CN109948151A (en) | The method for constructing voice assistant | |
| CN108600083A (en) | Message prompt method and equipment | |
| Duplessis et al. | Purely corpus-based automatic conversation authoring | |
| CN107194151A (en) | Determine the method and artificial intelligence equipment of emotion threshold value | |
| CN114896498A (en) | Course recommendation method, system, terminal and storage medium | |
| CN109492085A (en) | Method, apparatus, terminal and storage medium are determined based on the answer of data processing | |
| US11055493B2 (en) | Conversation system | |
| CN117556802A (en) | User portrait method, device, equipment and medium based on large language model | |
| CN109190116A (en) | Semantic analytic method, system, electronic equipment and storage medium | |
| CN106096653B (en) | Ascribed characteristics of population estimating method based on cross-platform user social contact multimedia behavior | |
| CN112307166A (en) | Intelligent question and answer method and device, storage medium and computer equipment |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190716 |
|
| WD01 | Invention patent application deemed withdrawn after publication |