CN105512228B - A kind of two-way question and answer data processing method and system based on intelligent robot - Google Patents
A kind of two-way question and answer data processing method and system based on intelligent robot Download PDFInfo
- Publication number
- CN105512228B CN105512228B CN201510857525.1A CN201510857525A CN105512228B CN 105512228 B CN105512228 B CN 105512228B CN 201510857525 A CN201510857525 A CN 201510857525A CN 105512228 B CN105512228 B CN 105512228B
- Authority
- CN
- China
- Prior art keywords
- question
- user
- information
- module
- customer parameter
- 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
Abstract
The invention discloses a kind of two-way question and answer data processing method and system based on intelligent robot.Method includes the following steps: obtaining user's question information and carrying out intention assessment;It is intended to generate the first response result for corresponding to the question information according to user;When being associated with when the customer parameter composition stored in the question information and user information database, into process of actively asking a question;Problem of actively asking a question corresponding with the question information is generated according to preset session rules;Export first response result and the problem of actively asking a question.The application can not only answer the problem of user proposes, additionally it is possible to which the problem of design is consistent with user characteristics simultaneously puts question to user.User can also feed back in two-way question answering system the answer for problem of actively asking a question, to user information database carry out supplement and it is perfect, and also to system carry out intelligent training.
Description
Technical field
The present invention relates to intelligent robot technology fields, ask specifically, being related to a kind of two-way based on intelligent robot
Answer data processing method and two-way question answering system.
Background technique
Intelligent answer robot belongs to the crossing domain of artificial intelligence and natural language processing, can pass through natural language
Mode is exchanged and is answered a question with people, is answered the problem of user is proposed with natural language with accurate, succinct natural language, is guaranteed
People rapidly and accurately obtain information.
From the point of view of current question and answer robot carrys out current situation, current technology has been able to processing common-sense (such as pearl Mu Lang
The height above sea level at Ma peak is how many), open (such as feel blue and how to mediate), chat greet (such as you are preferably also OK), mandatory
The problems such as (such as putting song), brings great convenience to user and experiences with better services.
However, current question and answer robot, which is limited to user's one direction, proposes problem.That is, robot before this by
The dynamic enquirement for receiving user, then the answer of problem is provided for user.This unidirectional interrogation reply system causes the limitation in interactive process
Property, once user stops puing question to, robot will not actively initiate topic.Therefore cause the effect of man-machine communication bad, user
Experience Degree is poor.
Therefore, needing one kind can make question and answer robot actively initiate to put question to and receive what user's active initiation was putd question to
Two-way answering method and system.
Summary of the invention
The technological deficiency that cannot actively initiate to put question to it is an object of the present invention to solving existing question and answer robot.
The embodiment of the present invention provides a kind of two-way question and answer data processing method based on intelligent robot first, including with
Lower step:
It obtains user's question information and carries out intention assessment;
It is intended to generate the first response result for corresponding to the question information according to user;
When being associated with when the customer parameter composition stored in the question information and user information database, into active question stream
Journey;
Problem of actively asking a question corresponding with the question information is generated according to preset session rules;
Export first response result and the problem of actively asking a question.
In one embodiment, further includes:
Receive the second response result for the problem feedback of actively asking a question;
The related information of customer parameter is extracted in the second response result, and user information database is updated according to related information
Customer parameter.
In one embodiment, further includes:
Process of actively asking a question is trained according to second response result, updates session rules so that the active generated
Question problem is matched with the customer parameter after updating.
In one embodiment, further includes:
The scene conjunctive word in the question information is extracted, is known according to the context of the scene conjunctive word and current session
Other dialogue scenarios;
Include: in the progress intention assessment the step of
Semantic parsing is carried out to the question information according to domain model, intention assessment is carried out in the dialogue scenarios.
In one embodiment, problem of actively asking a question corresponding with the question information is generated according to preset session rules
Include:
Extract at least one problem to be selected corresponding with question information;
It selects to ask with the problem of customer parameter matching degree highest as active from least one described problem to be selected
Topic.
In one embodiment, further includes:
When it fails to match at least one described problem to be selected and customer parameter, the master adapted to the customer parameter is generated
Dynamic question problem.
In one embodiment, further includes:
Receive user's chat message;
Chat scenario is designed according to the customer parameter updated in user information database;
The dialogue of talking in professional jargon for meeting customer parameter is generated under the chat scenario.
The embodiment of the present invention also provides a kind of two-way question answering system based on intelligent robot, comprising:
Intention assessment module is configured to obtain user's question information and carries out intention assessment;
Responder module is configured to be intended to generate the first response result for corresponding to the question information according to user;
Matching module is configured to be associated with when the customer parameter composition stored in the question information and user information database
When, into process of actively asking a question;
It actively asks a question module, is configured to generate active corresponding with the question information according to preset session rules and send out
It asks questions;
Output module is configured as output to first response result and the problem of actively asking a question.
In one embodiment, further includes:
Receiving module is configured to receive the second response result for the problem feedback of actively asking a question;
Update module is configured to extract the related information of customer parameter in the second response result, and is believed according to association
Breath updates the customer parameter of user information database.
In one embodiment, further includes:
Training module is configured to be trained process of actively asking a question according to the second response result, updates session rules
So that the problem of actively asking a question generated is matched with the customer parameter after updating.
In one embodiment, further includes:
Scene identification module is configured to extract the scene conjunctive word in the question information, is associated with according to the scene
The context identification dialogue scenarios of word and current session;
The intention assessment module is also used to carry out semantic parsing to the question information according to domain model, described right
Intention assessment is carried out in words scene.
In one embodiment, the module of actively asking a question includes:
Extracting sub-module is used to extract at least one problem to be selected corresponding with question information;
Submodule is selected, is used to selecting from least one described problem to be selected and customer parameter matching degree is highest asks
Topic is as problem of actively asking a question.
In one embodiment, the module of actively asking a question further include:
Submodule is generated, is used for when it fails to match at least one described problem to be selected and customer parameter, is generated and institute
State the problem of actively asking a question of customer parameter adaptation.
In one embodiment, further includes:
Module is obtained, is configured to obtain the chat message of user;
Scenario Design module is configured to design chat scenario according to the customer parameter updated in user information database;
Chat module is configured to generate the dialogue of talking in professional jargon for meeting customer parameter under the chat scenario.
In an embodiment of the present invention, intelligent robot can not only answer user propose the problem of, additionally it is possible to design with
The problem of user characteristics are consistent simultaneously puts question to user, makes user to the interested of human-computer interaction process.User asks a question to active
The answer of problem can also be fed back in two-way question answering system, on the one hand carry out supplement and perfect, another aspect to user information database
The problem of intelligent training also is carried out to system, designs system in subsequent question answering process closer to user characteristics.
In addition, system is not only only capable of completing the question answering process with user, moreover it is possible to initiate symbol in the chat process for class of talking in professional jargon
The chat topic for closing user characteristics, is able to ascend the usage experience of user, has stronger practicability.Actively question is chatted with initiation
Its two process complements each other promotion.Actively the interaction point of question is more, can more form more detailed reaction user characteristics
Customer parameter, so that initiation user is interested in chat process or meets the topic of user's application demand, with user's shape
At more good interaction.
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 objectives and other advantages of the invention can be by specification, right
Specifically noted structure is achieved and obtained in claim and attached drawing.
Detailed description of the invention
Attached drawing is used to provide further understanding of the present invention, and constitutes part of specification, with reality of the 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 two-way question and answer data processing method of the embodiment of the present invention one;
Fig. 2 is the flow chart of the process of actively asking a question of the embodiment of the present invention one;
Fig. 3 is the step flow chart of the two-way question and answer data processing method of the embodiment of the present invention two;
Fig. 4 is the step flow chart of the two-way chat data processing method of the embodiment of the present invention three;
Fig. 5 is the structural schematic diagram of the two-way question answering system of the embodiment of the present invention four;
Fig. 6 is the structural schematic diagram of the module of actively asking a question of the embodiment of the present invention four;
Fig. 7 is the structural schematic diagram of the two-way question answering system of the embodiment of the present invention five;
Fig. 8 is the structural schematic diagram of the two-way chat system of the embodiment of the present invention six.
Specific embodiment
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 intelligent answer robot of personalization, especially a kind of user characteristics that meet
Personalized two-way question answering system, can be widely applied to the intelligence such as intelligent sound assistant, chat robots, automatic customer service and expert system
Energy service system can be putd question to actively to user, or actively initiate some topic to user.Wherein, it is proposed to user
Problem or the topic of initiation are related to personality preference, age, gender and the occupation of user etc., these problems and user characteristics
More match.To analyse in depth the behavioural habits of user by the two-way interactive of user and question and answer robot, and will be to user's
Analysis result reacts in active question mechanism, forms benign cycle to improve the degree of intelligence of system.
Before carrying out active enquirement, needs to construct domain model and/or semanteme for carrying out semantic parsing and return
Class model, the user to identify that the input information representation of user comes out are intended to.In a preferred example, it is intended that identification engine
The possible intention of user is judged by domain model.Above-mentioned neck is obtained by being trained to the respective data of every field
Domain model is substantially a kind of classification model, the order that user inputs can be referred to difference using semantic subsumption algorithm
Field.If user inputs " Kung Pao chicken ", determine that user's intention may be with restaurant, menu and encyclopaedia by semantic subsumption algorithm
There is relationship.The semantic subsumption algorithm is one by using regular expression, syntactic analysis, syntactic analysis or keyword
For the modes such as parsing come the Semantic Similarity Measurement mode realized, data basis is then a large number of users of trained domain model, neck
Numeric field data.
Before carrying out active enquirement, it is also necessary to construct user information database store with the gender of user, the age, occupation,
The relevant customer parameter such as hobby and trip mode.The user information database be in order to it is subsequent actively ask a question during identify user
Interaction topic.For example, user information includes age, gender, birthday, hobby, constellation and local etc., it further include occupation, traffic
Trip mode, time of getting up, taste of diet and frequent activity venue etc..
For example, the information of party A-subscriber is " 8 years old, male, October 1 birthday, local Beijing ".If intelligent robot receives A use
The question information of family input is " where you are ", then system judges that interaction topic can be related to customer parameter " local ",
It is exactly using " local Beijing " this information of party A-subscriber as reference information.What system was designed have the problem of actively question, and " I am
Pekinese, you are also Pekinese ", or " I is Shanghai people, you are Pekinese "
System can also actively initiate topic within user's sluggish period for a long time.For the letter of above-mentioned party A-subscriber
Breath, the topic that system is actively initiated are, for example, " place which Beijing has joyful " or " Beijing Safari Park is joyful " etc.
Deng.
Two-way question and answer data processing method and system are described in detail below in conjunction with specific embodiment.
Embodiment one
Fig. 1 is the step flow chart of two-way question and answer implementation method.
In step s101, domain model and/or semantic classification model as described above are constructed.
In step s 102, building is stored with the user information database of user information.The user information database is stored with initial
Customer parameter, and user information database can be improved and be updated in subsequent question answering process.
In step s 103, it obtains user's question information and carries out intention assessment.
Specifically, the command information that first acquisition user inputs, then command information is converted into normative text format letter
Breath, and text formatting information is pre-processed, to obtain question information to be identified.Wherein, the instruction letter of user's input
Breath can be at least one of phonetic order, text instruction or positioning instruction.
Pretreatment operation may include denoising, intelligent correction, word segmentation processing and name Entity recognition etc..Wherein, it goes
Processing of making an uproar mainly filters out the meaningless word such as invalid word, stop-word, does not influence the intention of user's input after filtering;At error correction
Reason be according to the modes such as phonetic error correction, statistics error correction, semantic error correction by the input of user's erroneous input or speech recognition errors into
Row correction process obtains relatively accurate input;Word segmentation processing and name Entity recognition are by the side such as Hidden Markov Model
Formula segments user's input, and is labeled to each part of speech, is also marked accordingly simultaneously for name entity.
Such as user inputs " going how Xizhimen is walked ", then by the way that taxis verb " going " and noun can be obtained after participle
" Xizhimen ", while the name entity for obtaining " Xizhimen " is place name, " how walking " indicates that user's is intended that inquiry route.
In this step, input information can be carried out based on the domain model constructed in advance and/or semantic classification model
Semanteme parses to identify that user is intended to.If the input information of user is specific asked questions, such as " today, how is weather "
" going how Xizhimen is walked " etc. then can directly recognize the intention of user from problem.
Further, it is also possible to the scene conjunctive word in user's question information first be extracted, according to the scene conjunctive word and currently
The context identification dialogue scenarios of dialogue.Semantic parsing is carried out to the question information further according to domain model, in the dialogue
Intention assessment is carried out in scene.Wherein, dialogue scenarios indicate User Status and user demand etc., for accurately judging that user is intended to
It is accurately replied with being provided during subsequent answer.In this course, it is essentially relying in the context of current session
The descriptor of extraction determines, the judging of scene can provide effective support as semanteme parsing, greatly improves semantic parsing
Accuracy rate.
For example, the information of party B-subscriber is " male, local Shanghai, bank clerk drive to go on a journey ".And party B-subscriber and robot system
Dialogue in once there are information such as " the too congestion of the road of today, again be late " " road vehicle are seldom ".If the enquirement of party B-subscriber
Information is " going how Xizhimen is walked ", it is determined that " Xizhimen " is scene conjunctive word, and system identification to current dialogue scenarios is B
User will drive to go to Xizhimen, and party B-subscriber is recognized in the dialogue scenarios is intended to " acquisition current location to Xizhimen
Jam situation ".
In step S104, it is intended to generate the first response result for corresponding to question information according to user.For example, for using
The question information " going how Xizhimen is walked " at family, if the inquiry that is intended that system recognizes user in step s 103 is driven road
Line, then the first response result provided are " the shortest drive routes in path from current location to Xizhimen ";Or in step
User is recognized in S103 is intended to " jam situation of acquisition current location to Xizhimen ", then the first response result provided
It is " jam situation of acquisition current location to Xizhimen, and reasonable bypass route is provided ".
In step s105, the customer parameter stored in question information and user information database is matched, in the two structure
When at association, following step S106 is executed, into process of actively asking a question.If the two cannot form association, it is back to step
S103。
For example, the information of above-mentioned party A-subscriber is " 8 years old, male, October 1 birthday, local Beijing ", then customer parameter is " year
Age ", " gender ", " birthday " or " local ".If the question information that system receives party A-subscriber's input is that " nearby have what meal
The Room ", then recognize user is intended to " dining room of finding nearby ", and user is intended to not constitute with customer parameter to be associated at this time, directly
Connect the first response result that output is intended to user.For example, the first response result of output is " to have Malus spectabilis for 500 meters eastwards
Dining room ".
If the question information that system receives party A-subscriber's input is " you this year how old ", user is intended to and customer parameter " year
Age " constitutes association, triggers mechanism of actively asking a question.System is to providing the first response result in the next steps, and generates question
Problem " my 5 years old this year, you have 8 years old ".
In step s 106, problem of actively asking a question corresponding with question information is generated according to preset session rules.As
Described in last example, if the question information that system receives party A-subscriber's input is " you this year how old ", asking for actively question is generated
Topic is " you have 8 years old ".In step s 107, first response result and the problem of actively asking a question are exported.
Fig. 2 is the preferred example of process of actively asking a question.System can calculate more according to the question information of user
A possible the problem of actively puing question to.Problem base also can be set in system, for storing all problems that may be putd question to.It is preset
Session rules can maximum matching rule between active asked questions and customer parameter.
In Fig. 2, at least one problem to be selected corresponding with question information is extracted in step s 201 first, then,
The matching degree of at least one problem to be selected and customer parameter is calculated in step S202.Next, in step S203, selection
With the problem of customer parameter matching degree highest as actively asking a question problem;Also, when the matching degree being calculated is too low, then sentence
Breaking, it fails to match with customer parameter at least one described problem to be selected, and the master adapted to customer parameter is generated in step S204
Dynamic question problem, so that the problem of actively asking a question generated more meets user characteristics.
For example, the little girl that C user is 6 years old, if the C user that system receives puts question to as " we play games together
", and be not matched in user information database the hobby of C user, then design the problem of more meeting little girl's feature, system master
Dynamic question is " good, to come together how to play Barbie doll ".
For example, D user is 50 years old, if it is " having the game what is joyful " that the D user that system receives, which puts question to, and believe in user
The hobby of D user is not matched in breath library, then designs the problem of actively puing question to is " how carrying out fighting landlord ".
In addition, if the problem to be selected for corresponding to highest matching degree in the result for calculating acquisition in step S202 is not unique
, then it is randomly choosed in step S203, i.e., randomly chooses one from multiple problems to be selected of corresponding highest matching degree
Problem is as problem of actively asking a question.
For example, the information of above-mentioned party B-subscriber is that " male, local Shanghai, bank clerk drive to go on a journey, hobby music and outdoor fortune
It is dynamic ".If the enquirement that system receives party B-subscriber is " what activity weekend has ", highest matching degree is calculated in step S202
Problem to be selected there are two, be " weekend goes to hear a concert " or " weekend go drift about " respectively, then in step S203
One is randomly choosed to export as the problem of actively asking a question.This design allows in current user information database not
It can determine that most suitable problem of actively asking a question, need to be putd question to again to user to get more, more detailed user's letter
Breath.Therefore, a problem output is randomly choosed.
From the above analysis as can be seen that according to two-way question and answer data processing method provided in this embodiment, intelligent robot
The problem of user proposes can not only be answered, additionally it is possible to which the problem of design is consistent with user characteristics simultaneously puts question to user, makes user
To the interested of human-computer interaction process.
Embodiment two
Fig. 3 is the step flow chart of two-way question and answer data processing method provided in this embodiment.It is different from embodiment one
It is that user can also feed back in two-way question answering system the answer for problem of actively asking a question, on the one hand user information database is mended
Fill and perfect, on the other hand also to system carry out intelligent training, make system designed in subsequent question answering process closer to
The problem of family feature.
In Fig. 3, the step of being the same as example 1 identical symbol logo, herein mainly to the feedback of the present embodiment
The training process of process and process of actively asking a question is described in detail, and other content repeats no more.
In step S108, user is received for the second response result of the problem feedback of actively asking a question.For above-mentioned
Party B-subscriber, if party B-subscriber to intelligent robot put question to " you how old ", robot answer and actively put question to " I have five years old, you
".Party B-subscriber is " I have 30 years old " to the second response result that " you " this active asked questions are fed back.
In step S109, the related information of customer parameter is extracted in the second response result, and more according to related information
The customer parameter of new user information database.It is closed that is, intelligent robot extracts customer parameter " age " in the second response result
The information of connection is " 30 years old ".Then the user information database of party B-subscriber is updated, addition " age is 30 years old " this information.
In step s 110, process of actively asking a question is trained according to the second response result, so that the active hair generated
It asks questions and is matched with the customer parameter after updating.Step S105 and step S106 are returned to later, generate the problem of actively asking a question.
Specifically, the process being trained to process of actively asking a question includes the update of matching process and the update for talking with criterion.It embodies
After being added to new customer parameter in user information database or being revised as new customer parameter, need according to after update
Customer parameter matched, it is determined whether there is association;Correspondingly, in the next steps according to the selection of dialogue criterion and update
The problem of customer parameter matching degree highest later.Since in this way, by carrying out intelligent training to system, make system subsequent
The problem of being designed in question answering process closer to user characteristics.
For example, robot updates session rules if party B-subscriber puts question to " having good-looking film recently " in dialogue next time
To calculate matching degree, the first response of generation according to the parameter " 30 years old age " of question information and party B-subscriber and " gender is male "
It as a result is " " trump secret service " is hunky-dory " that generation is " you like seeing action movie " the problem of actively question.
Embodiment three
Fig. 4 is the step flow chart of two-way chat data processing method provided in this embodiment.With both examples above
It compares, the method for the present embodiment is not only able to achieve nan-machine interrogation, moreover it is possible to initiate to meet user spy in the chat process for class of talking in professional jargon
The chat topic of sign is able to ascend the usage experience of user, has stronger practicability.
Fig. 4 is the improvement made on the basis of Fig. 3, and the step identical as Fig. 3 repeats no more.
After step s 102, the chat message that step S111 receives user is executed.Then, in step S112 according to
The customer parameter design chat scenario updated in the information bank of family, and generated under chat scenario in step S113 and meet user's ginseng
Several dialogues of talking in professional jargon.The answer that step S108 receives user is executed again.
For example, party B-subscriber initiate first chat " playing game everyday, It rs boring really ", robot according in party B-subscriber's information
The parameter " 30 years old age " of update, the scene for designing chat is " user is on furlough ", and the conversation message for generating class of talking in professional jargon is
" 30 years old also play game everyday, not promising ".
It should be noted that actively question complements each other promotion with the two processes of initiating to chat.The actively interaction of question
Point is more, can more form the customer parameter of more detailed reaction user characteristics, so that it is interested to initiate user in chat process
Or topic that meet user's application demand, formed with user and more well interacted.
Example IV
Fig. 5 is the structural schematic diagram of the two-way question answering system of the present embodiment.The system mainly includes intention assessment module
503, responder module 504, matching module 505, module of actively asking a question 506 and output module 507.In addition, being additionally provided with field mould
Type library 501 and user information database 502.
Field of storage module and/or semantic classification model in domain model library 501, for completing intention assessment.User's letter
Breath stores user information in library 502, is stored with initial customer parameter at first, and can in subsequent question answering process user information
Library 502 is constantly improved and is updated.
Intention assessment module 503 is configured to obtain user's question information and carries out intention assessment.Specifically, first acquisition is used
The command information of family input, then command information is converted into normative text format information, and carry out to text formatting information pre-
Processing, to obtain question information to be identified.Input is believed based on the domain model constructed in advance and/or semantic classification model
Breath carries out semantic parsing to identify that user is intended to.
Responder module 504 is configured to be intended to generate the first response result for corresponding to question information according to user.Match mould
When block 505 is configured to be associated with when the customer parameter composition stored in question information and user information database, into process of actively asking a question.
It is associated with if the customer parameter stored in question information and user information database cannot be constituted, triggers module 506 of actively asking a question, designed
The problem of actively asking a question.
Module of actively asking a question 506 is configured to generate active hair corresponding with the question information according to preset session rules
It asks questions.Output module 507 is configured as output to first response result and the problem of actively asking a question.
Fig. 6 be actively ask a question module 506 specific structure a preferable example.Module of actively asking a question 506 can basis
The question information of user calculates multiple possible the problem of actively puing question to.Problem base also can be set in system, can for storing
The all problems that can be putd question to.Preset session rules can between active asked questions and customer parameter maximum match rule
Then.
As shown in fig. 6, module 506 of actively asking a question includes extracting sub-module 601, selection submodule 602 and generation submodule
603.Extracting sub-module 601 is for extracting at least one problem to be selected corresponding with question information.Select submodule 602 based on
The matching degree of at least one problem to be selected and customer parameter is calculated, therefrom selection is made with the problem of customer parameter matching degree highest
For problem of actively asking a question.When the matching degree being calculated is too low, then at least one described problem to be selected is matched with customer parameter
Failure.It generates submodule 603 and generates the problem of actively asking a question adapted to the customer parameter, so that the problem of actively asking a question generated
More meet user characteristics.
Further, it is not if corresponding to the problem to be selected of highest matching degree in the result that selection submodule 602 obtains
Uniquely, then it is randomly choosed, i.e., randomly chooses a problem conduct from multiple problems to be selected of corresponding highest matching degree
It actively asks a question problem.
Two-way question answering system provided in this embodiment can not only answer the problem of user proposes, additionally it is possible to design and user
The problem of feature is consistent simultaneously puts question to user, makes user to the interested of human-computer interaction process.
Embodiment five
Fig. 7 is the structural schematic diagram of the two-way question answering system of the present embodiment.Unlike example IV, user is to actively
The answer of question problem can also be fed back in two-way question answering system, on the one hand to user information database carry out supplement and it is perfect, it is another
The problem of aspect also carries out intelligent training to system, designs system in subsequent question answering process closer to user characteristics.
In Fig. 7, the identical symbol logo of structure identical with example IV herein mainly sets the feedback of the present embodiment and training
The structure of meter is described in detail, and other content repeats no more.
The present embodiment further includes receiving module 701, update module 702 and training module 703.Wherein, receiving module 701 is used
The second response result that the problem of actively asking a question is fed back is directed in receiving user.Update module 702 is used in the second response knot
The related information of customer parameter is extracted in fruit, and the customer parameter of user information database is updated according to related information.Training module 703
Process of actively asking a question is trained according to the second response result, so that the user after actively ask a question problem and the update that generate
Parameter matching.
Specifically, the process being trained to process of actively asking a question includes the update of matching process and talks with criterion more
Newly.Training module 703 is also capable of calling matching module 505 and module 506 of actively asking a question.It is embodied in and is added to when in user information database
New customer parameter is revised as after new customer parameter, and matching module 505 is needed according to the customer parameter after update
It is matched, it is determined whether there is association;Correspondingly, after module 506 of actively asking a question is according to the selection of dialogue criterion and updating
The problem of customer parameter matching degree highest.Since in this way, by carrying out intelligent training to system, make system in subsequent question and answer mistake
The problem of being designed in journey closer to user characteristics.
Embodiment six
Fig. 8 is the structural schematic diagram of the two-way chat system of the present embodiment.Compared with both examples above, the present embodiment
System be not only able to achieve nan-machine interrogation, moreover it is possible to initiate to meet the chat topics of user characteristics in the chat process for class of talking in professional jargon,
It is able to ascend the usage experience of user, has stronger practicability.
Fig. 8 is the improvement made on the basis of Fig. 7, and structure same as figure 7 repeats no more.
The two-way chat system of the present embodiment further includes obtaining module 801, Scenario Design module 802 and chat module 803.
Wherein, the chat message that module 801 is used to receive the class of talking in professional jargon of user is obtained.Scenario Design module 802 according to user for believing
The customer parameter design chat scenario updated in breath library, chat module 803, which is used to generate under chat scenario, meets customer parameter
Dialogue of talking in professional jargon.
Wherein, receiving module 701 can also receive dialogue of the user in chat process, and updated and used by update module 702
Family information bank, while 703 pairs of training module processes of actively asking a question carry out intelligent training.
It should be noted that actively question complements each other promotion with the two processes of initiating to chat.The actively interaction of question
Point is more, can more form the customer parameter of more detailed reaction user characteristics, so that it is interested to initiate user in chat process
Or topic that meet user's application demand, formed with user and more well interacted.
While it is disclosed that embodiment content as above but described only to facilitate understanding the present invention and adopting
Embodiment is not intended to limit the 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 scope of patent protection of the invention, still should be subject to the scope of the claims as defined in the appended claims.
Claims (14)
1. a kind of two-way question and answer data processing method based on intelligent robot, which comprises the following steps:
It obtains user's question information and carries out intention assessment;
It is intended to generate the first response result for corresponding to the question information according to user;
When being associated with when the customer parameter composition stored in the question information and user information database, into process of actively asking a question;
Topic is interacted based on the associated customer parameter determination of the question information, according to the interactive topic according to preset right
Words rule generates problem of actively asking a question corresponding with the question information;
Export first response result and the problem of actively asking a question.
2. two-way question and answer data processing method as described in claim 1, which is characterized in that further include:
Receive the second response result for the problem feedback of actively asking a question;
The related information of customer parameter is extracted in the second response result, and the user of user information database is updated according to related information
Parameter.
3. two-way question and answer data processing method as claimed in claim 2, which is characterized in that further include:
Process of actively asking a question is trained according to second response result, updates session rules so that the active question generated
Problem is matched with the customer parameter after updating.
4. two-way question and answer data processing method as described in claim 1, which is characterized in that further include:
The scene conjunctive word in the question information is extracted, according to the context identification pair of the scene conjunctive word and current session
Talk about scene;
Include: in the progress intention assessment the step of
Semantic parsing is carried out to the question information according to domain model, intention assessment is carried out in the dialogue scenarios.
5. such as two-way question and answer data processing method of any of claims 1-4, which is characterized in that according to preset right
Words rule generates the actively problem of asking a question corresponding with the question information
Extract at least one problem to be selected corresponding with question information;
The problem of being selected from least one described problem to be selected with customer parameter matching degree highest is as problem of actively asking a question.
6. two-way question and answer data processing method as claimed in claim 5, which is characterized in that further include:
When it fails to match at least one described problem to be selected and customer parameter, generates the active adapted to the customer parameter and send out
It asks questions.
7. two-way question and answer data processing method as claimed in claim 5, which is characterized in that further include:
Receive user's chat message;
Chat scenario is designed according to the customer parameter updated in user information database;
The dialogue of talking in professional jargon for meeting customer parameter is generated under the chat scenario.
8. a kind of two-way question answering system based on intelligent robot characterized by comprising
Intention assessment module is configured to obtain user's question information and carries out intention assessment;
Responder module is configured to be intended to generate the first response result for corresponding to the question information according to user;
Matching module, when being configured to be associated with when the customer parameter composition stored in the question information and user information database, into
Enter process of actively asking a question;
It actively asks a question module, is configured to interact topic with the associated customer parameter determination of the question information, according to institute
It states interactive topic and generates problem of actively asking a question corresponding with the question information according to preset session rules;
Output module is configured as output to first response result and the problem of actively asking a question.
9. two-way question answering system as claimed in claim 8, which is characterized in that further include:
Receiving module is configured to receive the second response result for the problem feedback of actively asking a question;
Update module is configured to extract the related information of customer parameter in the second response result, and more according to related information
The customer parameter of new user information database.
10. two-way question answering system as claimed in claim 9, which is characterized in that further include:
Training module is configured to be trained process of actively asking a question according to the second response result, update session rules so that
The problem of actively asking a question of generation is matched with the customer parameter after updating.
11. two-way question answering system as claimed in claim 8, which is characterized in that further include:
Scene identification module is configured to extract the scene conjunctive word in the question information, according to the scene conjunctive word and
The context identification dialogue scenarios of current session;
The intention assessment module is also used to carry out semantic parsing to the question information according to domain model, in the dialogue feelings
Intention assessment is carried out in scape.
12. the two-way question answering system as described in any one of claim 8-11, which is characterized in that the module packet of actively asking a question
It includes:
Extracting sub-module is used to extract at least one problem to be selected corresponding with question information;
Submodule is selected, is used to select from least one described problem to be selected and work the problem of customer parameter matching degree highest
For problem of actively asking a question.
13. two-way question answering system as claimed in claim 12, which is characterized in that the module of actively asking a question further include:
Submodule is generated, is used to generate and the use when it fails to match at least one described problem to be selected and customer parameter
The problem of actively asking a question of family parameter adaptation.
14. two-way question answering system as claimed in claim 12, which is characterized in that further include:
Module is obtained, is configured to obtain the chat message of user;
Scenario Design module is configured to design chat scenario according to the customer parameter updated in user information database;
Chat module is configured to generate the dialogue of talking in professional jargon for meeting customer parameter under the chat scenario.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510857525.1A CN105512228B (en) | 2015-11-30 | 2015-11-30 | A kind of two-way question and answer data processing method and system based on intelligent robot |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510857525.1A CN105512228B (en) | 2015-11-30 | 2015-11-30 | A kind of two-way question and answer data processing method and system based on intelligent robot |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105512228A CN105512228A (en) | 2016-04-20 |
CN105512228B true CN105512228B (en) | 2018-12-25 |
Family
ID=55720210
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510857525.1A Active CN105512228B (en) | 2015-11-30 | 2015-11-30 | A kind of two-way question and answer data processing method and system based on intelligent robot |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105512228B (en) |
Families Citing this family (56)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105912697B (en) * | 2016-04-25 | 2019-08-27 | 北京光年无限科技有限公司 | A kind of optimization method and device of conversational system knowledge base |
CN105898487B (en) * | 2016-04-28 | 2019-02-19 | 北京光年无限科技有限公司 | A kind of exchange method and device towards intelligent robot |
CN105798918B (en) * | 2016-04-29 | 2018-08-21 | 北京光年无限科技有限公司 | A kind of exchange method and device towards intelligent robot |
CN105945949A (en) * | 2016-06-01 | 2016-09-21 | 北京光年无限科技有限公司 | Information processing method and system for intelligent robot |
CN107547354A (en) * | 2016-06-27 | 2018-01-05 | 南京中兴软件有限责任公司 | Business datum provides method and device |
WO2018000205A1 (en) * | 2016-06-28 | 2018-01-04 | 深圳狗尾草智能科技有限公司 | Question answering method and system based on multiple intents and multiple skill packets, and robot |
CN106462384B (en) * | 2016-06-29 | 2019-05-31 | 深圳狗尾草智能科技有限公司 | Based on multi-modal intelligent robot exchange method and intelligent robot |
WO2018006367A1 (en) * | 2016-07-07 | 2018-01-11 | 深圳狗尾草智能科技有限公司 | Item purchasing method and system based on multimodal input in games |
CN106113054B (en) * | 2016-07-12 | 2018-09-14 | 上海木木机器人技术有限公司 | Service processing method based on robot |
WO2018010635A1 (en) * | 2016-07-14 | 2018-01-18 | 腾讯科技(深圳)有限公司 | Method of generating random interactive data, network server, and smart conversation system |
CN106528522A (en) * | 2016-08-26 | 2017-03-22 | 南京威卡尔软件有限公司 | Scenarized semantic comprehension and dialogue generation method and system |
CN106372195A (en) * | 2016-08-31 | 2017-02-01 | 北京光年无限科技有限公司 | Human-computer interaction method and device for intelligent robot |
CN106446141B (en) * | 2016-09-21 | 2020-05-19 | 北京光年无限科技有限公司 | Interactive data processing method for intelligent robot system and robot system |
CN107885744B (en) * | 2016-09-29 | 2023-01-03 | 微软技术许可有限责任公司 | Conversational data analysis |
JP6929539B2 (en) * | 2016-10-07 | 2021-09-01 | 国立研究開発法人情報通信研究機構 | Non-factoid question answering system and method and computer program for it |
CN106426203A (en) * | 2016-11-02 | 2017-02-22 | 旗瀚科技有限公司 | Communication system and method of active trigger robot |
CN107861961A (en) * | 2016-11-14 | 2018-03-30 | 平安科技(深圳)有限公司 | Dialog information generation method and device |
CN106599124A (en) * | 2016-11-30 | 2017-04-26 | 竹间智能科技(上海)有限公司 | System and method for actively guiding user to perform continuous conversation |
CN106559321A (en) * | 2016-12-01 | 2017-04-05 | 竹间智能科技(上海)有限公司 | The method and system of dynamic adjustment dialog strategy |
CN106776926A (en) * | 2016-12-01 | 2017-05-31 | 竹间智能科技(上海)有限公司 | Improve the method and system of responsibility when robot talks with |
CN106777081A (en) * | 2016-12-13 | 2017-05-31 | 竹间智能科技(上海)有限公司 | Method and device for determining conversational system acknowledgment strategy |
CN106649704B (en) * | 2016-12-20 | 2020-04-07 | 竹间智能科技(上海)有限公司 | Intelligent conversation control method and system |
WO2018112896A1 (en) * | 2016-12-23 | 2018-06-28 | 深圳前海达闼云端智能科技有限公司 | Chat interaction method and apparatus, and electronic device thereof |
CN106960005A (en) * | 2017-02-20 | 2017-07-18 | 北京光年无限科技有限公司 | A kind of output intent and robot for artificial intelligence robot |
CN106909677B (en) * | 2017-03-02 | 2020-09-08 | 腾讯科技(深圳)有限公司 | Method and device for generating question |
WO2018157349A1 (en) * | 2017-03-02 | 2018-09-07 | 深圳前海达闼云端智能科技有限公司 | Method for interacting with robot, and interactive robot |
US10503739B2 (en) * | 2017-04-20 | 2019-12-10 | Breville USA, Inc. | Crowdsourcing responses in a query processing system |
CN107133349B (en) * | 2017-05-24 | 2018-02-23 | 北京无忧创新科技有限公司 | One kind dialogue robot system |
CN107053208B (en) * | 2017-05-24 | 2018-06-01 | 北京无忧创新科技有限公司 | A kind of method of active dialog interaction robot system and the system active interlocution |
CN107225579A (en) * | 2017-05-27 | 2017-10-03 | 众德云格机器人(苏州)有限公司 | The machine question answering system of rhetorical question formula |
CN107273487A (en) * | 2017-06-13 | 2017-10-20 | 北京百度网讯科技有限公司 | Generation method, device and the computer equipment of chat data based on artificial intelligence |
CN109145096A (en) * | 2017-06-27 | 2019-01-04 | 中国海洋大学 | The daily robot automatically request-answering system of accompanying and attending to of personalization in rule-based library |
CN107480766B (en) * | 2017-07-18 | 2021-01-22 | 北京光年无限科技有限公司 | Method and system for content generation for multi-modal virtual robots |
CN107357787B (en) * | 2017-07-26 | 2021-04-02 | 微鲸科技有限公司 | Semantic interaction method and device and electronic equipment |
CN107562788B (en) * | 2017-07-28 | 2021-02-19 | 深圳前海微众银行股份有限公司 | Interaction method, interaction device and computer-readable storage medium |
JP7010000B2 (en) * | 2017-11-14 | 2022-01-26 | 富士フイルムビジネスイノベーション株式会社 | Information processing equipment and programs |
CN107967308B (en) | 2017-11-16 | 2021-04-23 | 百度在线网络技术(北京)有限公司 | Intelligent interaction processing method, device, equipment and computer storage medium |
US11836592B2 (en) | 2017-12-15 | 2023-12-05 | International Business Machines Corporation | Communication model for cognitive systems |
CN108109689B (en) * | 2017-12-29 | 2023-09-29 | 李向坤 | Diagnosis and treatment session method and device, storage medium and electronic equipment |
CN108114469A (en) * | 2018-01-29 | 2018-06-05 | 北京神州泰岳软件股份有限公司 | Game interaction method, apparatus, terminal and game interaction model based on dialogue |
TWI676902B (en) * | 2018-01-31 | 2019-11-11 | 中華電信股份有限公司 | Method and system of analyzing conversation contents using chatting robot |
CN108319978B (en) * | 2018-02-01 | 2021-01-22 | 北京捷通华声科技股份有限公司 | Semantic similarity calculation method and device |
CN108564833B (en) * | 2018-04-28 | 2021-08-03 | 北京比特智学科技有限公司 | Intelligent interactive conversation control method and device |
CN110750626B (en) * | 2018-07-06 | 2022-05-06 | 中国移动通信有限公司研究院 | Scene-based task-driven multi-turn dialogue method and system |
CN109460451A (en) * | 2018-09-29 | 2019-03-12 | 厦门快商通信息技术有限公司 | Actively obtain method, the intelligent customer service method and system of user information |
CN109473101B (en) * | 2018-12-20 | 2021-08-20 | 瑞芯微电子股份有限公司 | Voice chip structure and method for differentiated random question answering |
CN109787881A (en) * | 2018-12-26 | 2019-05-21 | 广州灵聚信息科技有限公司 | A kind of dialogue method and device with forecast function |
CN109800430B (en) * | 2019-01-18 | 2023-06-27 | 广东小天才科技有限公司 | Semantic understanding method and system |
CN110166640B (en) * | 2019-05-31 | 2021-05-14 | 重庆先特服务外包产业有限公司 | System and method for establishing call center general information model |
CN110245218A (en) * | 2019-06-18 | 2019-09-17 | 广州智伴人工智能科技有限公司 | A kind of voice response exchange method |
CN112133147A (en) * | 2019-06-24 | 2020-12-25 | 武汉慧人信息科技有限公司 | Online automatic teaching and interaction system based on teaching plan and preset question bank |
CN110610705B (en) * | 2019-09-20 | 2023-07-25 | 上海数鸣人工智能科技有限公司 | Voice interaction prompter based on artificial intelligence |
CN111400472A (en) * | 2020-03-16 | 2020-07-10 | 上海百事通信息技术股份有限公司 | Legal consultation intelligent question-answering method and system |
CN112000786A (en) * | 2020-06-30 | 2020-11-27 | 北京来也网络科技有限公司 | Dialogue robot problem processing method, device and equipment combining RPA and AI |
CN114416931A (en) * | 2020-10-28 | 2022-04-29 | 华为云计算技术有限公司 | Label generation method and device and related equipment |
CN114697280A (en) * | 2022-03-01 | 2022-07-01 | 西安博纳吉生物科技有限公司 | Instant messaging method for preset content |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20150074112A1 (en) * | 2012-05-14 | 2015-03-12 | Huawei Technologies Co., Ltd. | Multimedia Question Answering System and Method |
CN104978360A (en) * | 2014-04-11 | 2015-10-14 | 俞志晨 | Realization method of question-answering system with account identity |
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 |
-
2015
- 2015-11-30 CN CN201510857525.1A patent/CN105512228B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20150074112A1 (en) * | 2012-05-14 | 2015-03-12 | Huawei Technologies Co., Ltd. | Multimedia Question Answering System and Method |
CN104978360A (en) * | 2014-04-11 | 2015-10-14 | 俞志晨 | Realization method of question-answering system with account identity |
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 |
Also Published As
Publication number | Publication date |
---|---|
CN105512228A (en) | 2016-04-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105512228B (en) | A kind of two-way question and answer data processing method and system based on intelligent robot | |
CN107665706B (en) | Rapid voice interaction method and system | |
CN107846350B (en) | Method, computer readable medium and system for context-aware network chat | |
KR102170563B1 (en) | Human machine interactive method and apparatus based on artificial intelligence | |
CN108804698A (en) | Man-machine interaction method, system, medium based on personage IP and equipment | |
US9753914B2 (en) | Natural expression processing method, processing and response method, device, and system | |
CN107193978A (en) | A kind of many wheel automatic chatting dialogue methods and system based on deep learning | |
CN103458056B (en) | Speech intention judging system based on automatic classification technology for automatic outbound system | |
KR101634086B1 (en) | Method and computer system of analyzing communication situation based on emotion information | |
CN107665708A (en) | Intelligent sound exchange method and system | |
US9818405B2 (en) | Dialog management system | |
WO2017112813A1 (en) | Multi-lingual virtual personal assistant | |
CN108345692A (en) | A kind of automatic question-answering method and system | |
CN105704013A (en) | Context-based topic updating data processing method and apparatus | |
CN106297773A (en) | A kind of neutral net acoustic training model method | |
CN106326440A (en) | Human-computer interaction method and device facing intelligent robot | |
CN105798918A (en) | Interactive method and device for intelligent robot | |
CN106294854A (en) | A kind of man-machine interaction method for intelligent robot and device | |
CN111831798A (en) | Information processing method, information processing device, electronic equipment and computer readable storage medium | |
CN106803422A (en) | A kind of language model re-evaluation method based on memory network in short-term long | |
CN111145721A (en) | Personalized prompt language generation method, device and equipment | |
CN106952648A (en) | A kind of output intent and robot for robot | |
WO2021135457A1 (en) | Recurrent neural network-based emotion recognition method, apparatus, and storage medium | |
CN108376144A (en) | Man-machine more wheel dialogue methods that scene based on deep neural network automatically switches | |
JP2019036171A (en) | System for assisting in creation of interaction scenario corpus |
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 |