CN108228637A - Natural language client auto-answer method and system - Google Patents
Natural language client auto-answer method and system Download PDFInfo
- Publication number
- CN108228637A CN108228637A CN201611185779.4A CN201611185779A CN108228637A CN 108228637 A CN108228637 A CN 108228637A CN 201611185779 A CN201611185779 A CN 201611185779A CN 108228637 A CN108228637 A CN 108228637A
- Authority
- CN
- China
- Prior art keywords
- answer
- candidate
- user
- module
- slot position
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/3332—Query translation
- G06F16/3334—Selection or weighting of terms from queries, including natural language queries
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/3332—Query translation
- G06F16/3338—Query expansion
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Human Computer Interaction (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention discloses a kind of natural language client auto-answer method and system.This method includes：Database is inquired, obtains the answer of user's primal problem, wherein example of the answer to meet condition in database, all properties of example are as candidate slot position；Judge whether answer quantity is more than predetermined value；If answer quantity is more than predetermined value, propose to ask in reply to user；If answer quantity is not more than predetermined value, judge whether to inquire answer；If inquiring answer and answer quantity being no more than predetermined value, answer is returned to user.Using user demand as the starting point of service, user only needs description problem, does not need to user and understand service content in advance the present invention；Can the required information of user directly can obtain the natural language description of actual demand according to user.
Description
Technical field
The present invention relates to natural language processing field, more particularly to a kind of natural language client auto-answer method and it is
System.
Background technology
Customer service system is the system that an operation flow, strategy are mutually coordinated with technology, it has provided acquisition to the user
The appropriate channel of information creates customer value and enterprise value by way of interactive dialogue.
Operator attendance response and automatic-answering back device are current most important two kinds of client's question answering systems.
Although operator attendance response is capable of providing better user service experience, but still have following defect：
1st, high labor cost needs a large amount of personnel to reduce user and waits for service time；
2nd, repeated work is more, when handling general service content, there is the labour of a large amount of simple repeatability；
3rd, profession requirement is high, in processing complicated service procedure and business information, the plenty of time is needed to improve service ripe
Practice degree.
Automatic-answering back device question answering system can greatly make up the deficiency of operator attendance response, has the following advantages：
1st, human cost is saved, can partly replace the function of operator attendance response；
2nd, fast response time, for content it is clear and definite the problem of can efficiently respond；
3rd, persistent service, automatic-answering back device can accomplish 7x24 hours without not service.
But the existing most logic of automatic answering system is simple, function is single, have ignored user's history problem up and down
Literary information；When user view is complicated or not clear and definite enough, often gives an irrelevant answer or refuse to answer, user experience is poor；Meanwhile it passes
The automatic answering system of system understands the energy force difference of user demand, and user is it should be understood that correctly could select and receive after service item
Service, it is impossible to provide correct service content for user demand.Meanwhile the chat robots such as existing siri, the small ice of Microsoft
More application scenarios are chat rather than inquiry fine knowledge, and different field is adhered to separately with professional domain question answering system described herein.
Existing automatically request-answering system is broadly divided into three kinds, is the automatic-answering back device based on key command respectively, based on keyword
The automatic-answering back device of extraction and the automatic-answering back device based on general natural language processing.Three classes system is respectively present following deficiency：
Automatic-answering back device based on key command：
1st, passive service：User can just select correct service type after have to be understood that all service items；
2nd, flexibility is poor：Interaction logic is fixed, and understands the energy force difference of user demand, poor user experience.
Automatic-answering back device based on keyword extraction：
1st, intelligent difference：Do not have natural language processing and understandability, it is impossible to it is fuzzy to understand user demand, do not have reason
The ability of user's request is solved, user must provide correct keyword, could carry out service.
2nd, expansion is poor：The rule of keyword extraction is artificial setting, does not have fast automatic expansion capability, needs people
Work continuous updating.
Automatic-answering back device based on general natural language processing：
1st, do not have the ability of processing complex services logic：Simple answering can only be carried out, in specific application area
Processing capacity is insufficient.
Invention content
In view of above technical problem, the present invention provides a kind of natural language client auto-answer method and systems, are not required to
User is wanted, which to understand service item in advance, can both start service process.
According to an aspect of the present invention, a kind of natural language client auto-answer method is provided, including：
Database is inquired, obtains the answer of user's primal problem, wherein reality of the answer to meet condition in database
Example, all properties of example are as candidate slot position；
Judge whether answer quantity is more than predetermined value；
If answer quantity is not more than predetermined value, judge whether to inquire answer；
If inquiring answer and answer quantity being no more than predetermined value, answer is returned to user.
In one embodiment of the invention, the method further includes：
If not inquiring answer, determining whether subject, it fails to match；
If it fails to match for subject, fuzzy matching is carried out, returns to possible subject list；
If subject matching does not fail, the attributes match failure of inquiry is determined whether；
If the attributes match failure of inquiry, returns to the corresponding list of attribute values of subject example；
If inquiry attributes match do not fail, subject is expanded and condition of relaxing the restriction after inquire again.
In one embodiment of the invention, the carry out fuzzy matching returns to possible subject list and includes：
User's primal problem is analyzed using natural language processing method, extracts the subject of user's primal problem, predicate and fixed
Language restrictive condition；
If accurately it fails to match for subject, qualified example is recalled according to predicate and attribute restrictive condition, as time
Selected works close；
If accurately it fails to match for predicate, according to subject and attribute restrictive condition, the corresponding attribute of subject is recalled, as time
Selected works close；
In candidate translation example set, scored using Method of Fuzzy Matching each matching；
Selection scoring returns to user more than the candidate collection of predetermined threshold, and next step confirmation is carried out by user.
In one embodiment of the invention, it is described subject is expanded and condition of relaxing the restriction after inquire again including：
If subject is the example in database, recursively inquire in all classes belonging to the example upwards, if exist
Some class can meet querying condition, stop when inquiring answer；
If subject is the class in database, recursively inquire in such all parent upwards, if there are some
Parent meets querying condition, stops when inquiring answer；
If subject restrictive condition excessively causes inquiry to attempt to release every restrictive condition, and record successively less than answer
Answer number is returned answer number is minimum as answer.
In one embodiment of the invention, the method further includes：
Whether judgement obtains answer after inquiring again；
If obtaining answer after inquiring again, answer is returned to user；
If answer can not be obtained after inquiring again, user's primal problem is not answered.
In one embodiment of the invention, the method further includes：If answer quantity is more than predetermined value, carried to user
Go out rhetorical question.
In one embodiment of the invention, it is described to propose that rhetorical question includes to user：
Judge whether there is candidate slot position；
If without candidate slot position, answer is blocked；
If there is candidate slot position, score candidate slot position；
Judge whether directly to ask in reply the highest candidate slot position of scoring；
If the corresponding probable value in the candidate slot position is asked in reply in the directly highest candidate slot position of rhetorical question scoring；
If not asking in reply the highest candidate slot position of scoring directly, the candidate slot list after sequence is returned, is selected by user
Interested candidate's slot position.
In one embodiment of the invention, it is described to candidate slot position carry out scoring include：
Obtain the attribute set of the answer set for meeting user's primal problem and candidate slot position；
Entropy after being grouped to each slot position calculation basis slot position；
User is obtained to score to the interest-degree of each slot position；
Summation is weighted to interest-degree scoring and the entropy；
Answer set is updated for the new answer set after user's interaction.
In one embodiment of the invention, described after user returns to answer, the method further includes：
Recall the corresponding example of the answer or all properties of class；
Count the frequency of each candidate attribute；
Obtain the degree of correlation of each candidate attribute and user's primal problem；
Candidate attribute is ranked up according to the frequency and the degree of correlation；
According to the candidate attribute after sequence, the candidate problem of recommendation is generated.
According to another aspect of the present invention, a kind of natural language client automatic answering system is provided, including enquiry module, is answered
Case quantity judgment module, query result judgment module and answer return to module, wherein：
Enquiry module for inquiring database, obtains the answer of user's primal problem, wherein the answer is in database
Meet the example of condition, all properties of example are as candidate slot position；
Answer quantity judgment module, for judging whether answer quantity is more than predetermined value；
Query result judgment module for the judging result according to answer quantity judgment module, is not more than in answer quantity
In the case of predetermined value, judge whether to inquire answer；
Answer returns to module, for the judging result according to query result judgment module, in the case where inquiring answer,
Answer is returned to user.
In one embodiment of the invention, the system also includes subject matching judgment module, fuzzy matching module, categories
Property matching judgment module, attribute list return to module and subject enlargement module, wherein：
Subject matching judgment module for the judging result according to query result judgment module, is not inquiring answer
In the case of, determining whether subject, it fails to match；
Fuzzy matching module, for the judging result according to subject matching judgment module, in the subject situation that it fails to match
Under, fuzzy matching is carried out, returns to possible subject list；
Attributes match judgment module for the judging result according to subject matching judgment module, does not fail in subject matching
In the case of, determine whether the attributes match failure of inquiry；
Attribute list returns to module, for the judging result according to attributes match judgment module, in the attributes match of inquiry
In the case of failure, the corresponding list of attribute values of subject example is returned；
Subject enlargement module for the judging result according to attributes match judgment module, does not lose in the attributes match of inquiry
In the case of losing, subject is expanded and condition of relaxing the restriction after inquire again.
In one embodiment of the invention, the fuzzy matching module includes condition extraction unit, candidate collection determines
Unit, vague marking unit and candidate collection returning unit, wherein：
Condition extraction unit, for natural language processing method to be used to analyze user's primal problem, user is original asks for extraction
Subject, predicate and the attribute restrictive condition of topic；
In the case that accurately it fails to match in subject, item is limited according to predicate and attribute for candidate collection determination unit
Part recalls qualified example, as candidate collection；In the case that in predicate, accurately it fails to match, according to subject and attribute
Restrictive condition recalls the corresponding attribute of subject, as candidate collection；
Vague marking unit, in candidate translation example set, being scored using Method of Fuzzy Matching each matching；
Candidate collection returning unit, for scoring to be selected to return to user more than the candidate collection of predetermined threshold, by user
Carry out next step confirmation.
In one embodiment of the invention, the subject enlargement module includes the first query unit, the second query unit
With condition lifting unit, wherein：
First query unit in the case of in subject for the example in database, recursively inquires the example upwards
In affiliated all classes, if there are some classes can meet querying condition, stop when inquiring answer；
In the case of in subject for the class in database, it is all recursively to inquire such upwards for second query unit
Parent in, if there are some parents to meet querying condition, stops when inquiring answer；
Condition lifting unit in the case of excessively leading to inquiry less than answer in subject restrictive condition, is attempted successively
Every restrictive condition is released, and records answer number, is returned answer number is minimum as answer.
In one embodiment of the invention, the system also includes query result judgment module again, wherein：
Query result judgment module again, for judging whether obtain answer after inquiring again；Answer is obtained after inquiring again
In the case of, instruction answer returns to module and returns to answer to user；And it in the case of after inquiring again can not obtaining answer, does not return
Answer user's primal problem.
In one embodiment of the invention, the system also includes rhetorical question module, wherein：
Module is asked in reply, for the judging result according to answer quantity judgment module, is more than the feelings of predetermined value in answer quantity
Under condition, propose to ask in reply to user.
In one embodiment of the invention, the rhetorical question module includes candidate slot position judging unit, unit is blocked in answer,
Score unit and rhetorical question judging unit for slot position, wherein：
Candidate slot position judging unit for the judging result according to answer quantity judgment module, is more than pre- in answer quantity
In the case of definite value, candidate slot position is judged whether there is；
Unit is blocked in answer, for the judging result according to candidate slot position judging unit, the situation in no candidate slot position
Under, answer is blocked；
Score unit for slot position, for the judging result according to candidate slot position judging unit, in the case where there is candidate slot position,
It scores candidate slot position；
Judging unit is asked in reply, after scoring in slot position scoring unit candidate slot position, is judged whether directly anti-
Ask scoring highest candidate slot position；In the case of the highest candidate slot position of directly rhetorical question scoring, it is right to ask in reply the candidate slot position
The probable value answered；With in the case where not asking in reply the highest candidate slot position of scoring directly, the candidate slot list after sequence is returned,
Interested candidate slot position is selected by user.
In one embodiment of the invention, it is true to include answer set acquisition submodule, entropy for the slot position scoring unit
Stator modules, interest-degree determination sub-module, weighted sum submodule and answer update submodule, wherein：
Answer set acquisition submodule meets the answer set of user's primal problem and the attribute of candidate slot position for obtaining
Set；
Entropy determination sub-module, for the entropy after the grouping of each slot position calculation basis slot position；
Interest-degree determination sub-module scores to the interest-degree of each slot position for obtaining user；
Weighted sum submodule, for being weighted summation to interest-degree scoring and the entropy；
Answer updates submodule, for updating answer set for the new answer set after user's interaction.
In one embodiment of the invention, the system also includes attributes to recall module, frequency statistics module, the degree of correlation
Determining module, sorting module and recommendation problem generation module, wherein：
Attribute recalls module, and for returning to module after user returns to answer in answer, it is corresponding to recall the answer
The all properties of example or class；
Frequency statistics module, for counting the frequency of each candidate attribute；
Degree of correlation determining module, for obtaining the degree of correlation of each candidate attribute and user's primal problem；
Sorting module, for being ranked up according to the frequency and the degree of correlation to candidate attribute；
Recommendation problem generation module for the candidate attribute after sorting according to sorting module, generates the candidate problem of recommendation.
Using user demand as the starting point of service, user only needs description problem, it is advance not to need to user the present invention
Understand service content；Can the required letter of user directly can obtain the natural language description of actual demand according to user
Breath.
Description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, to embodiment or will show below
There is attached drawing needed in technology description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention, for those of ordinary skill in the art, without creative efforts, can be with
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is the schematic diagram of natural language client auto-answer method first embodiment of the present invention.
Fig. 2 is the schematic diagram of natural language client auto-answer method second embodiment of the present invention.
Fig. 3 is the schematic diagram of the dynamic groove position selection method based on kolmogorov sinai entropy in one embodiment of the invention.
Fig. 4 is the flow diagram of the intelligent interaction function based on fuzzy matching in one embodiment of the invention.
Fig. 5 is the flow diagram of the intelligent answer retrieval based on body construction in one embodiment of the invention.
Fig. 6 is the schematic diagram of natural language client auto-answer method 3rd embodiment of the present invention.
Fig. 7 is the schematic diagram of natural language client automatic answering system first embodiment of the present invention.
Fig. 8 is the schematic diagram of natural language client automatic answering system second embodiment of the present invention.
Fig. 9 is the schematic diagram of fuzzy matching module in one embodiment of the invention.
Figure 10 is the schematic diagram of subject enlargement module in one embodiment of the invention.
Figure 11 is the schematic diagram that module is asked in reply in one embodiment of the invention.
Figure 12 is the schematic diagram of one embodiment of the invention bracket groove position scoring unit.
Figure 13 is the schematic diagram of natural language client automatic answering system 3rd embodiment of the present invention.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present invention, the technical solution in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, instead of all the embodiments.Below
Description only actually at least one exemplary embodiment is illustrative, is never used as to the present invention and its application or makes
Any restrictions.Based on the embodiments of the present invention, those of ordinary skill in the art are not making creative work premise
Lower all other embodiments obtained, shall fall within the protection scope of the present invention.
Unless specifically stated otherwise, the component and positioned opposite, the digital table of step otherwise illustrated in these embodiments
It is not limited the scope of the invention up to formula and numerical value.
Simultaneously, it should be appreciated that for ease of description, the size of the various pieces shown in attached drawing is not according to reality
Proportionate relationship draw.
Technology, method and apparatus known to person of ordinary skill in the relevant may be not discussed in detail, but suitable
In the case of, the technology, method and apparatus should be considered as authorizing part of specification.
In shown here and discussion all examples, any occurrence should be construed as merely illustrative, without
It is as limitation.Therefore, the other examples of exemplary embodiment can have different values.
It should be noted that：Similar label and letter represents similar terms in following attached drawing, therefore, once a certain Xiang Yi
It is defined in a attached drawing, then in subsequent attached drawing does not need to that it is further discussed.
Fig. 1 is the schematic diagram of natural language client auto-answer method first embodiment of the present invention.Preferably, the present embodiment
It can be performed by natural language client automatic answering system of the present invention.This method includes the following steps：
Step 101, inquiry database (ontology) obtains the answer of user's primal problem, wherein the answer is database
The middle example for meeting condition, all properties of example are as candidate slot position.
In one embodiment of the invention, the database is ontology knowledge base, i.e. service business knowledge base.
Step 102, judge whether answer quantity is more than predetermined value.
Step 103, if answer quantity is not more than predetermined value, judge whether to inquire answer.
Step 104, if inquiring answer and answer quantity no more than predetermined value, answer is returned to user.
Based on natural language client's auto-answer method that the above embodiment of the present invention provides, using user demand as service
Starting point, user only need description problem, do not need to user and understand service content in advance；It can be directly according to user to reality
The natural language description of border demand can obtain the required information of user；It can intelligently be searched in service business knowledge base
Rope, reasoning obtain best suiting the service content of service logic design object.
Fig. 2 is the schematic diagram of natural language client auto-answer method second embodiment of the present invention.Preferably, the present embodiment
It can be performed by natural language client automatic answering system of the present invention.This method includes the following steps：
Step 201, user's primal problem input by user is received.
Step 202, answer initialization is carried out.Natural language analysis is carried out, and use SPARQL to user's primal problem
Meet in (Simple Protocol and RDF Query Language, query language and data acquisition protocols) inquiry ontology
The example of condition is as initial answer.Wherein, all properties of example are as candidate slot position.
Step 203, judge whether answer quantity is more than predetermined value (judging whether that answer is excessive).If answer quantity is more than
Predetermined value (i.e. answer is excessive), then perform step 204；Otherwise, if answer quantity is not more than predetermined value, step 210 is performed.
Step 204, candidate slot position (optional slot position) is judged whether there is.If without candidate slot position, step 205 is performed；It is no
Then, if there is candidate slot position, step 206 is performed.
Step 205, answer is blocked；Other steps of the present embodiment are no longer performed later.
Step 206, it according to answer initialization result, scores candidate slot position.Based on dynamic groove position assessment algorithm pair
Each slot position is scored, and obtains the priority of each slot position.
Step 207, erotetic logic selects：According to the marking situation of each slot position, this module can select different rhetorical questions to patrol
Volume.That is, it is judged that whether directly ask in reply the highest candidate slot position of scoring.If the directly highest candidate slot position of rhetorical question scoring, performs
Step 208；Otherwise, if not asking in reply the highest candidate slot position of scoring directly, step 209 is performed.
Step 208, if the slot position scoring of highest priority reaches threshold value or scoring is far above other candidate slot positions, directly
The corresponding probable value in the highest candidate's slot position of rhetorical question scoring；
Step 209, the candidate slot list after sequence is returned, interested candidate slot position or slot position is selected to belong to by user
Property.
In one embodiment of the invention, after step 208 or step 209, the method can also include：Generation is anti-
Ask sentence and option, that is, according to erotetic logic, natural language sentences are generated, and be presented to based on spatial term technology
User.Later, other steps of the present embodiment are no longer performed.
Step 210, judge whether not inquire answer.If inquiring answer, step 211 is performed；Otherwise, if not looking into
Answer is ask, then performs step 212.
Step 211, answer is returned to user；Other steps of this implementation are no longer performed later.
Step 212, determine whether that it fails to match for subject.If it fails to match for subject, step 213 is performed；Otherwise, it is if main
Language matching does not fail, then performs step 214.
Step 213, fuzzy matching is carried out, returns to possible subject list；Other steps of this implementation are no longer performed later.
Step 214, determine whether the attributes match failure of inquiry.If the attributes match failure of inquiry, performs step
215；Otherwise, if the attributes match of inquiry does not fail, step 216 is performed.
Step 215, the corresponding list of attribute values of subject example is returned；Other steps of this implementation are no longer performed later.
Step 216, subject is expanded and condition of relaxing the restriction after inquire again.
Step 217, judge whether obtain answer after inquiring again.If obtaining answer after inquiring again, step 211 is performed；It is no
Then, if answer can not be obtained after inquiring again, step 218 is performed.
Step 218, user's primal problem is not answered.
The specific refinement for the step of step 204- steps 209 is the proposition rhetorical questions to user in the above embodiment of the present invention
Step.
Using user demand as the starting point of service, user only needs description problem, is not required to the above embodiment of the present invention
User is wanted to understand service content in advance.
The above embodiment of the present invention directly can obtain use according to user to the natural language description of actual demand
The required information in family.
The above embodiment of the present invention can be putd question to actively to user, and based on suitable interaction sequences are calculated, be promoted
Interactive efficiency.
The above embodiment of the present invention can intelligently search in service business knowledge base, reasoning, obtain best suiting business
The service content of logical design target.
The above embodiment of the present invention has the ability of processing complicated business logic and targetedly to meet user definite
The ability of demand.
The above embodiment of the present invention can be interacted by talking with, gradually clear and definite user's query intention of stratification, so as to real
The complicated business logic now serviced.
The above embodiment of the present invention can help user's selection to meet the clothes of its demand according to user demand and service logic
Business.
In one embodiment of the invention, in Fig. 1 or Fig. 2 embodiments, the method can also include：Conversation history is believed
Breath is preserved and recovered.
The above embodiment of the present invention can be adapted for the dialogue interaction mechanism of natural language client's question answering system.
Several committed steps of Fig. 2 embodiments of the present invention are further described below by specific embodiment.
Fig. 3 is the schematic diagram of the dynamic groove position selection method based on kolmogorov sinai entropy in one embodiment of the invention.Such as Fig. 3 institutes
Show, the step 206 in Fig. 2 embodiments, that is, described the step of scoring candidate slot position can include：
Step 2061, the attribute set of the answer set for meeting user's primal problem and candidate slot position is obtained.
Step 2062 is to the entropy entr after the grouping of each slot position (attribute) calculation basis slot position.
Step 2063, with reference in electric business website label, comment information, to each slot position (attribute) respectively statistical attribute
The keyword frequency, as the interest-degree of attribute scoring pop after normalization.
Step 2064, summation is weighted to interest-degree scoring and the entropy.
Final score score (slot)=w1×entr+w2× pop understands w based on experience value1Preferably 0.4, w2It is preferred that
It is 0.6.
Step 2065, update answer set is the new answer set after user's interaction, performs step 2061 later.By repeatedly
In generation, calculates step 2061- steps 2065, until answer number is met the requirements.
The above embodiment of the present invention realizes the dynamic groove position assessment based on comentropy, so as to make user more efficiently
Final result is obtained, reduces interaction times.
Fig. 4 is the flow diagram of the intelligent interaction function based on fuzzy matching in one embodiment of the invention.Such as Fig. 4 institutes
Show, the step 213 in Fig. 2 embodiments, that is, the carry out fuzzy matching, the step of returning to possible subject list can include：
Step 2131, using natural language processing method analyze user's primal problem, extract user's primal problem subject,
Predicate and attribute restrictive condition.
Step 2132, judging subject, accurately whether matching fails.If accurately it fails to match for subject, step 2133 is performed；
If accurately matching does not fail subject, step 2134 is performed.
Step 2133, qualified example is recalled according to predicate and attribute restrictive condition, as candidate collection.
Step 2134, judging subject, accurately whether matching fails.If accurately it fails to match for subject, step 2135 is performed；
If accurately matching does not fail subject, step 2136 is performed.
Step 2135, according to subject and attribute restrictive condition, the corresponding attribute of subject is recalled, as candidate collection.
Step 2136, it can not answer.
In one embodiment of the invention, after step 2133 and step 2135, the above method can also include：It is waiting
It selects in example collection, using Method of Fuzzy Matching (editing distance, longest common subsequence), scores each matching；With
And selection scoring returns to user more than the candidate collection of predetermined threshold, and next step confirmation is carried out by user.
Since the problem of family inputs often has certain fuzziness, ontology knowledge base cannot be accurately matched under normal circumstances
In example or class.So we have used Method of Fuzzy Matching in the entity link stage, it is the fuzzy of Case-based Reasoning name respectively
Matching and the fuzzy matching based on attribute, and propose the interaction mechanism based on fuzzy matching.
Fig. 5 is the flow diagram of the intelligent answer retrieval based on body construction in one embodiment of the invention.Such as Fig. 5 institutes
Show, the step 216 in Fig. 2 embodiments, that is, it is described subject is expanded and condition of relaxing the restriction after the step of inquiring again can be with
Including：
Step 2161, when inquiry is less than answer, judge whether subject is example in database.If subject is database
In example, then perform step 2162；Otherwise, if subject is not the example in database, step 2163 is performed.
Step 2162, it recursively inquires upwards in all classes belonging to the example, if can meet there are some class and look into
Inquiry condition stops when inquiring answer；Step 2165 is performed later.
Step 2163, judge whether subject is class in database.If subject is the class in database, step is performed
2164；Otherwise, if subject is not the class in database, step 2166 is performed.
Step 2164, it recursively inquires upwards in such all parent, if there are some parents to meet inquiry item
Part stops when inquiring answer.
Step 2165, judge whether to obtain answer.If obtaining answer, answer is returned；Otherwise, if not obtaining answer,
Perform step 2166.
Step 2166, attempt to release every restrictive condition successively, and record answer number, using answer number it is minimum as
Answer returns.
Fig. 6 is the schematic diagram of natural language client auto-answer method 3rd embodiment of the present invention.Preferably, the present embodiment
It can be performed by natural language client automatic answering system of the present invention.In Fig. 1 or Fig. 2 embodiments, answer is returned in successful inquiring
Afterwards, this method is further comprising the steps of：
Step 601, the corresponding example of the answer or all properties of class are recalled.
Step 602, the frequency cp of each candidate attribute is counted1。
Step 603, the degree of correlation of each candidate attribute and user's primal problem is obtained.
In an embodiment of the invention, step 603 can include：By attribute vector, calculate and primitive attribute vector
Cosine similarity, as attribute and the degree of correlation information sim of primal problem；
Step 604, candidate attribute is ranked up according to the frequency and the degree of correlation.
In an embodiment of the invention, step 604 can include：Final score is obtained according to the frequency and the degree of correlation
Score (p1, pori)=w1×cp1+w2× sim, wherein understanding w based on experience value1Preferably 0.4, w2Preferably 0.6；According to
Final score carries out descending sort.
Step 605, according to the candidate attribute after descending sort, the candidate problem of recommendation is generated.
In an embodiment of the invention, step 605 can include：According to the candidate attribute after descending sort, by descending
First five after sequence is as the relevant issues recommended.
The above embodiment of the present invention shows user when user clearly inquires in ontology some example or the specific object of class
There is query intention to the example or class, this system, can be intelligently by its of instant example or class while customer problem is answered
The relevant issues that his attribute may be concerned about as user, return to user.
The above embodiment of the present invention actively can recommend other problems associated with the query to user, and user is helped to obtain more
Multi information.
Fig. 7 is the schematic diagram of natural language client automatic answering system first embodiment of the present invention.It is as shown in fig. 7, described
Natural language client automatic answering system can include enquiry module 701, answer quantity judgment module 702, query result and judge
Module 703 and answer return to module 704, wherein：
Enquiry module 701 for inquiring database, obtains the answer of user's primal problem, wherein the answer is data
Meet the example of condition in library, all properties of example are as candidate slot position.
Answer quantity judgment module 702, for judging whether answer quantity is more than predetermined value.
Query result judgment module 703, for the judging result according to answer quantity judgment module 702, in answer quantity
In the case of no more than predetermined value, judge whether to inquire answer.
Answer returns to module 704, for the judging result according to query result judgment module 703, is inquiring answer
In the case of, return to answer to user.
Based on natural language client's automatic answering system that the above embodiment of the present invention provides, using user demand as service
Starting point, user only need description problem, do not need to user and understand service content in advance；It can be directly according to user to reality
The natural language description of border demand can obtain the required information of user；It can intelligently be searched in service business knowledge base
Rope, reasoning obtain best suiting the service content of service logic design object.
Fig. 8 is the schematic diagram of natural language client automatic answering system second embodiment of the present invention.With embodiment illustrated in fig. 7
Compare, in the embodiment shown in fig. 8, the system can also include subject matching judgment module 705, fuzzy matching module 706,
Attributes match judgment module 707, attribute list return to module 708 and subject enlargement module 709, wherein：
Subject matching judgment module 705 for the judging result according to query result judgment module 703, is not inquiring
In the case of answer, determining whether subject, it fails to match.
Fuzzy matching module 706, for the judging result according to subject matching judgment module 705, in subject, it fails to match
In the case of, fuzzy matching is carried out, returns to possible subject list.
Attributes match judgment module 707 for the judging result according to subject matching judgment module 705, is matched in subject
In the case of not failing, the attributes match failure of inquiry is determined whether.
Attribute list returns to module 708, for the judging result according to attributes match judgment module 707, in the category of inquiry
Property in the case that it fails to match, return to the corresponding list of attribute values of subject example.
Subject enlargement module 709, for the judging result according to attributes match judgment module 707, in the attribute of inquiry
With do not fail in the case of, subject is expanded and condition of relaxing the restriction after inquire again.
Fig. 9 is the schematic diagram of fuzzy matching module in one embodiment of the invention.As shown in figure 9, the mould in Fig. 8 embodiments
Condition extraction unit 7061, candidate collection determination unit 7062,7063 and of vague marking unit can be included by pasting matching module 706
Candidate collection returning unit 7064, wherein：
Condition extraction unit 7061, for natural language processing method to be used to analyze user's primal problem, extraction user is former
Subject, predicate and the attribute restrictive condition of beginning problem.
Candidate collection determination unit 7062 in the case that in subject, accurately it fails to match, is limited according to predicate and attribute
Condition processed recalls qualified example, as candidate collection；With it is accurate in the case that it fails to match in predicate, according to subject and
Attribute restrictive condition recalls the corresponding attribute of subject, as candidate collection.
Vague marking unit 7063, in candidate translation example set, being carried out using Method of Fuzzy Matching to each matching
Scoring.
Candidate collection returning unit 7064, for scoring to be selected to return to user more than the candidate collection of predetermined threshold, by
User carries out next step confirmation.
Figure 10 is the schematic diagram of subject enlargement module in one embodiment of the invention.As shown in Figure 10, in Fig. 8 embodiments
Subject enlargement module 709 can include the first query unit 7091, the second query unit 7092 and condition lifting unit 7093,
In：
First query unit 7091, in the case of in subject for the example in database, recursively inquiry should upwards
In all classes belonging to example, if there are some classes can meet querying condition, stop when inquiring answer.
Second query unit 7092 in the case of in subject for the class in database, recursively inquires such upwards
In all parents, if there are some parents to meet querying condition, stops when inquiring answer.
Condition lifting unit 7093, in the case of excessively leading to inquiry less than answer in subject restrictive condition, successively
It attempts to release every restrictive condition, and record answer number, be returned answer number is minimum as answer.
In one embodiment of the invention, as shown in figure 8, the system can also include query result judgment module again
710, wherein：
Query result judgment module 710 again, for judging whether subject enlargement module 709 obtains answer after inquiring again；
In the case of obtaining answer after inquiring again, instruction answer returns to module 704 and returns to answer to user；It and can not after inquiring again
In the case of obtaining answer, user's primal problem is not answered.
In one embodiment of the invention, as shown in figure 8, the system can also include rhetorical question module 711, wherein：
Module 711 is asked in reply, for the judging result according to answer quantity judgment module, is more than predetermined value in answer quantity
In the case of, it proposes to ask in reply to user.
Figure 11 is the schematic diagram that module is asked in reply in one embodiment of the invention.As shown in figure 11, the rhetorical question in Fig. 8 embodiments
Module 711 can include candidate slot position judging unit 7111, unit 7112 is blocked in answer, slot position scoring unit 7113 and rhetorical question are sentenced
Disconnected unit 7114, wherein：
Candidate slot position judging unit 7111, for the judging result according to answer quantity judgment module 702, in answer quantity
In the case of more than predetermined value, candidate slot position is judged whether there is.
Unit 7112 is blocked in answer, for the judging result according to candidate slot position judging unit 7111, in no candidate slot
In the case of position, answer is blocked.
Score unit 7113 for slot position, for the judging result according to candidate slot position judging unit 7111, is there is candidate slot position
In the case of, it scores candidate slot position.
Judging unit 7114 is asked in reply, after scoring in slot position scoring unit 7113 candidate slot position, judgement is
The no highest candidate slot position of directly rhetorical question scoring；In the case of the highest candidate slot position of directly rhetorical question scoring, the time is asked in reply
Select the corresponding probable value in slot position；With in the case where not asking in reply the highest candidate slot position of scoring directly, the candidate after sequence is returned
Slot list selects interested candidate slot position by user.
Figure 12 is the schematic diagram of one embodiment of the invention bracket groove position scoring unit.As shown in figure 12, in Figure 11 embodiments
Slot position scoring unit 7113 can include answer set acquisition submodule 71131, entropy determination sub-module 71132, interest-degree
Determination sub-module 71133, weighted sum submodule 71134 and answer update submodule 71135, wherein：
Answer set acquisition submodule 71131, for obtaining the answer set for meeting user's primal problem and candidate slot position
Attribute set.
Entropy determination sub-module 71132, for the entropy after the grouping of each slot position calculation basis slot position.
Interest-degree determination sub-module 71133 scores to the interest-degree of each slot position for obtaining user.
Weighted sum submodule 71134, for being weighted summation to interest-degree scoring and the entropy.
Answer updates submodule 71135, for updating answer set for the new answer set after user's interaction.
Using user demand as the starting point of service, user only needs description problem, is not required to the above embodiment of the present invention
User is wanted to understand service content in advance.
The above embodiment of the present invention directly can obtain use according to user to the natural language description of actual demand
The required information in family.
The above embodiment of the present invention can be putd question to actively to user, and based on suitable interaction sequences are calculated, be promoted
Interactive efficiency.
The above embodiment of the present invention can intelligently search in service business knowledge base, reasoning, obtain best suiting business
The service content of logical design target.
The above embodiment of the present invention has the ability of processing complicated business logic and targetedly to meet user definite
The ability of demand.
The above embodiment of the present invention can be interacted by talking with, gradually clear and definite user's query intention of stratification, so as to real
The complicated business logic now serviced.
The above embodiment of the present invention can help user's selection to meet the clothes of its demand according to user demand and service logic
Business.
Figure 13 is the schematic diagram of natural language client automatic answering system 3rd embodiment of the present invention.With implementation shown in Fig. 8
Example compare, in the embodiment shown in fig. 13, the system can also be recalled including attribute module 715, frequency statistics module 716,
Degree of correlation determining module 717, sorting module 718 and recommendation problem generation module 719, wherein：
Attribute recalls module 715, for returning to module 704 after user returns to answer in answer, recalls the answer
The all properties of corresponding example or class.
Frequency statistics module 716, for counting the frequency of each candidate attribute.
Degree of correlation determining module 717, for obtaining the degree of correlation of each candidate attribute and user's primal problem.
Sorting module 718, for being ranked up according to the frequency and the degree of correlation to candidate attribute.
Recommendation problem generation module 719, for according to the candidate attribute after the sequence of sorting module 718, generating the time of recommendation
Select problem.
The above embodiment of the present invention shows user when user clearly inquires in ontology some example or the specific object of class
There is query intention to the example or class, this system, can be intelligently by its of instant example or class while customer problem is answered
The relevant issues that his attribute may be concerned about as user, return to user.
The above embodiment of the present invention actively can recommend other problems associated with the query to user, and user is helped to obtain more
Multi information.
Natural language client's automatic answering system described above can be implemented as described herein for performing
General processor, programmable logic controller (PLC) (PLC), digital signal processor (DSP), the application-specific integrated circuit of function
(ASIC), field programmable gate array (FPGA) either other programmable logic device, discrete gate or transistor logic,
Discrete hardware components or it is arbitrary appropriately combined.
So far, the present invention is described in detail.In order to avoid the design of the masking present invention, it is public that this field institute is not described
Some details known.Those skilled in the art as described above, can be appreciated how to implement technology disclosed herein completely
Scheme.
One of ordinary skill in the art will appreciate that hardware can be passed through by realizing all or part of step of above-described embodiment
It completes, relevant hardware can also be instructed to complete by program, the program can be stored in a kind of computer-readable
In storage medium, storage medium mentioned above can be read-only memory, disk or CD etc..
Description of the invention provides for the sake of example and description, and is not exhaustively or will be of the invention
It is limited to disclosed form.Many modifications and variations are obvious for the ordinary skill in the art.It selects and retouches
It states embodiment and is to more preferably illustrate the principle of the present invention and practical application, and those of ordinary skill in the art is enable to manage
The solution present invention is so as to design the various embodiments with various modifications suitable for special-purpose.
Claims (18)
1. a kind of natural language client auto-answer method, which is characterized in that including：
Database is inquired, obtains the answer of user's primal problem, wherein example of the answer to meet condition in database, real
The all properties of example are as candidate slot position；
Judge whether answer quantity is more than predetermined value；
If answer quantity is not more than predetermined value, judge whether to inquire answer；
If inquiring answer and answer quantity being no more than predetermined value, answer is returned to user.
2. it according to the method described in claim 1, it is characterized in that, further includes：
If not inquiring answer, determining whether subject, it fails to match；
If it fails to match for subject, fuzzy matching is carried out, returns to possible subject list；
If subject matching does not fail, the attributes match failure of inquiry is determined whether；
If the attributes match failure of inquiry, returns to the corresponding list of attribute values of subject example；
If inquiry attributes match do not fail, subject is expanded and condition of relaxing the restriction after inquire again.
3. according to the method described in claim 2, it is characterized in that, the carry out fuzzy matching, returns to possible subject list
Including：
User's primal problem is analyzed using natural language processing method, extracts subject, predicate and the attribute limit of user's primal problem
Condition processed；
If accurately it fails to match for subject, qualified example is recalled according to predicate and attribute restrictive condition, as Candidate Set
It closes；
If accurately it fails to match for predicate, according to subject and attribute restrictive condition, the corresponding attribute of subject is recalled, as Candidate Set
It closes；
In candidate translation example set, scored using Method of Fuzzy Matching each matching；
Selection scoring returns to user more than the candidate collection of predetermined threshold, and next step confirmation is carried out by user.
4. according to the method described in claim 2, it is characterized in that, it is described subject is expanded and condition of relaxing the restriction after again
Inquiry includes：
If subject is the example in database, recursively inquire in all classes belonging to the example upwards, if there are a certain
A class can meet querying condition, stop when inquiring answer；
If subject is the class in database, recursively inquire in such all parent upwards, if there are some parents
Meet querying condition, stop when inquiring answer；
If subject restrictive condition excessively causes inquiry to attempt to release every restrictive condition, and record answer successively less than answer
Number is returned answer number is minimum as answer.
5. according to the method described in any one of claim 2-4, which is characterized in that further include：
Whether judgement obtains answer after inquiring again；
If obtaining answer after inquiring again, answer is returned to user；
If answer can not be obtained after inquiring again, user's primal problem is not answered.
6. according to the described method of any one of claim 1-4, which is characterized in that further include：
If answer quantity is more than predetermined value, propose to ask in reply to user.
7. according to the method described in claim 6, it is characterized in that, described propose that rhetorical question includes to user：
Judge whether there is candidate slot position；
If without candidate slot position, answer is blocked；
If there is candidate slot position, score candidate slot position；
Judge whether directly to ask in reply the highest candidate slot position of scoring；
If the corresponding probable value in the candidate slot position is asked in reply in the directly highest candidate slot position of rhetorical question scoring；
If not asking in reply the highest candidate slot position of scoring directly, the candidate slot list after sequence is returned, is selected to feel emerging by user
The candidate slot position of interest.
8. the method according to the description of claim 7 is characterized in that it is described to candidate slot position carry out scoring include：
Obtain the attribute set of the answer set for meeting user's primal problem and candidate slot position；
Entropy after being grouped to each slot position calculation basis slot position；
User is obtained to score to the interest-degree of each slot position；
Summation is weighted to interest-degree scoring and the entropy；
Answer set is updated for the new answer set after user's interaction.
9. according to the described method of any one of claim 1-4, which is characterized in that it is described after user returns to answer, also
Including：
Recall the corresponding example of the answer or all properties of class；
Count the frequency of each candidate attribute；
Obtain the degree of correlation of each candidate attribute and user's primal problem；
Candidate attribute is ranked up according to the frequency and the degree of correlation；
According to the candidate attribute after sequence, the candidate problem of recommendation is generated.
10. a kind of natural language client automatic answering system, which is characterized in that including enquiry module, answer quantity judgment module,
Query result judgment module and answer return to module, wherein：
Enquiry module for inquiring database, obtains the answer of user's primal problem, wherein the answer is meets in database
The example of condition, all properties of example are as candidate slot position；
Answer quantity judgment module, for judging whether answer quantity is more than predetermined value；
Query result judgment module, for the judging result according to answer quantity judgment module, in answer quantity no more than predetermined
In the case of value, judge whether to inquire answer；
Answer returns to module, for the judging result according to query result judgment module, in the case where inquiring answer, Xiang Yong
Family returns to answer.
11. system according to claim 10, which is characterized in that further include subject matching judgment module, fuzzy matching mould
Block, attributes match judgment module, attribute list return to module and subject enlargement module, wherein：
Subject matching judgment module, for the judging result according to query result judgment module, in the situation for not inquiring answer
Under, determining whether subject, it fails to match；
Fuzzy matching module, for the judging result according to subject matching judgment module, in the case that in subject, it fails to match, into
Row fuzzy matching returns to possible subject list；
Attributes match judgment module, for the judging result according to subject matching judgment module, in the feelings that subject matching does not fail
Under condition, the attributes match failure of inquiry is determined whether；
Attribute list returns to module, for the judging result according to attributes match judgment module, fails in the attributes match of inquiry
In the case of, return to the corresponding list of attribute values of subject example；
Subject enlargement module for the judging result according to attributes match judgment module, does not fail in the attributes match of inquiry
In the case of, subject is expanded and condition of relaxing the restriction after inquire again.
12. system according to claim 11, which is characterized in that the fuzzy matching module include condition extraction unit,
Candidate collection determination unit, vague marking unit and candidate collection returning unit, wherein：
Condition extraction unit, for natural language processing method to be used to analyze user's primal problem, extraction user's primal problem
Subject, predicate and attribute restrictive condition；
Candidate collection determination unit in the case that in subject, accurately it fails to match, is called together according to predicate and attribute restrictive condition
Qualified example is returned, as candidate collection；In the case that in predicate, accurately it fails to match, limited according to subject and attribute
Condition recalls the corresponding attribute of subject, as candidate collection；
Vague marking unit, in candidate translation example set, being scored using Method of Fuzzy Matching each matching；
Candidate collection returning unit for scoring to be selected to return to user more than the candidate collection of predetermined threshold, is carried out by user
Confirm in next step.
13. system according to claim 11, which is characterized in that the subject enlargement module include the first query unit,
Second query unit and condition lifting unit, wherein：
First query unit in the case of in subject for the example in database, is recursively inquired belonging to the example upwards
All classes in, if there are some classes can meet querying condition, stop when inquiring answer；
Second query unit in the case of in subject for the class in database, recursively inquires such all father upwards
In class, if there are some parents to meet querying condition, stops when inquiring answer；
Condition lifting unit in the case of excessively leading to inquiry less than answer in subject restrictive condition, is attempted to release successively
Every restrictive condition, and answer number is recorded, it is returned answer number is minimum as answer.
14. according to the system described in any one of claim 11-13, which is characterized in that further include query result again and judge mould
Block, wherein：
Query result judgment module again, for judging whether obtain answer after inquiring again；The situation of answer is obtained after inquiring again
Under, instruction answer returns to module and returns to answer to user；And in the case of after inquiring again can not obtaining answer, do not answer use
Family primal problem.
15. according to the system described in any one of claim 10-13, which is characterized in that rhetorical question module is further included, wherein：
Module is asked in reply, for the judging result according to answer quantity judgment module, in the case where answer quantity is more than predetermined value,
It proposes to ask in reply to user.
16. system according to claim 15, which is characterized in that the rhetorical question module include candidate slot position judging unit,
Unit, slot position scoring unit and rhetorical question judging unit are blocked in answer, wherein：
Candidate slot position judging unit for the judging result according to answer quantity judgment module, is more than predetermined value in answer quantity
In the case of, judge whether there is candidate slot position；
Unit is blocked in answer, right in the case of no candidate slot position for the judging result according to candidate slot position judging unit
Answer is blocked；
Score unit for slot position, for the judging result according to candidate slot position judging unit, in the case where there is candidate slot position, to waiting
Slot position is selected to score；
Judging unit is asked in reply, after scoring in slot position scoring unit candidate slot position, judges whether directly to ask in reply and comment
Divide highest candidate slot position；In the case of the highest candidate slot position of directly rhetorical question scoring, it is corresponding to ask in reply the candidate slot position
Probable value；With directly in the case of the highest candidate slot position of rhetorical question scoring, do not returning to the candidate slot list after sequence, by with
The interested candidate slot position of family selection.
17. system according to claim 16, which is characterized in that the slot position scoring unit includes answer set and obtains son
Module, entropy determination sub-module, interest-degree determination sub-module, weighted sum submodule and answer update submodule, wherein：
Answer set acquisition submodule meets the answer set of user's primal problem and the property set of candidate slot position for obtaining
It closes；
Entropy determination sub-module, for the entropy after the grouping of each slot position calculation basis slot position；
Interest-degree determination sub-module scores to the interest-degree of each slot position for obtaining user；
Weighted sum submodule, for being weighted summation to interest-degree scoring and the entropy；
Answer updates submodule, for updating answer set for the new answer set after user's interaction.
18. according to the system described in any one of claim 10-13, which is characterized in that further include attribute and recall module, the frequency
Statistical module, degree of correlation determining module, sorting module and recommendation problem generation module, wherein：
Attribute recalls module, for returning to module after user returns to answer in answer, recalls the corresponding example of the answer
Or all properties of class；
Frequency statistics module, for counting the frequency of each candidate attribute；
Frequency statistics module, for counting the frequency of each candidate attribute；
Degree of correlation determining module, for obtaining the degree of correlation of each candidate attribute and user's primal problem；
Sorting module, for being ranked up according to the frequency and the degree of correlation to candidate attribute；
Recommendation problem generation module for the candidate attribute after sorting according to sorting module, generates the candidate problem of recommendation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611185779.4A CN108228637B (en) | 2016-12-21 | 2016-12-21 | Automatic response method and system for natural language client |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611185779.4A CN108228637B (en) | 2016-12-21 | 2016-12-21 | Automatic response method and system for natural language client |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108228637A true CN108228637A (en) | 2018-06-29 |
| CN108228637B CN108228637B (en) | 2020-09-04 |
Family
ID=62651726
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611185779.4A Active CN108228637B (en) | 2016-12-21 | 2016-12-21 | Automatic response method and system for natural language client |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108228637B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109753557A (en) * | 2018-12-26 | 2019-05-14 | 出门问问信息科技有限公司 | Answer output method, device, equipment and the storage medium of question answering system |
| CN109815320A (en) * | 2018-12-26 | 2019-05-28 | 出门问问信息科技有限公司 | Answer generation method, device, equipment and the storage medium of question answering system |
| CN110088748A (en) * | 2019-03-19 | 2019-08-02 | 京东方科技集团股份有限公司 | Problem generation method and device, interrogation system, computer readable storage medium |
| CN110737671A (en) * | 2018-07-03 | 2020-01-31 | 百度在线网络技术（北京）有限公司 | Table-based retrieval method and device |
| CN111694932A (en) * | 2019-03-13 | 2020-09-22 | 百度在线网络技术（北京）有限公司 | Conversation method and device |
| CN112337094A (en) * | 2020-10-27 | 2021-02-09 | 四川长虹电器股份有限公司 | Voice game interaction method |
| CN112667895A (en) * | 2020-12-28 | 2021-04-16 | 百果园技术(新加坡)有限公司 | Method, device and equipment for determining recommended item queue and storage medium |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070136246A1 (en) * | 2005-11-30 | 2007-06-14 | At&T Corp. | Answer determination for natural language questioning |
| CN101872349A (en) * | 2009-04-23 | 2010-10-27 | 国际商业机器公司 | Method and device for treating natural language problem |
| CN102236677A (en) * | 2010-04-28 | 2011-11-09 | 北京大学深圳研究生院 | Question answering system-based information matching method and system |
| CN102368260A (en) * | 2011-10-12 | 2012-03-07 | 北京百度网讯科技有限公司 | Method and device of producing domain required template |
| CN103280218A (en) * | 2012-12-31 | 2013-09-04 | 威盛电子股份有限公司 | Selection method based on voice recognition, mobile terminal device and information system thereof |
| CN104216913A (en) * | 2013-06-04 | 2014-12-17 | Sap欧洲公司 | Problem answering frame |
| CN104471568A (en) * | 2012-07-02 | 2015-03-25 | 微软公司 | Learning-Based Processing of Natural Language Problems |
| CN104809197A (en) * | 2015-04-24 | 2015-07-29 | 同程网络科技股份有限公司 | On-line question and answer method based on intelligent robot |
| CN105824933A (en) * | 2016-03-18 | 2016-08-03 | 苏州大学 | Automatic question answering system based on main statement position and implementation method thereof |
| CN106095872A (en) * | 2016-06-07 | 2016-11-09 | 北京高地信息技术有限公司 | Answer sort method and device for Intelligent Answer System |
-
2016
- 2016-12-21 CN CN201611185779.4A patent/CN108228637B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070136246A1 (en) * | 2005-11-30 | 2007-06-14 | At&T Corp. | Answer determination for natural language questioning |
| CN101872349A (en) * | 2009-04-23 | 2010-10-27 | 国际商业机器公司 | Method and device for treating natural language problem |
| CN102236677A (en) * | 2010-04-28 | 2011-11-09 | 北京大学深圳研究生院 | Question answering system-based information matching method and system |
| CN102368260A (en) * | 2011-10-12 | 2012-03-07 | 北京百度网讯科技有限公司 | Method and device of producing domain required template |
| CN104471568A (en) * | 2012-07-02 | 2015-03-25 | 微软公司 | Learning-Based Processing of Natural Language Problems |
| CN103280218A (en) * | 2012-12-31 | 2013-09-04 | 威盛电子股份有限公司 | Selection method based on voice recognition, mobile terminal device and information system thereof |
| CN104216913A (en) * | 2013-06-04 | 2014-12-17 | Sap欧洲公司 | Problem answering frame |
| CN104809197A (en) * | 2015-04-24 | 2015-07-29 | 同程网络科技股份有限公司 | On-line question and answer method based on intelligent robot |
| CN105824933A (en) * | 2016-03-18 | 2016-08-03 | 苏州大学 | Automatic question answering system based on main statement position and implementation method thereof |
| CN106095872A (en) * | 2016-06-07 | 2016-11-09 | 北京高地信息技术有限公司 | Answer sort method and device for Intelligent Answer System |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110737671A (en) * | 2018-07-03 | 2020-01-31 | 百度在线网络技术（北京）有限公司 | Table-based retrieval method and device |
| CN109753557A (en) * | 2018-12-26 | 2019-05-14 | 出门问问信息科技有限公司 | Answer output method, device, equipment and the storage medium of question answering system |
| CN109815320A (en) * | 2018-12-26 | 2019-05-28 | 出门问问信息科技有限公司 | Answer generation method, device, equipment and the storage medium of question answering system |
| CN111694932A (en) * | 2019-03-13 | 2020-09-22 | 百度在线网络技术（北京）有限公司 | Conversation method and device |
| CN110088748A (en) * | 2019-03-19 | 2019-08-02 | 京东方科技集团股份有限公司 | Problem generation method and device, interrogation system, computer readable storage medium |
| US11600389B2 (en) | 2019-03-19 | 2023-03-07 | Boe Technology Group Co., Ltd. | Question generating method and apparatus, inquiring diagnosis system, and computer readable storage medium |
| CN110088748B (en) * | 2019-03-19 | 2023-11-14 | 京东方科技集团股份有限公司 | Question generation method and device, consultation system, computer-readable storage medium |
| CN112337094A (en) * | 2020-10-27 | 2021-02-09 | 四川长虹电器股份有限公司 | Voice game interaction method |
| CN112337094B (en) * | 2020-10-27 | 2022-11-25 | 四川长虹电器股份有限公司 | Voice game interaction method |
| CN112667895A (en) * | 2020-12-28 | 2021-04-16 | 百果园技术(新加坡)有限公司 | Method, device and equipment for determining recommended item queue and storage medium |
| CN112667895B (en) * | 2020-12-28 | 2024-08-13 | 百果园技术(新加坡)有限公司 | Recommended item queue determining method, device, equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108228637B (en) | 2020-09-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108228637A (en) | Natural language client auto-answer method and system | |
| US11645517B2 (en) | Information processing method and terminal, and computer storage medium | |
| CN107818781B (en) | Intelligent interaction method, equipment and storage medium | |
| CN107797984B (en) | Intelligent interaction method, equipment and storage medium | |
| CN111400607B (en) | Search content output method and device, computer equipment and readable storage medium | |
| CN106295807B (en) | A method and device for information processing | |
| CN107766511A (en) | Intelligent answer method, terminal and storage medium | |
| CN109308321A (en) | A knowledge question answering method, knowledge question answering system and computer readable storage medium | |
| WO2020258662A1 (en) | Keyword determination method and apparatus, electronic device, and storage medium | |
| JP2017010517A (en) | Human-machine intelligence chatting method with artificial intelligence and device therefor | |
| CN107278302A (en) | A robot interaction method and an interactive robot | |
| CN110232447B (en) | In-depth reasoning method in legal cases | |
| CN104199810A (en) | Intelligent service method and system based on natural language interaction | |
| CN107239450B (en) | Method for processing natural language based on interactive context | |
| CN108268591A (en) | It is chatted based on active and builds the method and apparatus of user's portrait | |
| US11455501B2 (en) | Response based on hierarchical models | |
| CN114186048A (en) | Question answering method, device, computer equipment and medium based on artificial intelligence | |
| KR20180116104A (en) | Method and apparatus of dialog scenario database constructing for dialog system | |
| CN113076409A (en) | Dialogue system and method applied to robot, robot and readable medium | |
| CN118277588A (en) | Query request processing method, electronic device and storage medium | |
| CN111046151A (en) | Message processing method and device | |
| CN119474419A (en) | A media file query and retrieval method and system based on AI dialogue scenario | |
| CN107665447A (en) | Information processing method and message processing device | |
| CN109885668A (en) | A scalable field human-machine dialogue system state tracking method and device | |
| CN108491506A (en) | Method for pushing problem answers combination |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |