CN106095988A - Automatic question-answering method and device - Google Patents
Automatic question-answering method and device Download PDFInfo
- Publication number
- CN106095988A CN106095988A CN201610452735.7A CN201610452735A CN106095988A CN 106095988 A CN106095988 A CN 106095988A CN 201610452735 A CN201610452735 A CN 201610452735A CN 106095988 A CN106095988 A CN 106095988A
- Authority
- CN
- China
- Prior art keywords
- user
- matching
- answer
- participle
- module
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3322—Query formulation using system suggestions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/284—Lexical analysis, e.g. tokenisation or collocates
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Audiology, Speech & Language Pathology (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Machine Translation (AREA)
Abstract
The invention discloses a kind of automatic question-answering method and device.One of which method includes: obtain the solicited message of user;Solicited message is carried out automatic word segmentation, obtains participle information;Make requests on the matching treatment of information and knowledge point based on participle information, when the match is successful, send answer to user;Otherwise, according to user operation, participle information is adjusted, makes requests on the matching treatment of information and knowledge point based on the participle information after adjusting, when the match is successful, send answer to user;Otherwise, repeat above-mentioned adjustment and matching treatment, until the match is successful or reaches predetermined number of repetition.Technical scheme by means of the embodiment of the present invention, it is possible to increase intelligent robot semantics recognition and the accuracy rate of coupling, promotes Consumer's Experience, meets personalized use demand, preferably improves intelligent robot service quality and level.
Description
Technical field
The present invention relates to technical field of information processing, particularly relate to a kind of automatic question-answering method and device.
Background technology
In the prior art, including multiple knowledge point in intelligent answer knowledge base, each described knowledge point includes problem
And answer.Can mate according to the solicited message (i.e. problem) that user is inputted by intelligent answer knowledge base, mating
Time, prior art uses Intelligent Matching and two kinds of matching ways of fuzzy matching, specifically, after the solicited message input of user,
Engine can carry out participle according to the question sentence of user, after having carried out participle, needs to calculate this solicited message and intelligence according to part of speech
The semantic similarity of problem in question and answer knowledge base, if semantic similarity is more than or equal to the threshold value pre-set, then takes out
As semantic matches, i.e. Intelligent Matching;If being not reaching to the threshold value pre-set, then carry out Keywords matching, i.e. fuzzy matching.But
It is that in the prior art, the matching result after fuzzy matching is not likely to be very accurate, and matching result cannot be compiled by user
Collecting amendment, immediate system identification matching content can only be selected as self-defined question sentence, therefore, employing aforesaid way can not
Being well understood by the problem that user proposes, recognition accuracy is low, it is impossible to meet the requirement of user.
Summary of the invention
In view of the above problems, it is proposed that the present invention in case provide one overcome the problems referred to above or at least in part solve on
State automatic question-answering method and the device of problem.
The present invention provides a kind of automatic question-answering method, including:
Thering is provided question and answer knowledge base, question and answer knowledge base includes that multiple knowledge point, each knowledge point include problem and answer;
Obtain the solicited message of user;
Solicited message is carried out automatic word segmentation, obtains participle information;
Make requests on the matching treatment of information and knowledge point based on participle information, when the match is successful, send to user and answer
Case;Otherwise, according to user operation, participle information is adjusted, makes requests on information and knowledge based on the participle information after adjusting
The matching treatment of point, when the match is successful, sends answer to user;Otherwise, repeat above-mentioned adjustment and matching treatment, until
The match is successful or reaches predetermined number of repetition.
Present invention also offers a kind of automatic question-answering method, including:
Thering is provided question and answer knowledge base, question and answer knowledge base includes that multiple knowledge point, each knowledge point include problem and answer;
Obtain the solicited message of user;
Solicited message is carried out automatic word segmentation, obtains and show participle information;
According to user operation, participle information is adjusted;
The matching treatment of information and knowledge point, when the match is successful, Xiang Yong is made requests on based on the participle information after adjusting
Family sends answer;Otherwise, repeat above-mentioned display, adjustment and matching treatment, until the match is successful or reaches predetermined repetition
Number of times.
Present invention also offers a kind of automatic call answering arrangement, including: acquisition module, input module, word-dividing mode, coupling mould
Block, adjusting module, output module and control module, wherein:
Acquisition module, is used for obtaining multiple knowledge point, and each knowledge point includes problem and answer;
Input module, for obtaining the solicited message of user;
Word-dividing mode, for solicited message is carried out automatic word segmentation, obtains participle information;
Matching module, for making requests on mating of information and knowledge point based on participle information or the participle information after adjusting
Process;
Adjusting module, is used for when making requests on the matching treatment failure of information and knowledge point based on participle information, according to
Participle information is adjusted by user operation;
Output module, for when the match is successful, sends answer to user;
Control module, in the case of the match is successful at matching module, calls output module, loses in matching module coupling
In the case of losing, call adjusting module and matching module, until the match is successful or reaches predetermined number of repetition.
Present invention also offers a kind of automatic call answering arrangement, including: acquisition module, input module, word-dividing mode, display mould
Block, adjusting module, matching module, output module and control module, wherein:
Acquisition module, is used for obtaining multiple knowledge point, and each knowledge point includes problem and answer;
Input module, for obtaining the solicited message of user;
Word-dividing mode, for solicited message is carried out automatic word segmentation, obtains participle information;
Display module, is used for after word-dividing mode obtains participle information, and participle is believed by adjusting module according to user operation
Before breath is adjusted, show participle information;
Adjusting module, for being adjusted participle information according to user operation;
Matching module, for making requests on the matching treatment of information and knowledge point based on the participle information after adjusting;
Output module, for when the match is successful, sends answer to user;
Control module, in the case of the match is successful at described matching module, calls described output module, at described
In the case of joining module it fails to match, call described display module, described adjusting module and described matching module, until
It is made into merit or reaches predetermined number of repetition.
The present invention has the beneficial effect that:
By allowing user that participle is carried out edit-modify, re-start at the coupling of knowledge point further according to amended participle
Reason, thus it is met the answer of demand, solve and prior art uses Intelligent Matching and fuzzy matching can not well manage
Solving the problem that user proposes, recognition accuracy is low, it is impossible to the problem meeting the requirement of user, by means of the skill of the embodiment of the present invention
Art scheme, it is possible to increase intelligent robot semantics recognition and the accuracy rate of coupling, promotes Consumer's Experience, meets personalization and uses need
Ask, preferably improve intelligent robot service quality and level.
Described above is only the general introduction of technical solution of the present invention, in order to better understand the technological means of the present invention,
And can be practiced according to the content of description, and in order to allow above and other objects of the present invention, the feature and advantage can
Become apparent, below especially exemplified by the detailed description of the invention of the present invention.
Accompanying drawing explanation
By reading the detailed description of hereafter preferred implementation, various other advantage and benefit common for this area
Technical staff will be clear from understanding.Accompanying drawing is only used for illustrating the purpose of preferred implementation, and is not considered as the present invention
Restriction.And in whole accompanying drawing, it is denoted by the same reference numerals identical parts.In the accompanying drawings:
Fig. 1 is the flow chart of the automatic question-answering method of the example 1 of the inventive method embodiment one;
Fig. 2 is the flow chart processed in detail of the automatic question-answering method of the inventive method embodiment one;
Fig. 3 is the schematic diagram of the problem obtaining user's input of the inventive method embodiment;
Fig. 4 is that the question sentence according to user's input of the inventive method embodiment carries out the schematic diagram of Intelligent Matching;
Fig. 5 is the schematic diagram of the display participle information of the inventive method embodiment;
Fig. 6 is the schematic diagram that the user of the inventive method embodiment revises participle according to demand;
Fig. 7 be the inventive method embodiment amendment participle after carry out the schematic diagram of Intelligent Matching;
Fig. 8 is the schematic diagram that the user of the inventive method embodiment finally gives question sentence and the answer meeting demand;
Fig. 9 is the flow chart of the automatic question-answering method of the example 1 of the inventive method embodiment two;
Figure 10 is the flow chart of the automatic question-answering method of the example 2 of the inventive method embodiment one;
Figure 11 is the flow chart of example 2 automatic question-answering method of the inventive method embodiment two;
Figure 12 is the structural representation of the automatic call answering arrangement of apparatus of the present invention embodiment one;
Figure 13 be the automatic call answering arrangement of apparatus of the present invention embodiment two structural representation.
Detailed description of the invention
It is more fully described the exemplary embodiment of the disclosure below with reference to accompanying drawings.Although accompanying drawing shows the disclosure
Exemplary embodiment, it being understood, however, that may be realized in various forms the disclosure and should be by embodiments set forth here
Limited.On the contrary, it is provided that these embodiments are able to be best understood from the disclosure, and can be by the scope of the present disclosure
Complete conveys to those skilled in the art.
In order to solve prior art to use Intelligent Matching and fuzzy matching can not be well understood by the problem that user proposes,
Recognition accuracy is low, it is impossible to meet the problem that user requires, the invention provides a kind of automatic question-answering method and device, works as intelligence
Coupling and fuzzy matching be when all can not meet user's request, and user can carry out edit-modify, and root according to practical situation to participle
The matching treatment of knowledge point is re-started, the semantic recommendation results after being updated according to amended participle.
Below in conjunction with accompanying drawing and embodiment, the present invention is further elaborated.Should be appreciated that described herein
Specific embodiment only in order to explain the present invention, do not limit the present invention.
Embodiment of the method one
Example 1
According to embodiments of the invention, it is provided that a kind of automatic question-answering method, Fig. 1 is the reality of the inventive method embodiment one
The flow chart of the automatic question-answering method of example 1, as it is shown in figure 1, automatic question-answering method according to embodiments of the present invention includes locating as follows
Reason:
Step 101, it is provided that question and answer knowledge base, question and answer knowledge base include multiple knowledge point, each knowledge point include problem and
Answer.
In question answering system, knowledge base plays vital effect, and knowledge base includes multiple knowledge point, each knowledge point
Including question sentence and answer, described question sentence can include that a standard is asked.Specifically, the knowledge point in knowledge base is the most original and the simplest
Single form is exactly the FAQ commonly used at ordinary times, and general form is that " ask-answer " is right, wherein, should " asking " be exactly that standard is asked, and be somebody's turn to do " answering "
It it is exactly answer.Additionally, each knowledge point in knowledge base can also have the one or more extensions asked corresponding to a standard to ask,
This extension is asked and is asked expression-form slightly difference with standard, but the implication expressed is identical.
In knowledge base, standard is asked and extends that ask both can be to have used general problem form, it would however also be possible to employ semantic meaning representation
Formula form, it is same as the prior art, does not repeats them here.
Step 102, obtains the solicited message of user.
In embodiments of the present invention, this solicited message can be obtained by the various ways such as speech recognition, Text region.?
In the embodiment of the present invention, the solicited message of user generally user wants to obtain the problem information corresponding to certain answer.Such as,
The solicited message of user may is that how to arrange CRBT, how to apply for broadband etc..
Step 103, carries out automatic word segmentation to solicited message, obtains participle information.
In embodiments of the present invention, when carrying out automatic word segmentation, it is usually and carries out according to dictionary for word segmentation.In actual applications,
Automatic word segmentation process can use the one in the two-way maximum matching method of dictionary, viterbi method, HMM method and CRF method or
Multiple.
Such as, solicited message is: how to arrange CRBT, then participle information is: how, arrange, CRBT.
Step 104, carries out the matching treatment of described solicited message and described knowledge point, works as coupling based on described participle information
During success, send answer to user;Otherwise, show described participle information, according to user operation, described participle information is adjusted
Whole, the matching treatment of described solicited message and described knowledge point is carried out based on the participle information after adjusting, when the match is successful, to
User sends answer;Otherwise, repeat above-mentioned adjustment and matching treatment, until the match is successful or reaches predetermined repetition time
Number.
At step 104, matching treatment can be to use any one mode following:
Mode one: carry out Intelligent Matching by Semantic Similarity Measurement, when there is semantic similarity more than threshold value, coupling
Success;
Wherein, after pass-through mode one carries out Intelligent Matching success, described answer is sent in the following ways to user: will be
The answer of big semantic similarity correspondence problem is transmitted directly to user;
Mode two: carry out fuzzy matching by keyword search, when the problem having user to select in matching result, coupling
Success;
Mode three: first pass through Semantic Similarity Measurement and carry out Intelligent Matching, when there is semantic similarity less than described threshold value
Time, then carry out fuzzy matching by keyword search, when the problem having user to select in matching result, the match is successful.
Wherein, after pass-through mode two and mode three carry out fuzzy matching success, by successfully whole for Keywords matching or portion
Point problem is sent to user, and according to the selection of user, sends, to user, the answer that selected problem is corresponding.
During it should be noted that be equal to threshold value for semantic similarity, both it is believed that the match is successful, it is also possible to recognize
Unsuccessful for coupling, when matching result difference, follow-up carry out different process, it is not intended to protection scope of the present invention.
In Intelligent Matching, Semantic Similarity Measurement is a step crucial in question answering system, and its result directly influences whole
The accuracy of individual system.The object participating in Semantic Similarity Measurement must be reciprocity, for calculation knowledge point and user please
Seeking the semantic similarity of information (i.e. user's question sentence), defining this semantic similarity is user's question sentence and all question sentences in knowledge point
The maximum semantic similarity of (including that all standards are asked, extended and ask about template).In knowledge point, question sentence is calculating language with user's question sentence
Can be calculated by the co sinus vector included angle formula in tradition vector space model during justice similarity.
It should be noted that in embodiments of the present invention, confirm it is to use Intelligent Matching or employing in the following way
Fuzzy matching, specifically, after the input of the solicited message of user, engine can carry out participle according to the question sentence of user, is carrying out point
After word, need to calculate this solicited message and the semantic similarity of problem in intelligent answer knowledge base according to part of speech, if semantic similitude
Degree more than or equal to the threshold value pre-set, then carries out abstract semantics coupling, i.e. Intelligent Matching;If being not reaching to pre-set
Threshold value, then carry out Keywords matching, i.e. fuzzy matching.
In other embodiments of the invention, it is also possible to only with Intelligent Matching or fuzzy matching make requests on information with
The matching treatment of knowledge point, it is not intended to protection scope of the present invention.
Illustrate based on the semantic similarity between participle information computation requests information and problem in step 104 below
Explanation.
In embodiments of the present invention, it is assumed that user's question sentence is " CRBT is the most open-minded ", object knowledge point is " opening CRBT ".
Object knowledge point contains following extension asks about semantic template: " [how] [open-minded] [CRBT] ", " [open-minded] method of [CRBT] has
Which ", " [open-minded] [mode] of [introduction] [CRBT] ", " [CRBT] is [how] [open-minded] ", " you know how to open-minded
CRBT ", " I wants CRBT ".
First quantifying all of similar question sentence and semantic template in user's question sentence and destination document is n-dimensional vector.Quantifying
During, template part of speech is classified as same Feature Words with the word in the user's question sentence belonging to this part of speech, the narrow sense part of speech of synonym and
The word comprised has identical weights also can be classified as same Feature Words.In above-mentioned example, Feature Words has: { CRBT, [CRBT] }, { opens
Logical, [open-minded] }, { [how], how }, { [mode], method }, { which has }, { you }, { knowing }, { I }, { wanting }.Cause
All sentences in example and template can be converted into 9 dimensional vectors by this.The most successively user's question sentence vector is asked with standard, expanded
Exhibition is asked about vector corresponding to semantic template and is carried out the Semantic Similarity Measurement of co sinus vector included angle, obtains concrete Similarity value.
Then to these semantic similitude angle value maximizings, the maximum in this example is user's question sentence with " [how] [open-minded] is [color
Bell] " semantic similitude angle value.Can be to reduce computing in the way of using computing limit, limit record local maximum when maximizing
Amount.Finally using this maximum of trying to achieve as the Similarity value of user's question sentence Yu object knowledge point.
Additionally, when making requests on information and knowledge point based on participle information it fails to match, illustrate to believe based on this participle
Breath can not match suitable answer, and in these cases, the embodiment of the present invention can show participle information, and provides operation
Interface revises this participle information for user, is adjusted participle information according to user operation.In embodiments of the present invention, user
Being adjusted participle information can be to include at least one following several form: increases new participle information, delete former participle letter
Cease, replace former participle information.Such as, former participle information is: how, arrange, CRBT, amended participle information is: how to use
Mobile phone arranges CRBT, in this example, adds " with mobile phone " this new participle information.And for example, in another example, because of
Neologisms None-identified, causes participle mistake, identical solicited message can be provided different word segmentation result now by adjusting.
During the amendment of participle information or after the participle information that has been adjusted, can based on after adjusting point
Word information carries out the matching treatment of described solicited message and described knowledge point, if the match is successful, then pushes answer to user, no
Then, repeat above-mentioned display, adjustment and matching treatment, until the match is successful, or, reach predetermined number of repetition, or,
Reach other end condition.Wherein, above-mentioned display operation refers to: show currently used participle information;Above-mentioned adjustment operates
Refer to: currently used participle information is modified;Above-mentioned matching treatment refers to: carry out institute based on the participle information after adjusting
State the matching treatment of solicited message and described knowledge point.
In embodiments of the present invention, after participle information being adjusted according to user operation, it is also possible to after adjusting
Participle information updating dictionary for word segmentation, wherein, dictionary for word segmentation is for carrying out participle to subsequent request information.By above-mentioned process,
Follow-up word segmentation processing can be made more accurate, and extend one's service during using the range of choice to key word.
It can be seen from the above description that in embodiments of the present invention, the problem that intelligent robot system inputs for user
Corresponding knowledge coupling can be carried out, three kinds of matching ways can be included: Intelligent Matching, fuzzy matching and participle are revised (again
Semantic recommendation).Preferably, the order performing above-mentioned three kinds of matching ways is: Intelligent Matching, fuzzy matching, participle amendment, specifically
As follows:
1, Intelligent Matching: i.e. semantic recommendation, first passes through the abstract semantics matched and is identified, if the match is successful,
Then push corresponding matching result to user, perform Intelligent Matching and the match is successful shows that matching result fully meets user and needs
Asking, if mating unsuccessful or user to be unsatisfied with matching result, then carrying out fuzzy matching.
2, fuzzy matching: i.e. show participle, according to participle information, carries out Keywords matching, if the match is successful, then to
Family pushes corresponding matching result, performs fuzzy matching and the match is successful shows that matching result meets user's need to a certain extent
Asking, but there is the error of a part, if mating unsuccessful or user to be unsatisfied with matching result, then revising participle.
3, amendment participle (semantic recommendation again): i.e. user carries out edit-modify to participle, shows " Intelligent Matching " and " mould
Stick with paste coupling " result all cannot meet user's request and there is the biggest gap therewith, it is necessary to by system after manual amendment again
Semantic recommendation just can obtain the self-defined question sentence that user needs.Wherein, the process of participle amendment includes: split into by question sentence some
Individual participle, after user adjusts participle, first contrasts with original question sentence, adjusted for user word is entered into dictionary for word segmentation, the most again
Re-start semantic recommendation and/or fuzzy matching, until obtaining customer satisfaction system matching result or reaching predetermined repetition time
Number.
Below in conjunction with accompanying drawing, the technique scheme of the embodiment of the present invention is described in detail.
The technique scheme of the embodiment of the present invention may be used for iBot Cloud intelligent cloud service platform, below for
It is illustrated as a example by iBot Cloud intelligent cloud service platform.Fig. 2 is the automatic question answering side of the inventive method embodiment one
The flow chart processed in detail of method, as in figure 2 it is shown, specifically include following process:
Step 201, obtains the problem (corresponding to above-mentioned solicited message) of user's input;As it is shown on figure 3, obtain user's input
Problem: " how can arriving Disneyland ".
Step 202, carries out automatic word segmentation to the problem of user's input.
Step 203, calculates problem and the semantic similarity of problem in knowledge point of user's input based on participle information, if
This semantic similarity is more than threshold value, then perform step 204, otherwise, perform step 205.
Step 204, the answer to major general's maximum semantic similarity correspondence problem is sent to user, or, will be greater than threshold value
The part or all of problem corresponding to semantic similarity be shown to user, and according to the selection of user, push selected to user
The answer that problem is corresponding, end operation, if user is unsatisfied with for the answer pushed or problem, then perform step 205.
In embodiments of the present invention, can be initially with Intelligent Matching mode, as shown in Figure 4, according to asking of user's input
Sentence carries out semantics recognition, and the result obtaining Intelligent Matching is: " how to arrive Disneyland?", " Disneyland arrive
Method?", the two matching result can be shown to user, select for user.If as it is shown in figure 5, user is to Intelligent Matching
Question sentence when being unsatisfied with, then carry out fuzzy matching, the result of fuzzy matching can be obtained.
Step 205, it may be judged whether reach the predetermined number of times, if the judgment is Yes, end operation of adjusting, otherwise, perform step
206。
Step 206, displays to the user that this participle information.
Step 207, is adjusted participle information according to user operation, obtains the participle information after adjusting, and performs step
203。
As shown in Figure 6, user revises participle according to demand, obtains " the most whether can arrive Disneyland ".As
Shown in Fig. 7, in embodiments of the present invention, later participle can be revised for user and re-start Semantic Similarity Measurement, intelligence
Match question sentence: " Disneyland can be arrived by the subway?" result is identical with the question sentence of user's request, identification accurate
Rate is obviously improved.As shown in Figure 8, user finally gives question sentence and the answer meeting demand.
The technical scheme of the embodiment of the present invention allows user to carry out the result of system matches participating in getting involved.When system is carried out
" Intelligent Matching " and " fuzzy matching " when all can not meet the demand of user, can be according to user to the participle information in coupling
Amendment, i.e. obtain the semantic recommendation results after updating by editor's key word, thus be met the self-defined question sentence of demand.
Technical scheme by means of the embodiment of the present invention, it is possible to increase intelligent robot semantics recognition and the accuracy rate of coupling, promotes and uses
Family is experienced, and meets personalized use demand, preferably improves intelligent robot service quality and level.
Example 2
According to embodiments of the invention, it is provided that a kind of automatic question-answering method, Figure 10 is the inventive method embodiment one
The flow chart of the automatic question-answering method of example 2, as shown in Figure 10, automatic question-answering method according to embodiments of the present invention includes as follows
Process:
Step 1001, it is provided that question and answer knowledge base, question and answer knowledge base include multiple knowledge point, each knowledge point include problem and
Answer.
In question answering system, knowledge base plays vital effect, and knowledge base includes multiple knowledge point, each knowledge point
Including question sentence and answer, described question sentence can include that a standard is asked.Specifically, the knowledge point in knowledge base is the most original and the simplest
Single form is exactly the FAQ commonly used at ordinary times, and general form is that " ask-answer " is right, wherein, should " asking " be exactly that standard is asked, and be somebody's turn to do " answering "
It it is exactly answer.Additionally, each knowledge point in knowledge base can also have the one or more extensions asked corresponding to a standard to ask,
This extension is asked and is asked expression-form slightly difference with standard, but the implication expressed is identical.
In knowledge base, standard is asked and extends that ask both can be to have used general problem form, it would however also be possible to employ semantic meaning representation
Formula form, it is same as the prior art, does not repeats them here.
Step 1002, obtains the solicited message of user.
In embodiments of the present invention, this solicited message can be obtained by the various ways such as speech recognition, Text region.?
In the embodiment of the present invention, the solicited message of user generally user wants to obtain the problem information corresponding to certain answer.Such as,
The solicited message of user may is that how to arrange CRBT, how to apply for broadband etc..
Step 1003, carries out automatic word segmentation to solicited message, obtains and shows participle information.
In embodiments of the present invention, when carrying out automatic word segmentation, it is usually and carries out according to dictionary for word segmentation.In actual applications,
Automatic word segmentation process can use the one in the two-way maximum matching method of dictionary, viterbi method, HMM method and CRF method or
Multiple.
Additionally, unlike embodiment of the method one, in embodiments of the present invention, after carrying out participle operation, directly will
Participle information is shown to user.
Step 1004, carries out the matching treatment of described solicited message and described knowledge point based on described participle information, when
When being made into merit, send answer to user;Otherwise, according to user operation, described participle information is adjusted, after adjusting
Participle information carries out the matching treatment of described solicited message and described knowledge point, when the match is successful, sends answer to user;No
Then, repeat above-mentioned display, adjustment and matching treatment, until the match is successful or reaches predetermined number of repetition.
In step 1004, matching treatment can be to use any one mode following:
Mode one: carry out Intelligent Matching by Semantic Similarity Measurement, when there is semantic similarity more than threshold value, coupling
Success;
Wherein, after pass-through mode one carries out Intelligent Matching success, described answer is sent in the following ways to user: by language
Justice similarity is sent to user more than all or part of problem of threshold value, and according to the selection of user, to user send selected by ask
The answer that topic is corresponding;
Mode two: carry out fuzzy matching by keyword search, when the problem having user to select in matching result, coupling
Success;
Mode three: first pass through Semantic Similarity Measurement and carry out Intelligent Matching, when there is semantic similarity less than described threshold value
Time, then carry out fuzzy matching by keyword search, when the problem having user to select in matching result, the match is successful.
Wherein, after pass-through mode two and mode three carry out fuzzy matching success, by successfully whole for Keywords matching or portion
Point problem is sent to user, and according to the selection of user, sends, to user, the answer that selected problem is corresponding.
In Intelligent Matching, Semantic Similarity Measurement is a step crucial in question answering system, and its result directly influences whole
The accuracy of individual system.The object participating in Semantic Similarity Measurement must be reciprocity, for calculation knowledge point and user please
Seeking the semantic similarity of information (i.e. user's question sentence), defining this semantic similarity is user's question sentence and all question sentences in knowledge point
The maximum semantic similarity of (including that all standards are asked, extended and ask about template).In knowledge point, question sentence is calculating language with user's question sentence
Can be calculated by the co sinus vector included angle formula in tradition vector space model during justice similarity.
It should be noted that confirming in the following way is to use Intelligent Matching or use fuzzy matching, specifically, use
After the solicited message input at family, engine can carry out participle according to the question sentence of user, after having carried out participle, needs according to part of speech meter
Calculate this solicited message and the semantic similarity of problem in intelligent answer knowledge base, if semantic similarity is more than or equal to setting in advance
The value put, then carry out abstract semantics coupling, i.e. Intelligent Matching;If being not reaching to the value pre-set, then carry out Keywords matching,
I.e. fuzzy matching.
In other embodiments of the invention, it is also possible to only with Intelligent Matching or fuzzy matching make requests on information with
The matching treatment of knowledge point, it is not intended to protection scope of the present invention.
Illustrate based on the semantic similarity between participle information computation requests information and problem in step 104 below
Explanation.
In embodiments of the present invention, it is assumed that user's question sentence is " CRBT is the most open-minded ", object knowledge point is " opening CRBT ".
Object knowledge point contains following extension asks about semantic template: " [how] [open-minded] [CRBT] ", " [open-minded] method of [CRBT] has
Which ", " [open-minded] [mode] of [introduction] [CRBT] ", " [CRBT] is [how] [open-minded] ", " you know how to open-minded
CRBT ", " I wants CRBT ".
First quantifying all of similar question sentence and semantic template in user's question sentence and destination document is n-dimensional vector.Quantifying
During, template part of speech is classified as same Feature Words with the word in the user's question sentence belonging to this part of speech, the narrow sense part of speech of synonym and
The word comprised has identical weights also can be classified as same Feature Words.In above-mentioned example, Feature Words has: { CRBT, [CRBT] }, { opens
Logical, [open-minded] }, { [how], how }, { [mode], method }, { which has }, { you }, { knowing }, { I }, { wanting }.Cause
All sentences in example and template can be converted into 9 dimensional vectors by this.The most successively user's question sentence vector is asked with standard, expanded
Exhibition is asked about vector corresponding to semantic template and is carried out the Semantic Similarity Measurement of co sinus vector included angle, obtains concrete Similarity value.
Then to these semantic similitude angle value maximizings, the maximum in this example is user's question sentence with " [how] [open-minded] is [color
Bell] " semantic similitude angle value.Can be to reduce computing in the way of using computing limit, limit record local maximum when maximizing
Amount.Finally using this maximum of trying to achieve as the Similarity value of user's question sentence Yu object knowledge point.
During the amendment of participle information or after the participle information that has been adjusted, can based on after adjusting point
Word information carries out the matching treatment of described solicited message and described knowledge point, if the match is successful, then pushes answer to user, no
Then, repeat above-mentioned display, adjustment and matching treatment, until the match is successful, or, reach predetermined number of repetition, or,
Reach other end condition.Wherein, above-mentioned display operation refers to: show currently used participle information;Above-mentioned adjustment operates
Refer to: currently used participle information is modified;Above-mentioned matching treatment refers to: carry out institute based on the participle information after adjusting
State the matching treatment of solicited message and described knowledge point.
The technique scheme of the embodiment of the present invention may be used for iBot Cloud intelligent cloud service platform.
The technical scheme of the embodiment of the present invention allows user to carry out the result of system identification participating in getting involved.When user does not has
Selection system push coupling after problem time, can according to user's amendment to participle information, i.e. by editor key word obtain
Semantic recommendation results after must updating, thus it is met the self-defined question sentence of demand.Technology by means of the embodiment of the present invention
Scheme, it is possible to increase intelligent robot semantics recognition and the accuracy rate of coupling, promotes Consumer's Experience, meets personalization and uses need
Ask, preferably improve intelligent robot service quality and level.
Embodiment of the method two
Example 1
According to embodiments of the invention, it is provided that a kind of automatic question-answering method, Fig. 9 is the reality of the inventive method embodiment two
The flow chart of the automatic question-answering method of example 1, as it is shown in figure 9, automatic question-answering method according to embodiments of the present invention includes locating as follows
Reason:
Step 901, it is provided that question and answer knowledge base, question and answer knowledge base include multiple knowledge point, each knowledge point include problem and
Answer.
In question answering system, knowledge base plays vital effect, and knowledge base includes multiple knowledge point, each knowledge point
Including question sentence and answer, described question sentence can include that a standard is asked.Specifically, the knowledge point in knowledge base is the most original and the simplest
Single form is exactly the FAQ commonly used at ordinary times, and general form is that " ask-answer " is right, wherein, should " asking " be exactly that standard is asked, and be somebody's turn to do " answering "
It it is exactly answer.Additionally, each knowledge point in knowledge base can also have the one or more extensions asked corresponding to a standard to ask,
This extension is asked and is asked expression-form slightly difference with standard, but the implication expressed is identical.
In knowledge base, standard is asked and extends that ask both can be to have used general problem form, it would however also be possible to employ semantic meaning representation
Formula form, it is same as the prior art, does not repeats them here.
Step 902, obtains the solicited message of user.
In embodiments of the present invention, this solicited message can be obtained by the various ways such as speech recognition, Text region.?
In the embodiment of the present invention, the solicited message of user generally user wants to obtain the problem information corresponding to certain answer.Such as,
The solicited message of user may is that how to arrange CRBT, how to apply for broadband etc..
Step 903, carries out automatic word segmentation to solicited message, obtains and shows participle information.
In embodiments of the present invention, when carrying out automatic word segmentation, it is usually and carries out according to dictionary for word segmentation.In actual applications,
Automatic word segmentation process can use the one in the two-way maximum matching method of dictionary, viterbi method, HMM method and CRF method or
Multiple.
Additionally, unlike the example 1 of embodiment of the method one, in embodiments of the present invention, after carrying out participle operation,
Directly participle information is shown to user.
Step 904, is adjusted participle information according to user operation.In embodiments of the present invention, participle is believed by user
Be adjusted can be to include at least one following several form for breath: increase new participle information, delete former participle information, replace former
Participle information.
Unlike embodiment of the method one, before carrying out matching treatment, the embodiment of the present invention is directly according to user operation
Participle information is adjusted.
If it is to say, user needs to revise participle information, then can operating according to the adjustment of user and this participle believed
Breath is adjusted.
In embodiments of the present invention, after participle information being adjusted according to user operation, it is also possible to after adjusting
Participle information updating dictionary for word segmentation, wherein, dictionary for word segmentation is for carrying out participle to subsequent request information.By above-mentioned process, can
So that follow-up word segmentation processing is more accurate, extend one's service during using the range of choice to key word.
Step 905, carries out the matching treatment of described solicited message and described knowledge point based on the participle information after adjusting, when
When the match is successful, directly send answer to user;Otherwise, repeat above-mentioned display, adjustment and matching treatment, until mating into
Merit or reach predetermined number of repetition.
In step 905, matching treatment can be to use any one mode following:
Mode one: carry out Intelligent Matching by Semantic Similarity Measurement, when there is semantic similarity more than threshold value, coupling
Success;
Wherein, after pass-through mode one carries out Intelligent Matching success, described answer is sent in the following ways to user: will be
The answer of big semantic similarity correspondence problem is transmitted directly to user;
Mode two: carry out fuzzy matching by keyword search, when the problem having user to select in matching result, coupling
Success;
Mode three: first pass through Semantic Similarity Measurement and carry out Intelligent Matching, when there is semantic similarity less than described threshold value
Time, then carry out fuzzy matching by keyword search, when the problem having user to select in matching result, the match is successful.
Wherein, after pass-through mode two and mode three carry out fuzzy matching success, by successfully whole for Keywords matching or portion
Point problem is sent to user, and according to the selection of user, sends, to user, the answer that selected problem is corresponding.
In Intelligent Matching, Semantic Similarity Measurement is a step crucial in question answering system, and its result directly influences whole
The accuracy of individual system.The object participating in Semantic Similarity Measurement must be reciprocity, for calculation knowledge point and user please
Seeking the semantic similarity of information (i.e. user's question sentence), defining this semantic similarity is user's question sentence and all question sentences in knowledge point
The maximum semantic similarity of (including that all standards are asked, extended and ask about template).In knowledge point, question sentence is calculating language with user's question sentence
Can be calculated by the co sinus vector included angle formula in tradition vector space model during justice similarity.
It should be noted that confirming in the following way is to use Intelligent Matching or use fuzzy matching, specifically, use
After the solicited message input at family, engine can carry out participle according to the question sentence of user, after having carried out participle, needs according to part of speech meter
Calculate this solicited message and the semantic similarity of problem in intelligent answer knowledge base, if semantic similarity is more than or equal to setting in advance
The value put, then carry out abstract semantics coupling, i.e. Intelligent Matching;If being not reaching to the value pre-set, then carry out Keywords matching,
I.e. fuzzy matching.
It can be seen from the above description that in embodiments of the present invention, the problem that intelligent robot system inputs for user
Corresponding knowledge coupling can be carried out, can include three kinds of matching ways: participle amendment (again semantic recommend), Intelligent Matching, with
And fuzzy matching.Preferably, the order performing above-mentioned three kinds of matching ways is: participle amendment, Intelligent Matching, fuzzy matching, tool
Body is as follows:
1, amendment participle (semantic recommendation again): i.e. user carries out edit-modify to participle, and showing must be by manually repairing
Change rear system semantic recommendation again and just can obtain the self-defined question sentence that user needs.Wherein, the process of participle amendment includes: will ask
Sentence splits into several participles, after user adjusts participle, first contrasts with original question sentence, adjusted for user word is entered into participle
Dictionary, carries out semantic recommendation the most again.
2, Intelligent Matching: i.e. semantic recommendation, first passes through the abstract semantics matched and is identified, if the match is successful,
Then push corresponding matching result to user, perform Intelligent Matching and the match is successful shows that matching result fully meets user and needs
Asking, if mating unsuccessful or user to be unsatisfied with matching result, then carrying out fuzzy matching.
3, fuzzy matching: i.e. show participle, according to participle information, carries out Keywords matching, if the match is successful, then to
Family pushes corresponding matching result, performs fuzzy matching and the match is successful shows that matching result meets user's need to a certain extent
Asking, but there is the error of a part, if mating unsuccessful or user to be unsatisfied with matching result, then remodifying participle, until
Obtain customer satisfaction system matching result or reach predetermined number of repetition.
Put down it should be noted that the technique scheme of the embodiment of the present invention may be used for the service of iBot Cloud intelligent cloud
Platform.
The technical scheme of the embodiment of the present invention allows user to carry out the result of system identification participating in getting involved.When system is carried out
After participle, directly participle information is shown to user and modifies for user, i.e. obtain the language after updating by editor's key word
Justice recommendation results.Technical scheme by means of the embodiment of the present invention, it is possible to increase intelligent robot semantics recognition and the standard of coupling
Really rate, promotes Consumer's Experience, meets personalized use demand, preferably improves intelligent robot service quality and level.
Example 2
According to embodiments of the invention, it is provided that a kind of automatic question-answering method, Figure 11 is the inventive method embodiment two
The flow chart of example 2 automatic question-answering method, as shown in figure 11, automatic question-answering method according to embodiments of the present invention includes locating as follows
Reason:
Step 1101, it is provided that question and answer knowledge base, question and answer knowledge base include multiple knowledge point, each knowledge point include problem and
Answer;
In question answering system, knowledge base plays vital effect, and knowledge base includes multiple knowledge point, each knowledge point
Including question sentence and answer, described question sentence can include that a standard is asked.Specifically, the knowledge point in knowledge base is the most original and the simplest
Single form is exactly the FAQ commonly used at ordinary times, and general form is that " ask-answer " is right, wherein, should " asking " be exactly that standard is asked, and be somebody's turn to do " answering "
It it is exactly answer.Additionally, each knowledge point in knowledge base can also have the one or more extensions asked corresponding to a standard to ask,
This extension is asked and is asked expression-form slightly difference with standard, but the implication expressed is identical.
In knowledge base, standard is asked and extends that ask both can be to have used general problem form, it would however also be possible to employ semantic meaning representation
Formula form, it is same as the prior art, does not repeats them here.
Step 1102, obtains the solicited message of user;
In embodiments of the present invention, this solicited message can be obtained by the various ways such as speech recognition, Text region.?
In the embodiment of the present invention, the solicited message of user generally user wants to obtain the problem information corresponding to certain answer.Such as,
The solicited message of user may is that how to arrange CRBT, how to apply for broadband etc..
Step 1103, carries out automatic word segmentation to solicited message, obtains and shows participle information;
In embodiments of the present invention, when carrying out automatic word segmentation, it is usually and carries out according to dictionary for word segmentation.In actual applications,
Automatic word segmentation process can use the one in the two-way maximum matching method of dictionary, viterbi method, HMM method and CRF method or
Multiple.
Additionally, unlike the example 1 of embodiment of the method one, in embodiments of the present invention, after carrying out participle operation,
Directly participle information is shown to user.
Step 1104, is adjusted participle information according to user operation;
If it is to say, user needs to revise participle information, then can operating according to the adjustment of user and this participle believed
Breath is adjusted, if the user thinks that need not revise participle information, then keeps participle Information invariability.
In embodiments of the present invention, after participle information being adjusted according to user operation, it is also possible to after adjusting
Participle information updating dictionary for word segmentation, wherein, dictionary for word segmentation is for carrying out participle to subsequent request information.By above-mentioned process, can
So that follow-up word segmentation processing is more accurate, extend one's service during using the range of choice to key word.
Step 1105, carries out the matching treatment of described solicited message and described knowledge point based on the participle information after adjusting,
When the match is successful, after user's transmission problem, according to the answer selecting transmission correspondence of user;Otherwise, repeat above-mentioned display,
Adjust and matching treatment, until the match is successful or reaches predetermined number of repetition.
Specifically, in step 1105, matching treatment can be to use any one mode following:
Mode one: carry out Intelligent Matching by Semantic Similarity Measurement, when there is semantic similarity more than threshold value, coupling
Success;Wherein, after pass-through mode one carries out Intelligent Matching success, described answer is sent in the following ways to user: by semanteme
Similarity is sent to user more than all or part of problem of threshold value, and according to the selection of user, sends selected problem to user
Corresponding answer;
Mode two: carry out fuzzy matching by keyword search, when the problem having user to select in matching result, coupling
Success;
Mode three: first pass through Semantic Similarity Measurement and carry out Intelligent Matching, when there is semantic similarity less than described threshold value
Time, then carry out fuzzy matching by keyword search, when the problem having user to select in matching result, the match is successful.
Wherein, after pass-through mode two and mode three carry out fuzzy matching success, by successfully whole for Keywords matching or portion
Point problem is sent to user, and according to the selection of user, sends, to user, the answer that selected problem is corresponding.
The technical scheme of the embodiment of the present invention allows user to carry out the result of system identification participating in getting involved.When system is carried out
After participle, directly participle information is shown to user and modifies for user, i.e. obtain the language after updating by editor's key word
Justice recommendation results, subsequently, according to the problem (i.e. semantic recommendation results) of the selection of user, pushes corresponding answer to user.Borrow
Help the technical scheme of the embodiment of the present invention, it is possible to increase intelligent robot semantics recognition and the accuracy rate of coupling, promote user
Experience, meet personalized use demand, preferably improve intelligent robot service quality and level.
Device embodiment one
According to embodiments of the invention, it is provided that a kind of automatic call answering arrangement, Figure 12 is apparatus of the present invention embodiment one
The structural representation of automatic call answering arrangement, as shown in figure 12, automatic call answering arrangement according to embodiments of the present invention includes: obtain mould
Block 120, input module 121, word-dividing mode 122, matching module 123, adjusting module 124, output module 125 and control module
126, below in conjunction with accompanying drawing, the technique scheme of the embodiment of the present invention is described in detail.
Acquisition module 120, is used for obtaining multiple knowledge point, and each knowledge point includes problem and answer;
In question answering system, acquisition module 120 needs to obtain knowledge point from knowledge base, and knowledge base plays vital work
With, knowledge base includes that multiple knowledge point, each knowledge point include that question sentence and answer, described question sentence can include that a standard is asked.
Specifically, knowledge point in knowledge base is the most original and simplest form is exactly the FAQ commonly used at ordinary times, general form be " ask-
Answer " right, wherein, should " asking " be exactly that standard is asked, should " answering " be exactly answer.Additionally, each knowledge point in knowledge base can also have
The one or more extensions asked corresponding to a standard are asked, this extension is asked and asked expression-form slightly difference with standard, but expresses
Implication identical.
In knowledge base, standard is asked and extends that ask both can be to have used general problem form, it would however also be possible to employ semantic meaning representation
Formula form, it is same as the prior art, does not repeats them here.
Input module 121, for obtaining the solicited message of user;
In embodiments of the present invention, this solicited message can be obtained by the various ways such as speech recognition, Text region.?
In the embodiment of the present invention, the solicited message of user generally user wants to obtain the problem information corresponding to certain answer.Such as,
The solicited message of user may is that how to arrange CRBT, how to apply for broadband etc..
Word-dividing mode 122, for solicited message is carried out automatic word segmentation, obtains participle information;
In embodiments of the present invention, when carrying out automatic word segmentation, it is usually and carries out according to dictionary for word segmentation.In actual applications,
Automatic word segmentation process can use the one in the two-way maximum matching method of dictionary, viterbi method, HMM method and CRF method or
Multiple.
Such as, solicited message is: how to arrange CRBT, then participle information is: how, arrange, CRBT.
Matching module 123, for making requests on information and knowledge point based on the participle information after participle information or adjustment
Matching treatment;Matching module 123 specifically for:
Any one mode following is used to carry out matching treatment:
Mode one: carry out Intelligent Matching by Semantic Similarity Measurement, when there is semantic similarity more than threshold value, coupling
Success;
Mode two: carry out fuzzy matching by keyword search, when the problem having user to select in matching result, coupling
Success;
Mode three: first pass through Semantic Similarity Measurement and carry out Intelligent Matching, when there is semantic similarity less than threshold value, then
Carrying out fuzzy matching by keyword search, when the problem having user to select in matching result, the match is successful.
In Intelligent Matching, Semantic Similarity Measurement is a step crucial in question answering system, and its result directly influences whole
The accuracy of individual system.The object participating in Semantic Similarity Measurement must be reciprocity, for calculation knowledge point and user please
Seeking the semantic similarity of information (i.e. user's question sentence), defining this semantic similarity is user's question sentence and all question sentences in knowledge point
The maximum semantic similarity of (including that all standards are asked, extended and ask about template).In knowledge point, question sentence is calculating language with user's question sentence
Can be calculated by the co sinus vector included angle formula in tradition vector space model during justice similarity.
It should be noted that in embodiments of the present invention, confirm it is to use Intelligent Matching or employing in the following way
Fuzzy matching, specifically, after the input of the solicited message of user, engine can carry out participle according to the question sentence of user, is carrying out point
After word, need to calculate this solicited message and the semantic similarity of problem in intelligent answer knowledge base according to part of speech, if semantic similitude
Degree more than or equal to the value pre-set, then carries out abstract semantics coupling, i.e. Intelligent Matching;If being not reaching to pre-set
Value, then carry out Keywords matching, i.e. fuzzy matching.
In other embodiments of the invention, it is also possible to only with Intelligent Matching or fuzzy matching make requests on information with
The matching treatment of knowledge point, it is not intended to protection scope of the present invention.
Adjusting module 124, is used for when making requests on the matching treatment failure of information and knowledge point based on participle information, root
According to user operation, participle information is adjusted;
Additionally, when making requests on information and knowledge point based on participle information it fails to match, illustrate to believe based on this participle
Breath can not match suitable answer, and in these cases, the embodiment of the present invention can show participle information, and provides operation
Interface revises this participle information for user, is adjusted participle information according to user operation.In embodiments of the present invention, user
Being adjusted participle information can be to include at least one following several form: increases new participle information, delete former participle letter
Cease, replace former participle information.Such as, former participle information is: how, arrange, CRBT, amended participle information is: how to use
Mobile phone arranges CRBT, in this example, adds " with mobile phone " this new participle information.And for example, in another example, because of
Neologisms None-identified, causes participle mistake, identical solicited message can be provided different word segmentation result now by adjusting.
Output module 125, for when the match is successful, sends answer to user;Output module 125 specifically for:
After matching module 123 carries out Intelligent Matching success by Semantic Similarity Measurement, send answer to user and use
Any one mode below:
The answer of maximum semantic similarity correspondence problem is transmitted directly to user;
Semantic similarity is sent to user more than all or part of problem of threshold value, and according to the selection of user, Xiang Yong
Family sends the answer that selected problem is corresponding;
After matching module 123 carries out fuzzy matching success by keyword search, send the mode bag of answer to user
Include:
Keywords matching successfully all or part of problem is sent to user, and according to the selection of user, sends out to user
Send the answer that selected problem is corresponding.
Control module 126, in the case of the match is successful at matching module 123, calls output module 125, in coupling
Module 123 in the case of it fails to match, calls adjusting module 124 and matching module 123, until the match is successful or reaches
Predetermined number of repetition.
Preferably, in embodiments of the present invention, device also includes:
Display module, after word-dividing mode 122 carries out automatic word segmentation to solicited message and matching module 123 based on point
Before word information makes requests on the matching treatment of information and knowledge point, show participle information;Or, matching module 123 is based on dividing
After the matching treatment failure that word information makes requests on information and knowledge point and participle is believed by adjusting module 124 according to user operation
Before breath is adjusted, show participle information.
More new module, for according to the participle information updating dictionary for word segmentation after adjusting, wherein, dictionary for word segmentation is for follow-up
Solicited message carries out participle.
It can be seen from the above description that in embodiments of the present invention, the problem that intelligent robot system inputs for user
Corresponding knowledge coupling can be carried out, three kinds of matching ways can be included: Intelligent Matching, fuzzy matching and participle are revised (again
Semantic recommendation).Preferably, the order performing above-mentioned three kinds of matching ways is: Intelligent Matching, fuzzy matching, participle amendment, specifically
As follows:
1, Intelligent Matching: i.e. semantic recommendation, first passes through the abstract semantics matched and is identified, if the match is successful,
Then push corresponding matching result to user, perform Intelligent Matching and the match is successful shows that matching result fully meets user and needs
Asking, if mating unsuccessful or user to be unsatisfied with matching result, then carrying out fuzzy matching.
2, fuzzy matching: i.e. show participle, according to participle information, carries out Keywords matching, if the match is successful, then to
Family pushes corresponding matching result, performs fuzzy matching and the match is successful shows that matching result meets user's need to a certain extent
Asking, but there is the error of a part, if mating unsuccessful or user to be unsatisfied with matching result, then revising participle.
3, amendment participle (semantic recommendation again): i.e. user carries out edit-modify to participle, shows " Intelligent Matching " and " mould
Stick with paste coupling " result all cannot meet user's request and there is the biggest gap therewith, it is necessary to by system after manual amendment again
Semantic recommendation just can obtain the self-defined question sentence that user needs.Wherein, the process of participle amendment includes: split into by question sentence some
Individual participle, after user adjusts participle, first contrasts with original question sentence, adjusted for user word is entered into dictionary for word segmentation, the most again
Re-start semantic recommendation and/or fuzzy matching, until obtaining customer satisfaction system matching result or reaching predetermined repetition time
Number.
The technical scheme of the embodiment of the present invention allows user to carry out the result of system identification participating in getting involved.When system is carried out
" Intelligent Matching " and " fuzzy matching " when all can not meet the demand of user, can according to user in " fuzzy matching " point
The amendment of word information, i.e. obtains the semantic recommendation results after updating by editor's key word, thus is met making by oneself of demand
Justice question sentence.Technical scheme by means of the embodiment of the present invention, it is possible to increase intelligent robot semantics recognition and the accuracy rate of coupling,
Promote Consumer's Experience, meet personalized use demand, preferably improve intelligent robot service quality and level.Keywords matching
Device embodiment two
According to embodiments of the invention, it is provided that a kind of automatic call answering arrangement, Figure 13 is apparatus of the present invention embodiment two
Automatic call answering arrangement structural representation, as shown in figure 13, automatic call answering arrangement according to embodiments of the present invention includes: obtain
Module 130, input module 131, word-dividing mode 132, display module 133, adjusting module 134, matching module 135, output module
136 and control module 137, below in conjunction with accompanying drawing, the technical scheme of the embodiment of the present invention is illustrated.
Acquisition module 130, is used for obtaining multiple knowledge point, and each knowledge point includes problem and answer;
In embodiments of the present invention, acquisition module 130 obtains knowledge point from knowledge base, and in question answering system, knowledge base rises
Vital effect, knowledge base includes that multiple knowledge point, each knowledge point include that question sentence and answer, described question sentence can be wrapped
Include a standard to ask.Specifically, the knowledge point in knowledge base is the most original and simplest form is exactly the FAQ commonly used at ordinary times, and one
As form be that " ask-answer " is right, wherein, should " asking " be exactly that standard is asked, should " answering " be exactly answer.Additionally, each in knowledge base
Knowledge point can also have the one or more extensions asked corresponding to a standard to ask, this extension is asked and asked that expression-form is slightly with standard
Difference, but the implication expressed is identical.
In knowledge base, standard is asked and extends that ask both can be to have used general problem form, it would however also be possible to employ semantic meaning representation
Formula form, it is same as the prior art, does not repeats them here.
Input module 131, for obtaining the solicited message of user;In embodiments of the present invention, this solicited message can be led to
Cross the various ways such as speech recognition, Text region to obtain.In embodiments of the present invention, the solicited message of user generally user thinks
Obtain the problem information corresponding to certain answer.Such as, the solicited message of user may is that, how CRBT is set, how Shen
Please broadband etc..
Word-dividing mode 132, for solicited message is carried out automatic word segmentation, obtains participle information;
In embodiments of the present invention, when carrying out automatic word segmentation, it is usually and carries out according to dictionary for word segmentation.In actual applications,
Automatic word segmentation process can use the one in the two-way maximum matching method of dictionary, viterbi method, HMM method and CRF method or
Multiple.
Display module 133, is used for after word-dividing mode 132 obtains participle information, and adjusting module 134 is according to user operation
Before participle information is adjusted, show participle information;
Additionally, unlike device embodiment one, in embodiments of the present invention, after carrying out participle operation, directly will
Participle information is shown to user.
Adjusting module 134, for being adjusted participle information according to user operation;Adjusting module 134 uses with lower section
Described participle information is adjusted by least one formula: increases new participle information, delete former participle information, replace former participle letter
Breath.
If it is to say, user needs to revise participle information, then can operating according to the adjustment of user and this participle believed
Breath is adjusted.
Matching module 135, for making requests on the matching treatment of information and knowledge point based on the participle information after adjusting;?
Join module 135 specifically for:
Any one mode following is used to carry out matching treatment:
Carrying out Intelligent Matching by Semantic Similarity Measurement, when there is semantic similarity more than threshold value, the match is successful;
Carrying out fuzzy matching by keyword search, when the problem having user to select in matching result, the match is successful;
First pass through Semantic Similarity Measurement and carry out Intelligent Matching, when there is semantic similarity less than threshold value, then by closing
Keyword search carries out fuzzy matching, and when the problem having user to select in matching result, the match is successful.
Output module 136, for when the match is successful, sends answer to user;Output module 136 specifically for:
After matching module 135 carries out Intelligent Matching success by Semantic Similarity Measurement, send answer to user and use
Any one mode below:
The answer of maximum semantic similarity correspondence problem is transmitted directly to user;
Semantic similarity is sent to user more than all or part of problem of threshold value, and according to the selection of user, Xiang Yong
Family sends the answer that selected problem is corresponding;
After matching module 135 carries out fuzzy matching success by keyword search, send the mode bag of answer to user
Include:
Keywords matching successfully all or part of problem is sent to user, and according to the selection of user, sends out to user
Send the answer that selected problem is corresponding.
Matching treatment can be to use any one mode following:
Mode one: carry out Intelligent Matching by Semantic Similarity Measurement, when there is semantic similarity more than threshold value, coupling
Success;
Wherein, after pass-through mode one carries out Intelligent Matching success, answer is sent in the following ways to user: by maximum language
The answer of justice similarity correspondence problem is transmitted directly to user;
Mode two: carry out fuzzy matching by keyword search, when the problem having user to select in matching result, coupling
Success;
Mode three: first pass through Semantic Similarity Measurement and carry out Intelligent Matching, when there is semantic similarity less than threshold value, then
Carrying out fuzzy matching by keyword search, when the problem having user to select in matching result, the match is successful.
Wherein, after pass-through mode two and mode three carry out fuzzy matching success, by successfully whole for Keywords matching or portion
Point problem is sent to user, and according to the selection of user, sends, to user, the answer that selected problem is corresponding.
In Intelligent Matching, Semantic Similarity Measurement is a step crucial in question answering system, and its result directly influences whole
The accuracy of individual system.The object participating in Semantic Similarity Measurement must be reciprocity, for calculation knowledge point and user please
Seeking the semantic similarity of information (i.e. user's question sentence), defining this semantic similarity is user's question sentence and all question sentences in knowledge point
The maximum semantic similarity of (including that all standards are asked, extended and ask about template).In knowledge point, question sentence is calculating language with user's question sentence
Can be calculated by the co sinus vector included angle formula in tradition vector space model during justice similarity.
It should be noted that confirming in the following way is to use Intelligent Matching or use fuzzy matching, specifically, use
After the solicited message input at family, engine can carry out participle according to the question sentence of user, after having carried out participle, needs according to part of speech meter
Calculate this solicited message and the semantic similarity of problem in intelligent answer knowledge base, if semantic similarity is more than or equal to setting in advance
The value put, then carry out abstract semantics coupling, i.e. Intelligent Matching;If being not reaching to the value pre-set, then carry out Keywords matching,
I.e. fuzzy matching.
Control module 137, in the case of the match is successful at matching module 135, calls output module 136, in coupling
Module 135 in the case of it fails to match, calls display module 133, adjusting module 134 and matching module 135, until coupling
Success or reach predetermined number of repetition.
Preferably, said apparatus farther includes:
More new module, for according to the participle information updating dictionary for word segmentation after adjusting, wherein, dictionary for word segmentation is for follow-up
Solicited message carries out participle.
It can be seen from the above description that in embodiments of the present invention, the problem that intelligent robot system inputs for user
Corresponding knowledge coupling can be carried out, can include three kinds of matching ways: participle amendment (again semantic recommend), Intelligent Matching, with
And fuzzy matching.Preferably, the order performing above-mentioned three kinds of matching ways is: participle amendment, Intelligent Matching, fuzzy matching, tool
Body is as follows:
1, amendment participle (semantic recommendation again): i.e. user carries out edit-modify to participle, and showing must be by manually repairing
Change rear system semantic recommendation again and just can obtain the self-defined question sentence that user needs.Wherein, the process of participle amendment includes: will ask
Sentence splits into several participles, after user adjusts participle, first contrasts with original question sentence, adjusted for user word is entered into participle
Dictionary, carries out semantic recommendation the most again.
2, Intelligent Matching: i.e. semantic recommendation, first passes through the abstract semantics matched and is identified, if the match is successful,
Then push corresponding matching result to user, perform Intelligent Matching and the match is successful shows that matching result fully meets user and needs
Asking, if mating unsuccessful or user to be unsatisfied with matching result, then carrying out fuzzy matching.
3, fuzzy matching: i.e. show participle, according to participle information, carries out Keywords matching, if the match is successful, then to
Family pushes corresponding matching result, performs fuzzy matching and the match is successful shows that matching result meets user's need to a certain extent
Asking, but there is the error of a part, if mating unsuccessful or user to be unsatisfied with matching result, then remodifying participle, until
Obtain customer satisfaction system matching result or reach predetermined number of repetition.
Put down it should be noted that the technique scheme of the embodiment of the present invention may be used for the service of iBot Cloud intelligent cloud
Platform.
The technical scheme of the Keywords matching Keywords matching embodiment of the present invention allows user to enter the result of system identification
Row participates in getting involved.After system carries out participle, directly participle information is shown to user and modifies for user, i.e. by editor
Key word obtains the semantic recommendation results after updating, subsequently, according to the problem (i.e. semantic recommendation results) of the selection of user, Xiang Yong
Family pushes corresponding answer.Technical scheme by means of the embodiment of the present invention, it is possible to increase intelligent robot semantics recognition and
The accuracy rate joined, promotes Consumer's Experience, meets personalized use demand, preferably improves intelligent robot service quality and water
Flat.
Obviously, those skilled in the art can carry out various change and the modification essence without deviating from the present invention to the present invention
God and scope.So, if these amendments of the present invention and modification belong to the scope of the claims in the present invention and equivalent technologies thereof
Within, then the present invention is also intended to comprise these change and modification.
Algorithm and display are not intrinsic to any certain computer, virtual system or miscellaneous equipment relevant provided herein.
Various general-purpose systems can also be used together with based on teaching in this.As described above, construct required by this kind of system
Structure be apparent from.Additionally, the present invention is also not for any certain programmed language.It is understood that, it is possible to use various
Programming language realizes the content of invention described herein, and the description done language-specific above is to disclose this
Bright preferred forms.
In description mentioned herein, illustrate a large amount of detail.It is to be appreciated, however, that the enforcement of the present invention
Example can be put into practice in the case of not having these details.In some instances, it is not shown specifically known method, structure
And technology, in order to do not obscure the understanding of this description.
Similarly, it will be appreciated that one or more in order to simplify that the disclosure helping understands in each inventive aspect, exist
Above in the description of the exemplary embodiment of the present invention, each feature of the present invention is grouped together into single enforcement sometimes
In example, figure or descriptions thereof.But, the method for the disclosure should not be construed to reflect an intention that i.e. required guarantor
The application claims feature more more than the feature being expressly recited in each claim protected.More precisely, as following
Claims reflected as, inventive aspect is all features less than single embodiment disclosed above.Therefore,
The claims following detailed description of the invention are thus expressly incorporated in this detailed description of the invention, the most each claim itself
All as the independent embodiment of the present invention.
Those skilled in the art are appreciated that and the module in the client in embodiment can be carried out adaptivity
Ground changes and they is arranged in one or more clients different from this embodiment.Can be the module in embodiment
It is combined into a module, and multiple submodule or subelement or sub-component can be put them in addition.Except such spy
Levy and/or outside at least some in process or unit excludes each other, can use any combination that this specification (is included
Adjoint claim, summary and accompanying drawing) disclosed in all features and so disclosed any method or client
All processes or unit are combined.Unless expressly stated otherwise, this specification (includes adjoint claim, summary and attached
Figure) disclosed in each feature can be replaced by providing identical, equivalent or the alternative features of similar purpose.
Although additionally, it will be appreciated by those of skill in the art that embodiments more described herein include other embodiments
Some feature included by rather than further feature, but the combination of the feature of different embodiment means to be in the present invention's
Within the scope of and form different embodiments.Such as, in the following claims, embodiment required for protection appoint
One of meaning can mode use in any combination.
The all parts embodiment of the present invention can realize with hardware, or to run on one or more processor
Software module realize, or with combinations thereof realize.It will be understood by those of skill in the art that and can use in practice
Microprocessor or digital signal processor (DSP) realize the client being loaded with sequence network address according to embodiments of the present invention
In the some or all functions of some or all parts.The present invention is also implemented as performing as described herein
Part or all equipment of method or device program (such as, computer program and computer program).So
The program realizing the present invention can store on a computer-readable medium, or can have the shape of one or more signal
Formula.Such signal can be downloaded from internet website and obtain, or provides on carrier signal, or with any other shape
Formula provides.
The present invention will be described rather than limits the invention to it should be noted above-described embodiment, and ability
Field technique personnel can design alternative embodiment without departing from the scope of the appended claims.In the claims,
Any reference marks that should not will be located between bracket is configured to limitations on claims.Word " comprises " and does not excludes the presence of not
Arrange element in the claims or step.Word "a" or "an" before being positioned at element does not excludes the presence of multiple such
Element.The present invention and can come real by means of including the hardware of some different elements by means of properly programmed computer
Existing.If in the unit claim listing equipment for drying, several in these devices can be by same hardware branch
Specifically embody.Word first, second and third use do not indicate that any order.These word explanations can be run after fame
Claim.
Claims (22)
1. an automatic question-answering method, it is characterised in that including:
Thering is provided question and answer knowledge base, described question and answer knowledge base includes that multiple knowledge point, each described knowledge point include problem and answer;
Obtain the solicited message of user;
Described solicited message is carried out automatic word segmentation, obtains participle information;
The matching treatment of described solicited message and described knowledge point, when the match is successful, Xiang Yong is carried out based on described participle information
Family sends answer;Otherwise, according to user operation, described participle information is adjusted, carries out institute based on the participle information after adjusting
State the matching treatment of solicited message and described knowledge point, when the match is successful, send answer to user;Otherwise, above-mentioned tune is repeated
Whole and matching treatment, until the match is successful or reaches predetermined number of repetition.
2. the method for claim 1, it is characterised in that described method farther includes:
After described solicited message is carried out automatic word segmentation and carry out described solicited message based on described participle information to know with described
Before knowing the matching treatment of point, show described participle information;Or,
Carry out the matching treatment failure of described solicited message and described knowledge point based on described participle information after and according to user
Before described participle information is adjusted by operation, show described participle information.
3. the method for claim 1, it is characterised in that described matching treatment employing any one mode following:
Carrying out Intelligent Matching by Semantic Similarity Measurement, when there is semantic similarity more than threshold value, the match is successful;
Carrying out fuzzy matching by keyword search, when the problem having user to select in matching result, the match is successful;
First pass through Semantic Similarity Measurement and carry out Intelligent Matching, when there is semantic similarity less than described threshold value, then by closing
Keyword search carries out fuzzy matching, and when the problem having user to select in matching result, the match is successful.
4. method as claimed in claim 3, it is characterised in that
By Semantic Similarity Measurement carry out Intelligent Matching success after, described to user send answer use below any one side
Formula:
The answer of maximum semantic similarity correspondence problem is transmitted directly to user;
Semantic similarity is sent to user more than all or part of problem of threshold value, and according to the selection of user, sends out to user
Send the answer that selected problem is corresponding;
After carrying out fuzzy matching success by keyword search, described send the mode of answer to user and include:
Keyword match successfully all or part of problem is sent to user, and according to the selection of user, sends institute to user
Select the answer that problem is corresponding.
5. the method for claim 1, it is characterised in that the side described participle information being adjusted according to user operation
Formula includes at least one of: increases new participle information, delete former participle information, replace former participle information.
6. the method for claim 1, it is characterised in that described participle information is adjusted it according to user operation
After, described method farther includes:
According to the participle information updating dictionary for word segmentation after adjusting, wherein, described dictionary for word segmentation is for carrying out subsequent request information
Participle.
7. an automatic question-answering method, it is characterised in that including:
Thering is provided question and answer knowledge base, described question and answer knowledge base includes that multiple knowledge point, each described knowledge point include problem and answer;
Obtain the solicited message of user;
Described solicited message is carried out automatic word segmentation, obtains and show participle information;
According to user operation, described participle information is adjusted;
The matching treatment of described solicited message and described knowledge point is carried out based on the participle information after adjusting, when the match is successful,
Answer is sent to user;Otherwise, repeat above-mentioned display, adjustment and matching treatment, until the match is successful or it is predetermined to reach
Number of repetition.
8. method as claimed in claim 7, it is characterised in that described matching treatment employing any one mode following:
Carrying out Intelligent Matching by Semantic Similarity Measurement, when there is semantic similarity more than threshold value, the match is successful;
Carrying out fuzzy matching by keyword search, when the problem having user to select in matching result, the match is successful;
First pass through Semantic Similarity Measurement and carry out Intelligent Matching, when there is semantic similarity less than described threshold value, then by closing
Keyword search carries out fuzzy matching, and when the problem having user to select in matching result, the match is successful.
9. method as claimed in claim 7, it is characterised in that
By Semantic Similarity Measurement carry out Intelligent Matching success after, described to user send answer use below any one side
Formula:
The answer of maximum semantic similarity correspondence problem is transmitted directly to user;
Semantic similarity is sent to user more than all or part of problem of threshold value, and according to the selection of user, sends out to user
Send the answer that selected problem is corresponding;
After carrying out fuzzy matching success by keyword search, described send the mode of answer to user and include:
Keyword match successfully all or part of problem is sent to user, and according to the selection of user, sends institute to user
Select the answer that problem is corresponding.
10. method as claimed in claim 7, it is characterised in that described participle information is adjusted according to user operation
Mode includes at least one of: increases new participle information, delete former participle information, replace former participle information.
11. methods as claimed in claim 7, it is characterised in that described participle information is adjusted it according to user operation
After, described method farther includes:
According to the described participle information updating dictionary for word segmentation after adjusting, wherein, described dictionary for word segmentation is for subsequent request information
Carry out participle.
12. 1 kinds of automatic call answering arrangements, it is characterised in that including: acquisition module, input module, word-dividing mode, matching module,
Adjusting module, output module and control module, wherein:
Acquisition module, is used for obtaining multiple knowledge point, and each described knowledge point includes problem and answer;
Input module, for obtaining the solicited message of user;
Word-dividing mode, for described solicited message is carried out automatic word segmentation, obtains participle information;
Matching module, for carrying out described solicited message and described knowledge based on the participle information after described participle information or adjustment
The matching treatment of point;
Adjusting module, for when the matching treatment failure carrying out described solicited message and described knowledge point based on described participle information
Time, according to user operation, described participle information is adjusted;
Output module, for when the match is successful, sends answer to user;
Control module, in the case of the match is successful at described matching module, calls described output module, at described coupling mould
In the case of Block-matching failure, call described adjusting module and described matching module, until the match is successful or it is predetermined to reach
Number of repetition.
13. devices as claimed in claim 12, it is characterised in that described device also includes:
Display module, after described word-dividing mode carries out automatic word segmentation to described solicited message and described matching module based on
Before described participle information carries out the matching treatment of described solicited message and described knowledge point, show described participle information;Or,
After the matching treatment failure that described matching module carries out described solicited message and described knowledge point based on described participle information and
Before described participle information is adjusted by described adjusting module according to user operation, show described participle information.
14. devices as claimed in claim 12, it is characterised in that described matching module specifically for:
Any one mode following is used to carry out matching treatment:
Carrying out Intelligent Matching by Semantic Similarity Measurement, when there is semantic similarity more than threshold value, the match is successful;
Carrying out fuzzy matching by keyword search, when the problem having user to select in matching result, the match is successful;
First pass through Semantic Similarity Measurement and carry out Intelligent Matching, when there is semantic similarity less than described threshold value, then by closing
Keyword search carries out fuzzy matching, and when the problem having user to select in matching result, the match is successful.
15. devices as claimed in claim 14, it is characterised in that described output module specifically for:
After described matching module carries out Intelligent Matching success by Semantic Similarity Measurement, described send answer to user and use
Any one mode below:
The answer of maximum semantic similarity correspondence problem is transmitted directly to user;
Semantic similarity is sent to user more than all or part of problem of threshold value, and according to the selection of user, sends out to user
Send the answer that selected problem is corresponding;
After described matching module carries out fuzzy matching success by keyword search, the described mode bag sending answer to user
Include:
Keyword match successfully all or part of problem is sent to user, and according to the selection of user, sends institute to user
Select the answer that problem is corresponding.
16. devices as claimed in claim 12, it is characterised in that described adjusting module specifically for:
At least one in the following ways described participle information is adjusted: increase new participle information, delete former participle letter
Cease, replace former participle information.
17. devices as claimed in claim 12, it is characterised in that described device farther includes:
More new module, for according to the participle information updating dictionary for word segmentation after adjusting, wherein, described dictionary for word segmentation is for follow-up
Solicited message carries out participle.
18. 1 kinds of automatic call answering arrangements, it is characterised in that including: acquisition module, input module, word-dividing mode, display module,
Adjusting module, matching module, output module and control module, wherein:
Acquisition module, is used for obtaining multiple knowledge point, and each described knowledge point includes problem and answer;
Input module, for obtaining the solicited message of user;
Word-dividing mode, for described solicited message is carried out automatic word segmentation, obtains participle information;
Display module, for after described word-dividing mode obtains participle information, and adjusting module according to user operation to described point
Before word information is adjusted, show described participle information;
Adjusting module, for being adjusted described participle information according to user operation;
Matching module, for carrying out the matching treatment of described solicited message and described knowledge point based on the participle information after adjusting;
Output module, for when the match is successful, sends answer to user;
Control module, in the case of the match is successful at described matching module, calls described output module, at described coupling mould
In the case of Block-matching failure, call described display module, described adjusting module and described matching module, until mating into
Merit or reach predetermined number of repetition.
19. devices as claimed in claim 18, it is characterised in that described matching module specifically for:
Any one mode following is used to carry out matching treatment:
Carrying out Intelligent Matching by Semantic Similarity Measurement, when there is semantic similarity more than threshold value, the match is successful;
Carrying out fuzzy matching by keyword search, when the problem having user to select in matching result, the match is successful;
First pass through Semantic Similarity Measurement and carry out Intelligent Matching, when there is semantic similarity less than described threshold value, then by closing
Keyword search carries out fuzzy matching, and when the problem having user to select in matching result, the match is successful.
20. devices as claimed in claim 19, it is characterised in that described output module specifically for:
After described matching module carries out Intelligent Matching success by Semantic Similarity Measurement, described send answer to user and use
Any one mode below:
The answer of maximum semantic similarity correspondence problem is transmitted directly to user;
Semantic similarity is sent to user more than all or part of problem of threshold value, and according to the selection of user, sends out to user
Send the answer that selected problem is corresponding;
After described matching module carries out fuzzy matching success by keyword search, the described mode bag sending answer to user
Include:
Keyword match successfully all or part of problem is sent to user, and according to the selection of user, sends institute to user
Select the answer that problem is corresponding.
21. devices as claimed in claim 18, it is characterised in that described adjusting module specifically for:
At least one in the following ways described participle information is adjusted: increase new participle information, delete former participle letter
Cease, replace former participle information.
22. devices as claimed in claim 18, it is characterised in that described device farther includes:
More new module, for according to the participle information updating dictionary for word segmentation after adjusting, wherein, described dictionary for word segmentation is for follow-up
Solicited message carries out participle.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610452735.7A CN106095988A (en) | 2016-06-21 | 2016-06-21 | Automatic question-answering method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610452735.7A CN106095988A (en) | 2016-06-21 | 2016-06-21 | Automatic question-answering method and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106095988A true CN106095988A (en) | 2016-11-09 |
Family
ID=57238367
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610452735.7A Pending CN106095988A (en) | 2016-06-21 | 2016-06-21 | Automatic question-answering method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106095988A (en) |
Cited By (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108153868A (en) * | 2017-12-25 | 2018-06-12 | 安徽磐众信息科技有限公司 | A kind of voice encyclopaedia for option trade answers machine |
CN108153738A (en) * | 2018-02-10 | 2018-06-12 | 灯塔财经信息有限公司 | A kind of chat record analysis method and device based on hierarchical clustering |
CN108182287A (en) * | 2018-01-29 | 2018-06-19 | 百度在线网络技术(北京)有限公司 | A kind of automatic question-answering method, device and server |
CN108287823A (en) * | 2018-02-07 | 2018-07-17 | 平安科技(深圳)有限公司 | Message data processing method, device, computer equipment and storage medium |
CN108388558A (en) * | 2018-02-07 | 2018-08-10 | 平安普惠企业管理有限公司 | Problem matching process, device, customer service robot and storage medium |
CN109190997A (en) * | 2018-09-18 | 2019-01-11 | 广东电网有限责任公司 | The hierarchical parsing of Chinese address and specification handles method and system |
CN109241247A (en) * | 2018-06-28 | 2019-01-18 | 平安科技(深圳)有限公司 | The problem of multiparty collaboration project processing method, system and server |
CN109325182A (en) * | 2018-10-12 | 2019-02-12 | 平安科技(深圳)有限公司 | Dialogue-based information-pushing method, device, computer equipment and storage medium |
CN109509002A (en) * | 2018-06-15 | 2019-03-22 | 中国建设银行股份有限公司 | Automatically configure the system and method for realizing conversation process, storage medium |
CN109597874A (en) * | 2018-10-23 | 2019-04-09 | 重庆邂智科技有限公司 | Information recommendation method, device and server |
CN109597559A (en) * | 2018-12-10 | 2019-04-09 | 联想(北京)有限公司 | A kind of exchange method, device and electronic equipment |
CN109858007A (en) * | 2017-11-30 | 2019-06-07 | 上海智臻智能网络科技股份有限公司 | Semantic analysis answering method and device, computer equipment and storage medium |
CN109977204A (en) * | 2019-03-08 | 2019-07-05 | 科讯嘉联信息技术有限公司 | A kind of intelligent Answer System and method in knowledge based library |
CN110019838A (en) * | 2017-12-25 | 2019-07-16 | 上海智臻智能网络科技股份有限公司 | Intelligent Answer System and intelligent terminal |
CN110019731A (en) * | 2017-12-25 | 2019-07-16 | 上海智臻智能网络科技股份有限公司 | Intelligent interactive method, device, terminal device and storage medium |
CN110019723A (en) * | 2017-12-21 | 2019-07-16 | 上海智臻智能网络科技股份有限公司 | Answering method, terminal device and storage medium based on question answering system |
CN110209773A (en) * | 2017-12-21 | 2019-09-06 | 上海智臻智能网络科技股份有限公司 | A kind of question and answer system based on question answering system |
CN110347787A (en) * | 2019-06-12 | 2019-10-18 | 平安科技(深圳)有限公司 | A kind of interview method, apparatus and terminal device based on AI secondary surface examination hall scape |
CN110795540A (en) * | 2018-07-17 | 2020-02-14 | 上海智臻智能网络科技股份有限公司 | Attachment preview method and device of question-answering system, storage medium and terminal |
CN112015874A (en) * | 2020-07-30 | 2020-12-01 | 上海松鼠课堂人工智能科技有限公司 | Student mental health accompany conversation system |
CN113157868A (en) * | 2021-04-29 | 2021-07-23 | 青岛海信网络科技股份有限公司 | Method and device for matching answers to questions based on structured database |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102073665A (en) * | 2009-11-25 | 2011-05-25 | 陈新康 | Method for computer retrieval of client name based on lexical analysis |
CN102799676A (en) * | 2012-07-18 | 2012-11-28 | 上海语天信息技术有限公司 | Recursive and multilevel Chinese word segmentation method |
CN104077275A (en) * | 2014-06-27 | 2014-10-01 | 北京奇虎科技有限公司 | Method and device for performing word segmentation based on context |
CN104899187A (en) * | 2014-03-06 | 2015-09-09 | 武汉元宝创意科技有限公司 | Man-computer interaction word segmentation and semantic marking method and man-computer interaction word segmentation and semantic marking system |
CN104899304A (en) * | 2015-06-12 | 2015-09-09 | 北京京东尚科信息技术有限公司 | Named entity identification method and device |
CN104951456A (en) * | 2014-03-26 | 2015-09-30 | 上海智臻网络科技有限公司 | Method, device and equipment used for obtaining answer information |
-
2016
- 2016-06-21 CN CN201610452735.7A patent/CN106095988A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102073665A (en) * | 2009-11-25 | 2011-05-25 | 陈新康 | Method for computer retrieval of client name based on lexical analysis |
CN102799676A (en) * | 2012-07-18 | 2012-11-28 | 上海语天信息技术有限公司 | Recursive and multilevel Chinese word segmentation method |
CN104899187A (en) * | 2014-03-06 | 2015-09-09 | 武汉元宝创意科技有限公司 | Man-computer interaction word segmentation and semantic marking method and man-computer interaction word segmentation and semantic marking system |
CN104951456A (en) * | 2014-03-26 | 2015-09-30 | 上海智臻网络科技有限公司 | Method, device and equipment used for obtaining answer information |
CN104077275A (en) * | 2014-06-27 | 2014-10-01 | 北京奇虎科技有限公司 | Method and device for performing word segmentation based on context |
CN104899304A (en) * | 2015-06-12 | 2015-09-09 | 北京京东尚科信息技术有限公司 | Named entity identification method and device |
Non-Patent Citations (1)
Title |
---|
陈金华: "《智慧学习环境构建》", 30 September 2013, 国防工业出版社 * |
Cited By (28)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109858007A (en) * | 2017-11-30 | 2019-06-07 | 上海智臻智能网络科技股份有限公司 | Semantic analysis answering method and device, computer equipment and storage medium |
CN109858007B (en) * | 2017-11-30 | 2024-02-02 | 上海智臻智能网络科技股份有限公司 | Semantic analysis question-answering method and device, computer equipment and storage medium |
CN110209773A (en) * | 2017-12-21 | 2019-09-06 | 上海智臻智能网络科技股份有限公司 | A kind of question and answer system based on question answering system |
CN110019723A (en) * | 2017-12-21 | 2019-07-16 | 上海智臻智能网络科技股份有限公司 | Answering method, terminal device and storage medium based on question answering system |
CN108153868A (en) * | 2017-12-25 | 2018-06-12 | 安徽磐众信息科技有限公司 | A kind of voice encyclopaedia for option trade answers machine |
CN110019731A (en) * | 2017-12-25 | 2019-07-16 | 上海智臻智能网络科技股份有限公司 | Intelligent interactive method, device, terminal device and storage medium |
CN110019838A (en) * | 2017-12-25 | 2019-07-16 | 上海智臻智能网络科技股份有限公司 | Intelligent Answer System and intelligent terminal |
CN108182287A (en) * | 2018-01-29 | 2018-06-19 | 百度在线网络技术(北京)有限公司 | A kind of automatic question-answering method, device and server |
CN108388558B (en) * | 2018-02-07 | 2022-04-19 | 平安普惠企业管理有限公司 | Question matching method and device, customer service robot and storage medium |
CN108287823A (en) * | 2018-02-07 | 2018-07-17 | 平安科技(深圳)有限公司 | Message data processing method, device, computer equipment and storage medium |
CN108388558A (en) * | 2018-02-07 | 2018-08-10 | 平安普惠企业管理有限公司 | Problem matching process, device, customer service robot and storage medium |
CN108153738A (en) * | 2018-02-10 | 2018-06-12 | 灯塔财经信息有限公司 | A kind of chat record analysis method and device based on hierarchical clustering |
CN109509002A (en) * | 2018-06-15 | 2019-03-22 | 中国建设银行股份有限公司 | Automatically configure the system and method for realizing conversation process, storage medium |
CN109241247B (en) * | 2018-06-28 | 2023-06-30 | 平安科技(深圳)有限公司 | Multi-party collaboration project problem processing method, system and server |
CN109241247A (en) * | 2018-06-28 | 2019-01-18 | 平安科技(深圳)有限公司 | The problem of multiparty collaboration project processing method, system and server |
CN110795540A (en) * | 2018-07-17 | 2020-02-14 | 上海智臻智能网络科技股份有限公司 | Attachment preview method and device of question-answering system, storage medium and terminal |
CN109190997A (en) * | 2018-09-18 | 2019-01-11 | 广东电网有限责任公司 | The hierarchical parsing of Chinese address and specification handles method and system |
CN109325182A (en) * | 2018-10-12 | 2019-02-12 | 平安科技(深圳)有限公司 | Dialogue-based information-pushing method, device, computer equipment and storage medium |
CN109325182B (en) * | 2018-10-12 | 2024-05-14 | 平安科技(深圳)有限公司 | Information pushing method and device based on session, computer equipment and storage medium |
CN109597874A (en) * | 2018-10-23 | 2019-04-09 | 重庆邂智科技有限公司 | Information recommendation method, device and server |
CN109597874B (en) * | 2018-10-23 | 2023-05-23 | 重庆邂智科技有限公司 | Information recommendation method, device and server |
CN109597559A (en) * | 2018-12-10 | 2019-04-09 | 联想(北京)有限公司 | A kind of exchange method, device and electronic equipment |
CN109977204A (en) * | 2019-03-08 | 2019-07-05 | 科讯嘉联信息技术有限公司 | A kind of intelligent Answer System and method in knowledge based library |
CN110347787A (en) * | 2019-06-12 | 2019-10-18 | 平安科技(深圳)有限公司 | A kind of interview method, apparatus and terminal device based on AI secondary surface examination hall scape |
CN110347787B (en) * | 2019-06-12 | 2022-12-09 | 平安科技(深圳)有限公司 | Interview method and device based on AI auxiliary interview scene and terminal equipment |
CN112015874A (en) * | 2020-07-30 | 2020-12-01 | 上海松鼠课堂人工智能科技有限公司 | Student mental health accompany conversation system |
CN113157868A (en) * | 2021-04-29 | 2021-07-23 | 青岛海信网络科技股份有限公司 | Method and device for matching answers to questions based on structured database |
CN113157868B (en) * | 2021-04-29 | 2022-11-11 | 青岛海信网络科技股份有限公司 | Method and device for matching answers to questions based on structured database |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106095988A (en) | Automatic question-answering method and device | |
CN110008322B (en) | Method and device for recommending dialogues in multi-turn conversation scene | |
KR102431232B1 (en) | Method and apparatus for processing questions and answers, electronic device and storage medium | |
WO2020177282A1 (en) | Machine dialogue method and apparatus, computer device, and storage medium | |
US20180004729A1 (en) | State machine based context-sensitive system for managing multi-round dialog | |
CN107292696B (en) | Intelligent automobile shopping guide system and implementation method | |
US9269354B2 (en) | Semantic re-ranking of NLU results in conversational dialogue applications | |
US20140309990A1 (en) | Semantic re-ranking of nlu results in conversational dialogue applications | |
CN109408731A (en) | A kind of multiple target recommended method, multiple target recommended models generation method and device | |
US20140288913A1 (en) | Customizable and low-latency interactive computer-aided translation | |
US20140257792A1 (en) | Anaphora Resolution Using Linguisitic Cues, Dialogue Context, and General Knowledge | |
CN109920414A (en) | Nan-machine interrogation's method, apparatus, equipment and storage medium | |
CN110852695A (en) | Task approval method and device, server and storage medium | |
CN111339309B (en) | Corpus expansion method and system for user intention | |
US20200074999A1 (en) | Semi-supervised question answering machine | |
CN111191450A (en) | Corpus cleaning method, corpus entry device and computer-readable storage medium | |
CN111597313B (en) | Question answering method, device, computer equipment and storage medium | |
US20200327197A1 (en) | Document-based response generation system | |
CN105760361A (en) | Language model building method and device | |
CN111723270A (en) | RPA robot searching method, device and equipment | |
CN116956116A (en) | Text processing method and device, storage medium and electronic equipment | |
CN114365118A (en) | Knowledge graph-based queries in an artificial intelligence chat robot with basic query element detection and graphical path generation | |
CN111552787A (en) | Question and answer processing method, device, equipment and storage medium | |
CN112925894B (en) | Method, system and device for matching bid-asking questions in conversation | |
US11551012B2 (en) | Apparatus and method for providing personal assistant service based on automatic translation |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161109 |