CN110245240A - A kind of determination method and device of problem data answer - Google Patents
A kind of determination method and device of problem data answer Download PDFInfo
- Publication number
- CN110245240A CN110245240A CN201910472611.9A CN201910472611A CN110245240A CN 110245240 A CN110245240 A CN 110245240A CN 201910472611 A CN201910472611 A CN 201910472611A CN 110245240 A CN110245240 A CN 110245240A
- Authority
- CN
- China
- Prior art keywords
- problem data
- word
- answer
- data
- map
- 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
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/25—Integrating or interfacing systems involving database management systems
- G06F16/254—Extract, transform and load [ETL] procedures, e.g. ETL data flows in data warehouses
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/28—Databases characterised by their database models, e.g. relational or object models
- G06F16/284—Relational databases
- G06F16/288—Entity relationship models
-
- 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
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- 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
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- 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)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
This application discloses a kind of determination method and devices of problem data answer, are related to computer field, for accurately determining the answer of problem data.It include: the first problem data for obtaining user's input;The first problem data include at least one word;Determine the word relationship at least one word between each word;The corresponding service type of the first problem data is determined according at least one described word, the word relationship and intention map;Wherein, the service type for being intended to map and including at least at least one word association;The answer that the first problem data are determined in answer is preset from least one according to the corresponding service type of the first problem data.
Description
Technical field
The invention relates to field of computer technology more particularly to the determination methods and dress of a kind of problem data answer
It sets.
Background technique
The realization of question answering system needs based on powerful and comprehensive knowledge resource library.The knowledge resource library of early stage is most
It is the knowledge resource library that the respective professional domain of building is gone by the expert of every field.But expert constructs the shortcomings that knowledge resource library
It is time-consuming and laborious when constructing large-scale knowledge resource library, and when conversion art, it is necessary to construct another field
Knowledge resource library, and the structure in the knowledge resource library of the construction method in the knowledge resource library of every field and building completion is general
In the case of be inconsistent.Construction method does not have versatility and the structure in knowledge resource library does not have consistency, cannot generate
The question answering system of general field.
While constructing knowledge resource library, it is also necessary to deeply understand the problem of user proposes data.Due to calculating mechanism
It is relatively difficult to solve human language, solving this problem data at present, there are two types of methods: one is semantic analytic method (scmantic
Parsing), another kind is based on method for information retrieval.
The semantic most common method of analytic method is to utilize Combinatory Categorial Grammar (combinatory categorial
Grammar, CCG).But the vocabulary that plays an important role in CCG method therefore is limited by manually generated
The conversion and extension in field.If field is converted, the specific vocabulary of a collection of frontier must be regenerated, and needing to give birth to
It will take a lot of manpower and time when at large-scale words.
Based on method for information retrieval comparative maturity and simple and practical, and the manually generated vocabulary as CCG is not needed,
But a disadvantage is that requiring that the word or word in question sentence must be included at least in answer based on method for information retrieval, so being not so good as
Semantic analytic method is accurate.
Summary of the invention
Embodiments herein provides a kind of determination method and device of problem data answer, meets for accurately determining
The answer of user demand simultaneously saves human cost.
In order to achieve the above objectives, embodiments herein adopts the following technical scheme that
In a first aspect, a kind of determination method of problem data answer is provided, this method comprises:
Obtain the first problem data of user's input;First problem data include at least one word;Determine at least one
Word relationship in word between each word;First problem is determined according at least one word, word relationship and intention map
The corresponding service type of data;Wherein, it is intended that map includes at least the service type of at least one word association;It is asked according to first
The corresponding service type of topic data presets the answer that first problem data are determined in answer from least one.
The processing method of problem data provided by the embodiments of the present application is determined extremely by data the problem of identification user's input
The word relationship of each word in a few word and at least one word, so as to accurately parse problem data;Pass through
It determines the corresponding service type of problem data, reduces search range, can quickly determine the answer of problem data, therefore can be with
Meet user demand, and reduces cost of labor.
Second aspect, provides a kind of determining device of problem data answer, which can be server, or
Applied to the chip of server, the apparatus may include: communication unit, determination unit;Communication unit, for obtaining user's input
First problem data;First problem data include at least one word;Determination unit, it is every at least one word for determining
Word relationship between a word;Determination unit is also used to according at least one word, word relationship and is intended to map and determines the
The corresponding service type of one problem data;Wherein, it is intended that map includes at least the service type of at least one word association;
Determination unit is also used to preset that first problem is determined in answer from least one according to the corresponding service type of first problem data
The answer of data.
The third aspect provides a kind of readable storage medium storing program for executing, instruction is stored in readable storage medium storing program for executing, when instruction is performed
When, realize the method such as first aspect.
Fourth aspect provides a kind of computer program product, and computer program product includes at least one instruction, when extremely
When a few instruction is run on computers, so that computer executes the method such as first aspect.
5th aspect, provides a kind of chip, and chip includes at least one processor and communication interface, communication interface and at least
One processor coupling, at least one processor is for running computer program or instruction, the method to realize first aspect.
The device or computer storage medium or computer program product or chip of above-mentioned offer are used to execute institute above
The corresponding method provided, therefore, the attainable beneficial effect of institute can refer to counterparty in corresponding method provided above
The beneficial effect of case, details are not described herein again.
Detailed description of the invention
Fig. 1 is a kind of structural schematic diagram for question answering system that embodiments herein provides;
Fig. 2 is a kind of determination method flow schematic diagram one for problem data answer that embodiments herein provides;
Fig. 3 is a kind of determination method flow schematic diagram two for problem data answer that embodiments herein provides;
Fig. 4 is a kind of determination method flow schematic diagram three for problem data answer that embodiments herein provides;
Fig. 5 is a kind of structural schematic diagram for consciousness map that embodiments herein provides;
Fig. 6 is a kind of structural schematic diagram including comparing the consciousness map of service that embodiments herein provides;
Fig. 7 is a kind of data structure schematic diagram for calculation type that embodiments herein provides;
Fig. 8 is a kind of determination method flow schematic diagram four for problem data answer that embodiments herein provides;
Fig. 9 is a kind of determination method flow schematic diagram five for problem data answer that embodiments herein provides;
Figure 10 is a kind of determination apparatus structure schematic diagram one for problem data answer that embodiments herein provides;
Figure 11 is a kind of determination apparatus structure schematic diagram two for problem data answer that embodiments herein provides;
Figure 12 is a kind of determination apparatus structure schematic diagram three for problem data answer that embodiments herein provides;
Figure 13 is a kind of apparatus structure schematic diagram for chip that embodiments herein provides.
Specific embodiment
For the ease of clearly describing the technical solution of the embodiment of the present application, in embodiments herein, use " the
One ", the printed words such as " second " distinguish function and the essentially identical identical entry of effect or similar item.For example, the first history is asked
Data and the second historical problem data are inscribed just for the sake of distinguishing different historical problem data, its sequencing are not carried out
It limits.It will be appreciated by those skilled in the art that the printed words such as " first ", " second " are not defined quantity and execution order, and
And the printed words such as " first ", " second " also do not limit certain difference.
It should be noted that in the application, " illustrative " or " such as " etc. words make example, illustration for indicating or say
It is bright.Described herein as " illustrative " or " such as " any embodiment or design scheme be not necessarily to be construed as than it
His embodiment or design scheme more preferably or more advantage.Specifically, use " illustrative " or " such as " etc. words be intended to
Related notion is presented in specific ways.
In the application, "at least one" refers to one or more, and " multiple " refer to two or more."and/or",
The incidence relation of affiliated partner is described, indicates may exist three kinds of relationships, for example, A and/or B, can indicate: individualism A,
Exist simultaneously A and B, the case where individualism B, wherein A, B can be odd number or plural number.Character "/" typicallys represent front and back and closes
Connection object is a kind of relationship of "or".At least one of " following (a) " or its similar expression, refer to these in any group
It closes, any combination including individual event (a) or complex item (a).For example, at least one (a) in a, b or c, can indicate:
A, b, c, a-b, a-c, b-c or a-b-c, wherein a, b, c can be individually, be also possible to multiple.
As shown in Figure 1, Fig. 1 shows a kind of structural schematic diagram of question answering system provided by the embodiments of the present application.The question and answer
System includes: input layer, Entity recognition layer, intention assessment layer, problem data answer layer.
Wherein, input layer: problem data is inputted for retrieving user.Input layer may include: voice input, chat input
And search input.
Entity recognition layer: user inputs at least one word of problem data for identification.Entity recognition layer may include
Service type identification module, the identification module of knowledge based map and NLP identification module.
Intention assessment layer: for inputting the service type of problem data according at least one words recognition user.At least one
A word can be with are as follows: single entity, the corresponding attribute of entity+entity, entity+relationship, the problem of often asking data
(frequently asked questions, FAQ) matching threshold and other words.
Problem data answers layer: for determining answer according to user's input problem data and service type.Question answering layer
It may include: entity Search Module, semantic search module, figure calculating excavation module, FAQ retrieval module and full-text search mould
Block.Problem data, which answers layer, can determine that user inputs the answer of problem data by any one of the above or multiple combinations.
In a kind of mode in the cards, as shown in Figure 1, the question answering system can also include pretreatment layer, pretreatment layer
For asking that data are handled to user's input, such as check whether the problem of user inputs data have wrong word or have additional character
Deng.
In a kind of mode in the cards, as shown in Figure 1, the question answering system can also include problem data cue module.
The problem data cue module is used for: if the first problem data of user's input are imperfect or have ambiguity, sending prompt information.
In a kind of mode in the cards, as shown in Figure 1, the question answering system can also include session completion module.More
Under the scene for taking turns dialogue, session completion module is used to carry out problem data to active user's problem data based on contextual information
Completion.
It should be noted that more wheel dialogues refer in human-computer dialogue, server in preliminary clear user's input first
After the intention of problem data, necessary information is obtained in a manner of finally obtaining the first problem data of clear user's input.It is more
The applicable scene for taking turns session is user with specific purpose, it is desirable to obtain the information or clothes that meet specific restrictive condition
Business, such as: the electricity price electricity charge, troublshooting, metering device, geography information and business business etc..Since the demand of user is more multiple
It is miscellaneous, it may be necessary to which that a point more word sessions are stated that user constantly may also modify or improve the demand of oneself in dialog procedure.
In addition, server can also be by inquiring, clarifying when the demand of the first problem data of user is not specific enough or clear
Or confirmation is to help user to find satisfied result.Therefore, one decision process of more wheel dialogue, server in dialog procedure not
Be completely cured should be taken in next step according to current state decision optimal movement (such as: provide as a result, inquire specific restrictive condition, it is clear
Clear or confirmation demand etc.), so that most effective auxiliary user completes the task of information or service acquisition.
In a kind of mode in the cards, as shown in Figure 1, the question answering system can also include that visualization result shows mould
Block, the answer for user to be inputted first problem data show user.
In a kind of mode in the cards, as shown in Figure 1, the question answering system can also include problem data recommending module,
For may interested problem data to user recommended user.
In a kind of mode in the cards, as shown in Figure 1, the question answering system can also include user feedback module, it be used for
Receive the use opinion of user.For example, when user is by the input of user feedback module " being unable to get oneself desired answer ", then
System can recorde the opinion and the opinion notified system manager.
In a kind of mode in the cards, as shown in Figure 1, the question answering system can also include user journal module, it be used for
Record the answer of the historical problem data and/or historical problem data of user.
Below in conjunction with Fig. 2 to Fig. 9 to a kind of determination method progress of problem data answer provided by the embodiments of the present application
It is specifically described.
It should be pointed out that mutually can use for reference or refer between each embodiment of the application, for example, the same or similar step
Suddenly, between embodiment of the method and Installation practice, it can mutually refer to, not limit.
As shown in Fig. 2, Fig. 2 shows a kind of determination method of problem data answer provided by the embodiments of the present application, the party
The executing subject of method can may be terminal for the executing subject of server or this method.
It should be noted that server can be property server, it can also be virtual server, such as Cloud Server.
The equipment that terminal in the embodiment of the present application can be to provide problem data.For example, the terminal can be user hand
Communication equipment is held, for example, mobile phone, tablet computer, wearable device etc..
Following embodiments are described so that executing subject is server as an example, and this method may include:
S101, server obtain the first problem data of user's input.
Wherein, first problem data include at least one word.
In the embodiment of the present application, server can obtain the first problem data of user's input by the input layer in Fig. 1.
It should be noted that first problem data are referred to as problem data.Spy illustrates herein.
Wherein, the first problem data that server receives can be the forms such as voice, text.
S102, server determine the word relationship at least one word between each word.
In the embodiment of the present application, server can be determined in first problem data extremely by the Entity recognition layer in Fig. 1
The word relationship of each word in a few word and at least one word.
Wherein, word relationship may include subordinate relation, hyponymy etc..
Illustratively, the relationship between " electronic equipment " and " mobile phone " is subordinate relation, namely " mobile phone " belongs to that " electronics is set
It is standby ".Relationship between " mobile phone " and " A mobile phone " is hyponymy, namely " mobile phone " is upper word, and " A mobile phone " is bottom
Word.
S103, according at least one word, at least one word between each word word relationship and be intended to map
Determine the corresponding service type of first problem data.
Wherein, it is intended that map includes at least the service type of at least one word association.
In a kind of possible implementation, server can be according to each word at least one word and at least one word
Word relationship between language determines the corresponding service type of first problem data by the intention assessment layer in Fig. 1.
In alternatively possible implementation, server can determine service type according to the type of first problem data.
Wherein, the type of first problem data includes that figure calculates any of type, measurement type, comparative type.The embodiment of the present application
In, server can determine the service type of problem data by the way of template matching, for example, occurring in first problem data
" counting by XXX " then hits statistical fractals, " relationship of XXX and XXX " occurs and then hits figure calculating service.
S104, server, which are preset according to the corresponding service type of first problem data from least one, determines first in answer
The answer of problem data.
In a kind of possible implementation, server can be configured at least one default answer.
In a kind of possible implementation, server can obtain at least one default answer by network.
In the embodiment of the present application, server can be pre- from least one according to the corresponding service type of first problem data
If determining the answer of the first problem data in answer.
The processing method of problem data provided by the embodiments of the present application, the first problem data by identification user's input are true
The word relationship of each word at least one fixed word and at least one word, so as to accurately parse first problem number
According to;By determining the corresponding service type of first problem data, search range is reduced, can quickly determine first problem data
Answer, therefore can satisfy user demand, and reduce cost of labor.
As a kind of possible implementation, as shown in figure 3, S102 can specifically pass through S301- in the embodiment of the present application
S302 is realized:
S301, server determine at least one word of first problem data according to knowledge mapping.
Knowledge mapping includes the corresponding category of each entity in relationship and multiple entities between multiple entities, multiple entities
Property and attribute value.
Wherein, at least one word may include entity, concept, attribute, attribute value etc..
It should be noted that a word corresponds to an entity or a concept or a category in the embodiment of the present application
Property or an attribute value etc..
By taking the knowledge mapping of " price of A mobile phone and B mobile phone " as an example, in the knowledge mapping, entity or concept are A mobile phone, B
Mobile phone, relationship be and, attribute is price, and attribute value is height.Then word can for A mobile phone, B mobile phone, price and.
In a kind of possible implementation, server can divide first problem data according to Forward Maximum Method
Word.
It should be noted that first problem data are exactly separated by Forward Maximum Method, at least one word is obtained,
The length wherein separated is preset value;At least one word is matched with the word of knowledge mapping or dictionary from left to right,
Next round matching is carried out if successful match, until at least one word processing finishes.If a word match is unsuccessful,
Word is then removed into a word from end, then is matched.
Illustratively, with first problem data be " how is iPhone cruising ability? " for, then server is according to just
The result segmented to maximum matching be " apple, mobile phone, continuation of the journey, ability, how,? ".
In a kind of possible implementation, server can divide first problem data according to reverse maximum matching
Word.
It should be noted that the difference of reverse maximum matching and Forward Maximum Method is: by least one word from the right side
It is matched to a left side with the word of knowledge mapping or dictionary.Remaining matching process is consistent with Forward Maximum Method, no longer superfluous herein
It states.
Illustratively, with first problem data be " how is iPhone cruising ability? " for, then server is according to inverse
The result segmented to maximum matching be " iPhone, cruising ability, how,? ".
In a kind of possible implementation, server can be right according to the matched method of two-way maximum and knowledge mapping
First problem data participle, obtains at least one word of first problem data.
Specifically, two-way maximum matching method is the word segmentation result for obtaining Forward Maximum Method method and reverse maximum matching method
Obtained result is compared.The more the better according to bulky grain degree word, the more fewer better principle of non-dictionary word and monosyllabic word is chosen
The good result output of one of which participle effect.Wherein, granularity indicates the quantity of word in a word.
Illustratively, with first problem data be " how is iPhone cruising ability? " for, server is according to forward direction
The result that is segmented of maximum matching be " apple, mobile phone, continuation of the journey, ability, how ";Server is according to reverse maximum matching point
Word result is " iPhone, cruising ability, how ".According to the two-way matched principle of maximum, reverse maximum matching knot is chosen
Fruit is word segmentation result.
In a kind of possible implementation, server can carry out word segmentation result according to didactic rule further
It disambiguates, and then determines at least one word of first problem data.
Does is illustratively, it that ' how is iPhone cruising ability with first problem data? ' for, then server can be known
Not Chu the entity of the first problem data in knowledge mapping be ' iPhone ', entity attribute name ' cruising ability '.
It should be noted that in the embodiment of the present application, heuristic rule may include:
If A, the word segmentation result of Forward Maximum Method method is different from the word number of the word segmentation result of reverse maximum matching method,
Take the participle word segmentation result that quantity is few and/or individual character is few.
If B, the word number of the word segmentation result of the word segmentation result of Forward Maximum Method method and reverse maximum matching method is segmented and is tied
Fruit word number is identical, just illustrates no ambiguity, can return to any one.
S302, at least one word is determined according to natural language processing (natural language processing, NLP)
Word relationship in language between each word.
In the embodiment of the present application, server can determine in S302 word between at least one word by the identification of NLP
Relationship, and then solve the first problem data of the ambiguity of different terms combination.
It should be noted that only include a word in obtained word segmentation result when server is by above-mentioned segmenting method,
Then server can determine answer relevant to a word according to a word.
Illustratively, by taking word segmentation result only includes one, Beijing word as an example, then answer can be introduced for Pekinese, such as:
Beijing is the capital of China and/or Pekinese's population is the answer relevant with Beijing such as 21,710,000 people.
In a kind of possible implementation, server can by the syntactic structure of each word of interdependent syntactic analysis (such as:
" Subject, Predicate and Object ", " determine shape benefit "), determine the word relationship in first problem data between at least one word, and then determine this
Word relationship between the word of one problem data.
It should be noted that syntactic structure includes word relationship or the dependence between word and word
(dependency relations).One of word relationship includes a core word (head) and a qualifier
(dependant)。
In a kind of possible implementation, server can be based on condition random field (conditional random
Field, CRF) Model sequence mark method, carry out first problem data interdependent syntactic analysis.
It should be noted that CRF model is common model in sequence labelling scene, compare hidden Markov model
(hidden markov model, HMM) can utilize more features, than maximum entropy Markov model (maximum
Entropy markov model, MEMM) more resistant against sequence labelling bias the problem of.
It should be noted that including three features: starting point, terminal, relationship name according to the dependence that CRF model determines
Claim.Wherein, dependence may include two factors: direction and distance.Therefore we are by sequence labelling is defined as: [+| -]
dPOS。
Wherein, [+| -] indicate direction ,+indicate that position of the governing word in sentence appears in behind dependent;Indicate branch
It is appeared in front of dependent with word;POS indicates the part of speech classification that governing word has;D indicates distance.Wherein, governing word dominates
Dependent.
It should be noted that d refers to the word that d-th is POS with identical part of speech since some direction.
Illustratively, by taking " latest version iPhone was issued and begun to sell at September 16th " as an example, server can be according to above-mentioned
Segmenting method determine word segmentation result are as follows: ' newest (adjective), iPhone (noun), in (preposition), 16 (time of September
Word), publication (verb), simultaneously (preposition), begin to sell (verb).Wherein dependent are as follows: iPhone (noun) ', governing word are as follows: publication
(verb).The 1st word (beginning to sell) is identical as governing word (publication) part of speech since reverse, is all verb, therefore d is 1.
In the embodiment of the present application, server can determine multiple feature templates of CRF model.
It should be noted that feature templates are used to determine that the training of the state characteristic function and CRF model of CRF model to be special
Sign carries out interdependent syntactic analysis at least one word to keep CRF model more accurate, improves the standard of NLP identification
True rate.
It illustratively, as shown in table 1, is multiple feature templates of the CRF model determined in the embodiment of the present application in table 1.
1 feature templates of table
Serial number | Feature templates |
U00 | %x [- 2,0] |
U01 | %x [- 1,0] |
U02 | %x [0,0] |
U03 | %x [1,0] |
U04 | %x [2,0] |
U05 | %x [- 2,0]/%x [- 1,0]/%x [0,0] |
U06 | %x [- 1,0]/%x [0,0]/%x [1,0] |
U07 | %x [0,0]/%x [1,0]/%x [2,0] |
U08 | %x [- 1,0]/%x [0,0] |
U09 | %x [0,0]/%x [1,0] |
Illustratively, by taking first problem data are " Chinese packaging nets " as an example, the word segmentation result of the first problem data is
" in ", " state ", " packet ", " dress ", " net ", then this 5 subscripts for segmenting corresponding CRF models are respectively as follows: -2, -1,0,1,2, with
For " packet ", then according to U0--U4 feature templates: indicate the relationship between some position and the information of current location, such as
U00:%x [- 2,0], just refer to " in " and " packet " between connection.
U5--U7 feature templates: indicating the relationship between certain three position and the information of current location, such as U05:%x
[- 2,0]/%x [- 1,0]/%x [0,0], just refer to " in ", " state ", the connection between " packet " and " packet ".
U8--U9 feature templates: indicating the relationship between certain two position and the information of current location, such as U08:%x
[- 1,0]/%x [0,0], just refers to the connection between " state ", " packet " and " packet ".
As a kind of possible implementation, as shown in figure 4, S103 can be especially by S401- in the embodiment of the present application
S402 is realized:
S401, server determine first from least one default intention map according at least one word and word relationship
The corresponding intention map of problem data.
In the embodiment of the present application, server can carry out first problem data according at least one word of above-mentioned determination
It was found that understanding, and then the corresponding intention map of first problem data is determined from least one default intention map.
There is at least one default to be intended to map in a kind of possible implementation, in server.First problem data pair
The intention map answered belongs at least one default intention map.
In a kind of possible implementation, server can obtain at least one by network and default be intended to map.
It should be noted that being intended to map for reflecting entity in knowledge mapping, concept, attribute, attribute value and service class
Association between type.
In a kind of possible implementation, in the embodiment of the present application, server can be according to the level-one concept of word and two
Grade concept building is intended to map.Wherein, level-one concept includes: entity type, attribute, attribute value, four class of service type.Second level is general
Thought includes: text attribute value, integer attribute value, floating number attribute value and time attribute value.
Illustratively, as shown in figure 5, Fig. 5 shows a kind of intention map of the embodiment of the present application.
It should be noted that subclass sub Class Of describes the hyponymy between entity type in Fig. 5;Some categories
Property has Property describes the attribute that entity type possesses;Value values describes the attribute value that attribute is possessed;Triggering
Triggers describes the service type that entity type may trigger.
Illustratively, the consciousness map of Fig. 5 may include comprising one or more service types.One or more service class
Type is connected with entity type top mode, and service type can be triggered by representing all types of entities in intention map.
In a kind of possible implementation, server can be according at least one word in first problem data and at least one
A default similarity for being intended to each preset intention map in map determines the corresponding intention map of first problem data.
For example, server can be by least one word in first problem data and phase at least one default intention map
It is determined as the corresponding intention map of first problem data like highest default intention map is spent.
Illustratively, the meaning for being intended to map and iPhone 6 and constituting that a kind of type of cell phone is honor 8 is shown in Fig. 6
Figure map.
Illustratively, it by taking the intention map of type of cell phone in Fig. 6 as an example, is defined as in the intention map with text attribute
' buyer's guide ', integer attribute are defined as ' commodity price ', floating number attribute definition is ' size ', time attribute is defined as
' date of production '.Preset intention map may include being intended to map 1: true service, honor 8, commodity price, size and
The date of production;It is intended to map 2: true service, iPhone 6, commodity price, size and the date of production.Wherein, it is intended that figure
Spectrum 1 and intention map 2 include that common service type compares service.
By taking first problem data are the price of mobile phone honor 8 as an example, then server can determine the first problem data pair
That answers is intended to the intention map that map is honor 8.
By first problem data be mobile phone honor 8 than apple 6 it is cheap how many for, then server can determine that this first is asked
Inscribe the corresponding intention map of data are as follows: the intention map for being intended to map and iPhone 6 and constituting of honor 8.
S402, server determine the corresponding service type of first problem data according to the service type for being intended to map.
Wherein, service type include the fact that service, calculation type service, statistical fractals or compare service any of.
In the embodiment of the present application, server can determine that first problem data are corresponding according to the service type being intended in map
Service type.
It should be noted that with calculation type service and/or compared with define on the side that is connected of service " physical quantities " this
A attribute.Such as, attribute value 2 indicate that when including two entities in first problem data be that can trigger the service.And statistical
Service does not need to define " physical quantities " this attribute, is that a statistical is asked as " electric fan sells out how many today "
Data are inscribed, statistical problem data does not need to determine the number of entity in first problem data.
In a kind of possible implementation, server can be intended to map according to the multiple combinations of above-mentioned multiple words
Route matching is carried out in spectrum.
In a kind of possible implementation, the shortest path first that server can be constrained according to conditional is calculated multiple
Shortest path in path.Wherein, constraint condition are as follows: limitation side type is that subClassOf+ service type-concept+word closes
The set of system.
Specifically, server checks the path recognized, determines the path if including service type in path
Whether satisfaction triggers the condition serviced (such as: path must include two entities).If being unsatisfactory for the condition of triggering service, service
Device can guide user to input, until first problem data meet the condition of triggering service.
It should be noted that for different service types, shortest path can be calculated by different servers, then it can be with
It obtains any one or more in numerical result, single knowledge card, list of entities, subgraph, chart and multielement.
It should be noted that if not including service type in path, such as " state's netter takes telephone number ", then the fact is carried out
It answers.
In a kind of possible implementation, S104 can specifically include S1041-S1044:
If S1041, the corresponding service type of first problem data are true type service, server is looked into according to knowledge mapping
It askes, reasoning determines answer.
It should be noted that true type service is the answer to true type problem data.When first problem data are corresponding
When service type is true type service, server can be according to knowledge base question and answer (knowledge based question
Answering, KBQA), problem data is determined in analysis, by carrying out semantic understanding and parsing to first problem data, and then is utilized
Knowledge mapping is inquired, reasoning obtains answer.
In a kind of mode in the cards, server can by inquiring structuring method (query construction) come
It is answered.Inquiring structuring method can be divided into based on template, be based on information obtained in problem data analysis, based on machine learning
Query statement is constructed with based on semantic information.
Wherein, establishing inquiry based on template is to set up to realize designed query template, wherein including some empty slots, is needed
A complete inquiry is formed after relevant information is inserted.There are also work simultaneously is asked by the method foundation of machine learning
Inscribe the mapping relations between data and query statement.
It should be noted that true four class problem data of type service support, true sex service is looked into the embodiment of the present application
Inquiry method includes object query, entity attribute inquiry, (needs that multi-hop inquiry refers to are by repeatedly looking into for the multi-hop inquiry of entity attribute
Ask can just obtain result) and more attribute conditions object query.For the inquiring structuring of these four types of problem datas, the application is real
The query language that example defines knowledge based map is applied, is handled, is obtained according to the result of first problem data intention assessment
Final query statement.
Illustratively, as shown in table 2, it indicates to output the corresponding query language of multiple queries method.
2 query language of table
If S1042, the corresponding service type of first problem data are to compare type problem service, server is by first problem
Data are divided into the subproblem data of multiple facts, determine answer according to the subproblem of multiple facts.
It should be noted that comparing type service is expanded the function of true type service, on the basis of true sex service,
It further increases and compares type service.
It illustratively, is that " family's grid company is more than Southern Power Grid Company covering province " state is with first problem data
Example.In the stage of answer, server can be divided the problem data are as follows: the covering province of State Grid Corporation of China and south electric network are public
The subproblem data of department's covering two, province service type.Server can retrieve answering for the subproblem of the two facts respectively
Case.Final server compares the answer of the subproblem of the two facts, to generate final result.
If S1043, the corresponding service type of first problem data be statistical question answering, server by inquiry from
Answer is determined in multiple default answers.
It should be noted that the answer of statistical service is, the application reality for statistical analysis to the knowledge in knowledge mapping
It applies in example, server can reorganize knowledge data, be stored in a manner of relation data, and relation data is utilized
The statistical function that library provides carries out the answer of statistical problem.When the service type of first problem data is that statistical services,
Server needs carry out knowledge mapping data to extract-transposition-load (extract-transform-load, ETL) work in advance
Make.
Specifically, carrying out cutting by entity type to knowledge mapping, each type constructs an entity table, by the concept
The attribute possessed is mapped as the field in table, additionally adds entity name, entity synonym, entitative concept field.
Exemplary property, as shown in table 3, table 3 shows a kind of entity table of statistical problem data.
The entity table of the statistical problem data of table 3
It is the answer of statistical service for first problem data in the embodiment of the present application, server can pass through inquiry
The mode of building carries out.Pass through building structured query language (structured query language, SQL) form
Query language carries out question answering, is returned with specific value or statistics tabular form.
In a kind of possible implementation, server, which can be used, calculates COUNT statistical function.
Illustratively, table 4 shows the query language of various problems data and each problem data of various problems data.
4 query language of table
If the service type of S1044, first problem data is calculation type service, server is obtained by the method that figure calculates
To answer.
The method calculated by figure is needed to obtain answer it should be noted that the problem of calculation type services data refer to.
Illustratively by taking the relationship such as Tianjin electric business company and Beijing electric business company as an example.
It should be noted that due to current data query language can not coverage diagram calculate application scenarios, so can not lead to
The method for crossing inquiring structuring obtains the answer of problems data, and therefore, in the embodiment of the present application, server can be counted by figure
The method of calculation obtains answer, and user is returned in a manner of subgraph.
It should be noted that figure calculating is the abstract table of one kind " figure " structure to real world based on " graph theory "
It reaches, and the calculating mode in this data structure.In general, basic data structure can be in figure calculates are as follows: G=(V,
E, D) V=vertex (vertex or node) side E=edge) D=data (weight).
As shown in fig. 7, Fig. 7, which shows a kind of figure, calculates graph data structure in service.Wherein, V1、V2、V3、V4、V5For section
Point, connecting line indicate side, digital representation weight.For example, V1And V2Between connecting line indicate side, 8 indicate the two nodes power
Weight.
It should be noted that graph data structure expresses the relevance between data well, therefore, go out in many applications
Existing problem data can be abstracted into figure to indicate, establish model by the thought of graph theory or based on scheming to solve the problems, such as
Data.
It should be noted that figure calculate for solve on Large Scale Graphs how under the premise of being computed correctly guarantee calculate
Efficiency.
In a kind of mode in the cards, in the embodiment of the present application, server can be counted by using Spark GraphX
It calculates frame and carries out figure calculating service.
It should be noted that Spark GraphX Computational frame using point segmentation by the way of carry out large-scale graph data
Distributed storage, i.e., each node storage is primary, but the side having can be interrupted and assign on two servers, therefore can save
Memory space.The nomography kit that the embodiment of the present application can be provided by GraphX, realizes page-ranking PageRank, number
The nomographys such as triangle, largest connected figure and shortest path.
Illustratively, a simple shortest path calculation code is as follows:
In a kind of possible implementation, as shown in figure 8, this method can also include: S501- in the embodiment of the present application
S504:
S501, server obtain the Second Problem data of user's input.
S502, in the case where determining Second Problem data and first problem data correlation, server extracts at least one word
The first word in language.
In the embodiment of the present application, server can realize more wheels in such a way that session completion module and more wheels are intended to splicing
Session.Wherein, more wheels are intended to splicing and carry out intention assessment to first problem data by single-wheel session, by the intention recognized with
At least one words of first problem data carries out record storage, when getting Second Problem associated with first problem data
When data, Second Problem data can be updated according to by least one word of first problem data, or according to Second Problem number
According at least one word update first problem data.
In the embodiment of the present application, if after user inputs the first data problem, having input second again within a preset time and asking
Data are inscribed, if it is pronoun that Second Problem data, which do not have specific subject or the subject of Second Problem data, due at this time
Server not can determine that the purpose of Second Problem problem, therefore can extract in first problem data subject as Second Problem
The subject of data, and update Second Problem data.
It illustratively, be " how much electric one evening of Gree can save ", Second Problem data with first problem data is " it
For price ".It include its this pronoun in Second Problem data, " it " reference is indefinite, therefore server can determine second
Problem data is associated with first problem data, then server can replace the subject " Gree " in first problem data
The pronoun " it " of Second Problem data obtains updated Second Problem data: the price of Gree.
It illustratively, be " how much electric one evening of Gree can save ", Second Problem data with first problem data is " so
Set-up time? " for, server determines that Second Problem data include session mark words " so ", " ", and Second Problem number
According to no subject, then the subject Gree in first problem data can be added to Second Problem data by server, be obtained
One updated Second Problem data, namely " set-up time of Gree? ".
In a kind of possible implementation, Second Problem data belong to first problem data, then server can be by first
At least one word and Second Problem data of problem data, the problem data after obtaining a supplement, according to asking after supplement
Topic data determine answer.
Illustratively, by taking the electricity price that first problem data are China is how many as an example, since server possibly can not determine
Answer, alternatively, the answer that server provides is excessive.Server can input Second Problem data by session completion module.When
The Second Problem data that server is got are Beijing, since Beijing belongs to China, server can determine Second Problem
Data belong to first problem data." China, electricity price " in first problem data can be added to Second Problem by server
In data, a updated Second Problem data " electricity price of BeiJing, China " is obtained, server can be according to updated
Two problem datas determine answer.
In a kind of possible implementation, if at least one word in Second Problem data be unsatisfactory for it is primary correctly
KBQA or the condition of triggering service, such as " set-up time? ", then server can be empty by the subject Gree in first problem data
Tune adds to Second Problem data, obtains a updated Second Problem data, the namely " set-up time of Gree
? ".
It should be noted that if Second Problem data are unrelated to first problem data, then server can be by second
Problem data determines answer in such a way that single-wheel parses.
S503, server update Second Problem data according to the first word.
In Second Problem data situation associated with first problem data, server can will be in first problem data
The first word update to Second Problem data.
Illustratively, if the subject of first problem data is indefinite, or lack subject, server can be by first problem number
According to subject add to Second Problem data.
S504, server determine the answer of Second Problem data according to the Second Problem data of update.
In the case where Second Problem data update, server can be answered according to the determination of updated Second Problem data
Case.Server determines that updated Second Problem data answer scoring method can be with reference to the above method, and details are not described herein again.
In the embodiment of the present application, more wheel sessions are carried out based on being intended to, it is more more acurrate than simply splicing, also broadly know
Whether other Second Problem data recursively realize more wheel intents, avoid splicing still in the intention of first problem data
It is too long, splice excessively rough bring and accidentally recalls and owe to recall.
In a kind of possible implementation, as shown in figure 9, this method can also include: S601- in the embodiment of the present application
S602:
S601, server calculate the similarity of first problem data and the first historical problem data.
In a kind of mode in the cards, server can calculate first problem data and the first history with collaborative filtering
The similarity of problem data.
It illustratively, is " price of Gree " with first problem data, the first historical problem is that " how much is Gree
Money "." Gree " is identical as " Gree " in the first historical problem in first problem data, " valence in first problem data
Lattice " are similar to " how much " in the first historical problem.
In a kind of mode in the cards, which can be vector similarity.
If S602, similarity are greater than or equal to threshold value, server determines the answer of the second historical problem data.
Wherein, the second historical problem data and the first historical problem data at least one word having the same.
If the similarity of problem data and the first historical problem data is greater than threshold value, server can be from multiple historical problems
At least one second historical problem data is determined in data.
Illustratively, similarity can be 50%.By taking first problem data are the prices of Gree as an example, the first history
Problem data be Gree how much.First problem data and the first historical problem data similarity are more than 50%, then service
Device can according to the word " Gree " in the first historical problem data, determine in historical problem data " Gree
Guarantee period is how long ", the historical problems data such as " specification of Gree " be the second historical problem data.
The embodiment of the present application can carry out functional module according to determining device of the above method example to problem data answer
Perhaps the division of functional unit be for example, each functional module of each function division or functional unit can be corresponded to, can also will
Two or more functions are integrated in a processing module.Above-mentioned integrated module both can take the form of hardware reality
It is existing, it can also be realized in the form of software function module or functional unit.Wherein, in the embodiment of the present application to module or
The division of unit is schematically that only a kind of logical function partition, there may be another division manner in actual implementation.
The embodiment of the present application provides a kind of determining device of problem data answer, which can be server, can also
Think the chip applied to server, as shown in Figure 10, which may include communication unit 111, determination unit 112.
Communication unit 111, for obtaining the first problem data of user's input;First problem data include at least one word
Language;
Determination unit 112, for determining the word relationship at least one word between each word;
Determination unit 112 is also used to determine first problem data according at least one word, word relationship and intention map
Corresponding service type;Wherein, it is intended that map includes at least the service type of at least one word association;
Determination unit 112 is also used to be preset in answer according to the corresponding service type of first problem data from least one
Determine the answer of first problem data.
Optionally, determination unit 112 are specifically used for:
At least one word of first problem data is determined according to knowledge mapping;Knowledge mapping includes multiple entities, multiple
The corresponding attribute of each entity and attribute value in relationship and multiple entities between entity;
The word relationship at least one word between each word is determined according to natural language processing NLP.
Optionally, determination unit 112, also particularly useful for:
It is default from least one according to the word relationship at least one word and at least one word between each word
It is intended to determine the corresponding intention map of first problem data in map;First problem number is determined according to the service type for being intended to map
According to corresponding service type;Wherein, service type includes the fact that service, figure calculate service, statistical fractals or compare in service
Any one.
Optionally, communication unit 111 are also used to obtain the Second Problem data of user's input;
Determination unit 112 is extracted at least in the case where being also used to determine Second Problem data and first problem data correlation
The first word in one word;
Determination unit 112 is also used to update Second Problem data according to the first word;
Determination unit 112 is also used to determine the answer of Second Problem data according to the Second Problem data of update.
Optionally, as shown in figure 11, which further includes computing unit 113;
Computing unit 113, for calculating the similarity of first problem data Yu the first historical problem data;
Determination unit 112, if being also used to similarity more than or equal to threshold value, it is determined that the second historical problem data are answered
Case;Wherein, the second historical problem data and the first historical problem data at least one word having the same.
Figure 12 shows another possible structural representation of the determining device of involved problem data in above-described embodiment
Figure.When the device is server, which includes: one or more processors 161 and communication interface 162.Processor 161 is used
Control management is carried out in the movement to device, for example, executing the step of above-mentioned determination unit 112 executes, and/or for executing sheet
Other processes of technology described in text.
In the concrete realization, as one embodiment, processor 161 may include one or more CPU, such as in Figure 12
CPU0 and CPU1.
In the concrete realization, as one embodiment, communication equipment may include multiple processors, such as the place in Figure 12
Manage device 111.Each of these processors can be monokaryon (single-CPU) processor, be also possible to a multicore
(multi-CPU) processor.Here processor can refer to one or more equipment, circuit, and/or for handling problem data
The processing core of (such as computer program instructions).
Optionally, which can also include memory 163 and communication line 164, and memory 163 is for storage device
Program code and problem data.
Figure 13 is the structural schematic diagram of chip 170 provided by the embodiments of the present application.Chip 170 includes one or more
(including two) processor 1710 and communication interface 1730.
Optionally, which further includes memory 1740, and memory 1740 may include read-only memory and deposit at random
Access to memory, and operational order and problem data are provided to processor 1710.The a part of of memory 1740 can also include non-
Volatile random access memory (non-volatile random access memory, NVRAM).
In some embodiments, memory 1740 stores following element, execution module or problem data structure,
Perhaps their subset or their superset.
In the embodiment of the present application, by calling the operational order of the storage of memory 1740, (operational order is storable in
In operating system), execute corresponding operation.
Wherein, above-mentioned processor 1710 may be implemented or execute various exemplary in conjunction with described in present disclosure
Logic block, unit and circuit.The processor can be central processing unit, general processor, and digital signal processor is dedicated
Integrated circuit, field programmable gate array or other programmable logic device, transistor logic, hardware component or its
Any combination.It, which may be implemented or executes, combines various illustrative logic blocks described in present disclosure, unit
And circuit.The processor is also possible to realize the combination of computing function, such as combines comprising one or more microprocessors,
DSP and the combination of microprocessor etc..
Memory 1740 may include volatile memory, such as random access memory;The memory also may include
Nonvolatile memory, such as read-only memory, flash memory, hard disk or solid state hard disk;The memory can also include upper
State the combination of the memory of type.
Bus 1720 can be expanding the industrial standard structure (Extended Industry Standard
Architecture, EISA) bus etc..Bus 1720 can be divided into address bus, problem data bus, control bus etc..For
Convenient for indicating, only indicated with a line in Figure 13, it is not intended that an only bus or a type of bus.
Through the above description of the embodiments, it is apparent to those skilled in the art that, for description
It is convenienct and succinct, only with the division progress of above-mentioned each functional unit for example, in practical application, can according to need and will be upper
It states function distribution to be completed by different functional units, i.e., the internal structure of device is divided into different functional units, to complete
All or part of function described above.The specific work process of the system, apparatus, and unit of foregoing description, before can referring to
The corresponding process in embodiment of the method is stated, details are not described herein.
The embodiment of the present application also provides a kind of computer readable storage medium, and finger is stored in computer readable storage medium
It enables, when computer executes the instruction, which executes each step in method flow shown in above method embodiment.
Wherein, computer readable storage medium, such as electricity, magnetic, optical, electromagnetic, infrared ray can be but not limited to or partly led
System, device or the device of body, or any above combination.The more specific example of computer readable storage medium is (non-poor
The list of act) it include: the electrical connection with one or more conducting wires, portable computer diskette, hard disk.Random access memory
(Random Access Memory, RAM), read-only memory (Read-Only Memory, ROM), erasable type may be programmed read-only
It is memory (Erasable Programmable Read Only Memory, EPROM), register, hard disk, optical fiber, portable
Compact disc read-only memory (Compact Disc Read-Only Memory, CD-ROM), light storage device, magnetic memory
Part or above-mentioned people are in appropriate combination or the computer readable storage medium of any other form of this field numerical value.
A kind of illustrative storage medium is coupled to processor, to enable a processor to from the read information, and can be to
Information is written in the storage medium.Certainly, storage medium is also possible to the component part of processor.Pocessor and storage media can be with
In application-specific IC (Application Specific Integrated Circuit, ASIC).In the application
In embodiment, computer readable storage medium can be any tangible medium for including or store program, which can be referred to
Enable execution system, device or device use or in connection.
The embodiment of the present invention provides a kind of computer program product comprising instruction, when instruction is run on computers
When, so that computer executes the determination method of data answer the problem of as described in Fig. 2, Fig. 3, Fig. 4, Fig. 8, Fig. 9.
Due to the determining device of the problems in the embodiment of the present invention data, computer readable storage medium, computer journey
Sequence product can be applied to the above method, therefore, can be obtained technical effect see also above method embodiment, this hair
Details are not described herein for bright embodiment.
In several embodiments provided herein, it should be understood that disclosed system, apparatus and method, it can be with
It realizes by another way.For example, apparatus embodiments described above are merely indicative, for example, the unit
It divides, only a kind of logical function partition, there may be another division manner in actual implementation, such as multiple units or components
It can be combined or can be integrated into another system, or some features can be ignored or not executed.Another point, it is shown or
The mutual coupling, direct-coupling or communication connection discussed can be through some interfaces, the indirect coupling of equipment or unit
It closes or communicates to connect, can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, each functional unit in each embodiment of the application can integrate in one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.
The above, the only specific embodiment of the application, but the protection scope of the application is not limited thereto, it is any
Those familiar with the art within the technical scope of the present application, can easily think of the change or the replacement, and should all contain
Lid is within the scope of protection of this application.Therefore, the protection scope of the application should be based on the protection scope of the described claims.
Claims (12)
1. a kind of determination method of problem data answer characterized by comprising
Obtain the first problem data of user's input;The first problem data include at least one word;
Determine the word relationship at least one described word between each word;
The corresponding service of the first problem data is determined according at least one described word, the word relationship and intention map
Type;Wherein, the service type for being intended to map and including at least at least one word association;
It is preset according to the corresponding service type of the first problem data from least one and determines the first problem number in answer
According to answer.
2. the method according to claim 1, wherein at least one word of the determination between each word
Word relationship, comprising:
At least one described word is determined from the first problem data according to knowledge mapping;The knowledge mapping includes multiple
The corresponding attribute of each entity and attribute value in relationship and the multiple entity between entity, the multiple entity;
The word relationship at least one described word between each word is determined according to natural language processing NLP.
3. method according to claim 1 or 2, which is characterized in that described at least one word according to, the word
Relationship and intention map determine the corresponding service type of the first problem data, comprising:
Determine that described first asks from least one default intention map according at least one described word and the word relationship
Inscribe the corresponding intention map of data;
The corresponding service type for being intended to map of the first problem data is determined as the corresponding institute of the first problem data
State service type;Wherein, the service type includes the fact that service, figure calculate service, statistical fractals, compare any in servicing
It is a.
4. method according to claim 1 or 2, which is characterized in that the method also includes:
Obtain the Second Problem data of user's input;
In the case where determining Second Problem data and the first problem data correlation, at least one described word is extracted
One word;
The Second Problem data are updated according to first word;
The answer of the Second Problem data is determined according to the Second Problem data of update.
5. method according to claim 1 or 2, which is characterized in that the method also includes:
Calculate the similarity of the first problem data Yu the first historical problem data;
If the similarity is greater than or equal to threshold value, it is determined that the answer of the second historical problem data;Wherein, second history
Word having the same between problem data and the first historical problem data.
6. a kind of determining device of problem data answer characterized by comprising communication unit, determination unit;
The communication unit, for obtaining the first problem data of user's input;The first problem data include at least one
Word;
The determination unit, for determining the word relationship at least one described word between each word;
The determination unit is also used to according at least one described word, the word relationship and is intended to map and determines described the
The corresponding service type of one problem data;Wherein, the service for being intended to map and including at least at least one word association
Type;
The determination unit is also used to be preset in answer according to the corresponding service type of the first problem data from least one
Determine the answer of the first problem data.
7. device according to claim 6, which is characterized in that the determination unit is specifically used for:
At least one described word is determined from the first problem data according to knowledge mapping;The knowledge mapping includes multiple
The corresponding attribute of each entity and attribute value in relationship and the multiple entity between entity, the multiple entity;
The word relationship at least one described word between each word is determined according to NLP.
8. device according to claim 6 or 7, which is characterized in that the determination unit, also particularly useful for:
Determine that described first asks from least one default intention map according at least one described word and the word relationship
Inscribe the corresponding intention map of data;
The corresponding service type for being intended to map of the first problem data is determined as the corresponding institute of the first problem data
State service type;Wherein, the service type includes the fact that service, figure calculate service, statistical fractals, compare any in servicing
It is a.
9. device according to claim 6 or 7, which is characterized in that
The communication unit is also used to obtain the Second Problem data of user's input;
The determination unit extracts institute in the case where being also used to determine Second Problem data and the first problem data correlation
State the first word at least one word;
The determination unit is also used to update the Second Problem data according to first word;
The determination unit is also used to determine the answer of the Second Problem data according to the Second Problem data of update.
10. device according to claim 6 or 7, which is characterized in that described device further includes computing unit;
The computing unit, for calculating the similarity of the first problem data Yu the first historical problem data;
If the similarity is greater than or equal to threshold value, the determination unit is also used to determine the answer of the second historical problem data;
Wherein, the second historical problem data and the first historical problem data at least one word having the same.
11. a kind of readable storage medium storing program for executing, which is characterized in that instruction is stored in the readable storage medium storing program for executing, when described instruction quilt
When execution, the method as described in any one of claims 1 to 5 is realized.
12. a kind of chip, which is characterized in that the chip includes at least one processor and communication interface, the communication interface
It is coupled at least one described processor, at least one described processor is for running computer program or instruction, to realize power
Benefit require any one of 1 to 5 described in method.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910472611.9A CN110245240A (en) | 2019-05-31 | 2019-05-31 | A kind of determination method and device of problem data answer |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910472611.9A CN110245240A (en) | 2019-05-31 | 2019-05-31 | A kind of determination method and device of problem data answer |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110245240A true CN110245240A (en) | 2019-09-17 |
Family
ID=67885756
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910472611.9A Pending CN110245240A (en) | 2019-05-31 | 2019-05-31 | A kind of determination method and device of problem data answer |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110245240A (en) |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110837550A (en) * | 2019-11-11 | 2020-02-25 | 中山大学 | Knowledge graph-based question and answer method and device, electronic equipment and storage medium |
CN111241263A (en) * | 2020-04-24 | 2020-06-05 | 支付宝(杭州)信息技术有限公司 | Text generation method and device and electronic equipment |
CN111259668A (en) * | 2020-05-07 | 2020-06-09 | 腾讯科技(深圳)有限公司 | Reading task processing method, model training device and computer equipment |
CN111709235A (en) * | 2020-05-28 | 2020-09-25 | 上海发电设备成套设计研究院有限责任公司 | Text data statistical analysis system and method based on natural language processing |
CN111858877A (en) * | 2020-06-17 | 2020-10-30 | 平安科技(深圳)有限公司 | Multi-type question intelligent question answering method, system, equipment and readable storage medium |
CN112989010A (en) * | 2021-04-26 | 2021-06-18 | 广州思迈特软件有限公司 | Data query method, data query device and electronic equipment |
CN113420111A (en) * | 2021-06-17 | 2021-09-21 | 中国科学院声学研究所 | Intelligent question-answering method and device for multi-hop inference problem |
CN113434644A (en) * | 2021-05-25 | 2021-09-24 | 北京农业信息技术研究中心 | Agricultural technology knowledge service method and system |
WO2021208460A1 (en) * | 2020-04-15 | 2021-10-21 | 深圳前海微众银行股份有限公司 | Sentence completion method and device, and readable storage medium |
CN113657102A (en) * | 2021-08-17 | 2021-11-16 | 北京百度网讯科技有限公司 | Information extraction method, information extraction device, information extraction apparatus, storage medium, and program |
CN116432615A (en) * | 2023-06-12 | 2023-07-14 | 中国第一汽车股份有限公司 | Text processing method and device |
CN117851614A (en) * | 2024-03-04 | 2024-04-09 | 创意信息技术股份有限公司 | Searching method, device and system for mass data and storage medium |
CN118277588A (en) * | 2024-05-30 | 2024-07-02 | 阿里云飞天(杭州)云计算技术有限公司 | Query request processing method, electronic device and storage medium |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107918634A (en) * | 2017-06-27 | 2018-04-17 | 上海壹账通金融科技有限公司 | Intelligent answer method, apparatus and computer-readable recording medium |
CN109033223A (en) * | 2018-06-29 | 2018-12-18 | 北京百度网讯科技有限公司 | For method, apparatus, equipment and computer readable storage medium across type session |
CN109522465A (en) * | 2018-10-22 | 2019-03-26 | 国家电网公司 | The semantic searching method and device of knowledge based map |
-
2019
- 2019-05-31 CN CN201910472611.9A patent/CN110245240A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107918634A (en) * | 2017-06-27 | 2018-04-17 | 上海壹账通金融科技有限公司 | Intelligent answer method, apparatus and computer-readable recording medium |
CN109033223A (en) * | 2018-06-29 | 2018-12-18 | 北京百度网讯科技有限公司 | For method, apparatus, equipment and computer readable storage medium across type session |
CN109522465A (en) * | 2018-10-22 | 2019-03-26 | 国家电网公司 | The semantic searching method and device of knowledge based map |
Cited By (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110837550A (en) * | 2019-11-11 | 2020-02-25 | 中山大学 | Knowledge graph-based question and answer method and device, electronic equipment and storage medium |
CN110837550B (en) * | 2019-11-11 | 2023-01-17 | 中山大学 | Knowledge graph-based question answering method and device, electronic equipment and storage medium |
WO2021208460A1 (en) * | 2020-04-15 | 2021-10-21 | 深圳前海微众银行股份有限公司 | Sentence completion method and device, and readable storage medium |
CN111241263A (en) * | 2020-04-24 | 2020-06-05 | 支付宝(杭州)信息技术有限公司 | Text generation method and device and electronic equipment |
CN111259668A (en) * | 2020-05-07 | 2020-06-09 | 腾讯科技(深圳)有限公司 | Reading task processing method, model training device and computer equipment |
CN111709235A (en) * | 2020-05-28 | 2020-09-25 | 上海发电设备成套设计研究院有限责任公司 | Text data statistical analysis system and method based on natural language processing |
CN111709235B (en) * | 2020-05-28 | 2023-10-20 | 上海发电设备成套设计研究院有限责任公司 | Text data statistical analysis system and method based on natural language processing |
CN111858877A (en) * | 2020-06-17 | 2020-10-30 | 平安科技(深圳)有限公司 | Multi-type question intelligent question answering method, system, equipment and readable storage medium |
CN112989010A (en) * | 2021-04-26 | 2021-06-18 | 广州思迈特软件有限公司 | Data query method, data query device and electronic equipment |
CN113434644A (en) * | 2021-05-25 | 2021-09-24 | 北京农业信息技术研究中心 | Agricultural technology knowledge service method and system |
CN113434644B (en) * | 2021-05-25 | 2023-09-12 | 北京农业信息技术研究中心 | Agricultural technology knowledge service method and system |
CN113420111A (en) * | 2021-06-17 | 2021-09-21 | 中国科学院声学研究所 | Intelligent question-answering method and device for multi-hop inference problem |
CN113420111B (en) * | 2021-06-17 | 2023-08-11 | 中国科学院声学研究所 | Intelligent question answering method and device for multi-hop reasoning problem |
CN113657102A (en) * | 2021-08-17 | 2021-11-16 | 北京百度网讯科技有限公司 | Information extraction method, information extraction device, information extraction apparatus, storage medium, and program |
CN116432615A (en) * | 2023-06-12 | 2023-07-14 | 中国第一汽车股份有限公司 | Text processing method and device |
CN117851614A (en) * | 2024-03-04 | 2024-04-09 | 创意信息技术股份有限公司 | Searching method, device and system for mass data and storage medium |
CN117851614B (en) * | 2024-03-04 | 2024-05-14 | 创意信息技术股份有限公司 | Searching method, device and system for mass data and storage medium |
CN118277588A (en) * | 2024-05-30 | 2024-07-02 | 阿里云飞天(杭州)云计算技术有限公司 | Query request processing method, electronic device and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110245240A (en) | A kind of determination method and device of problem data answer | |
CN111125334B (en) | Search question-answering system based on pre-training | |
CN113704451B (en) | Power user appeal screening method and system, electronic device and storage medium | |
CN112667794A (en) | Intelligent question-answer matching method and system based on twin network BERT model | |
US12032911B2 (en) | Systems and methods for structured phrase embedding and use thereof | |
CN111797210A (en) | Information recommendation method, device and equipment based on user portrait and storage medium | |
CN104933081A (en) | Search suggestion providing method and apparatus | |
US12093648B2 (en) | Systems and methods for producing a semantic representation of a document | |
CN113127506B (en) | Target query statement construction method and device, storage medium and electronic device | |
CN110390052A (en) | Search for recommended method, the training method of CTR prediction model, device and equipment | |
CN113627797B (en) | Method, device, computer equipment and storage medium for generating staff member portrait | |
CN111274267A (en) | Database query method and device and computer readable storage medium | |
US11934392B2 (en) | Method and system for data conversations | |
CN111694940A (en) | User report generation method and terminal equipment | |
CN110990532A (en) | Method and device for processing text | |
CN112182145A (en) | Text similarity determination method, device, equipment and storage medium | |
CN116303537A (en) | Data query method and device, electronic equipment and storage medium | |
CN113821588A (en) | Text processing method and device, electronic equipment and storage medium | |
CN117743543A (en) | Sentence generation method and device based on large language model and electronic equipment | |
CN116821372A (en) | Knowledge graph-based data processing method and device, electronic equipment and medium | |
CN109710742B (en) | Method, system and equipment for processing individual stock announcement natural language query | |
CN116402166B (en) | Training method and device of prediction model, electronic equipment and storage medium | |
CN111737607B (en) | Data processing method, device, electronic equipment and storage medium | |
CN113379432A (en) | Sales system customer matching method based on machine learning | |
CN111126073B (en) | Semantic retrieval method and device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190917 |
|
RJ01 | Rejection of invention patent application after publication |