CN108717441A

CN108717441A - The determination method and device of predicate corresponding to question template

Info

Publication number: CN108717441A
Application number: CN201810468186.1A
Authority: CN
Inventors: 周辉阳; 饶孟良
Original assignee: Tencent Technology Shenzhen Co Ltd
Current assignee: Tencent Technology Shenzhen Co Ltd
Priority date: 2018-05-16
Filing date: 2018-05-16
Publication date: 2018-10-30
Anticipated expiration: 2038-05-16
Also published as: CN108717441B

Abstract

The invention discloses a kind of determination method and devices of the predicate corresponding to question template.Wherein, this method includes：It obtains the target problem template with correspondence and answers information, and obtain predicate and content information with correspondence；Each predicate answered corresponding to information is determined according to the similarity between information and the content information is each answered in the answer information；The most predicate of the corresponding quantity for answering information in the predicate is determined as the target predicate corresponding to the target problem template.Efficiency lower technical problem when the present invention solves the predicate determined in the related technology corresponding to question template.

Description

The determination method and device of predicate corresponding to question template

Technical field

The present invention relates to computer realms, in particular to a kind of determination method of the predicate corresponding to question template And device.

Background technology

The relevant technologies are typically using the method for rule when processing template is intended to, and are exactly artificial hand-written rule, system Determine mapping ruler.Such as：Artificial is much this year you, you today how old, how old are you age, you are which year birth , the problems such as your, your how old at age in this year, is mapped as " age " at much ages in this year, but the defect of this scheme is non- Chang Mingxian：It is limited to artificial experience, and can not possibly be complete.If user becomes a saying：How much is your age, if the way to put questions Not within rule, then this sentence can not be handled.Current technology problem essentially consists in artificial knowledge and ability is limited, And the way to put questions of user is multifarious, is made a lot of variety.Therefore, artificial scheme is only possible to reply partial picture, it is impossible to answer To all situations, support can not be just provided for the rule that do not summarize manually.In addition, excessive also can be manually team Increase appropriation budget, this is a kind of way got half the result with twice the effort, the effect for having spent money also not necessarily to have.

For above-mentioned problem, currently no effective solution has been proposed.

Invention content

An embodiment of the present invention provides a kind of determination method and devices of the predicate corresponding to question template, at least to solve Efficiency lower technical problem when determining the predicate corresponding to question template in the related technology.

One side according to the ... of the embodiment of the present invention provides a kind of determination method of the predicate corresponding to question template, Including：It obtains the target problem template with correspondence and answers information, and obtain with the predicate of correspondence and interior Hold information；It is determined described each time according to the similarity each answered in the answer information between information and the content information Answer the predicate corresponding to information；The most predicate of the corresponding quantity for answering information in the predicate is determined as the target problem Target predicate corresponding to template.

Another aspect according to the ... of the embodiment of the present invention, additionally provide the predicate corresponding to a kind of question template determines dress It sets, including：First acquisition module, for obtaining target problem template and answer information with correspondence, and acquisition tool There are the predicate and content information of correspondence；First determining module, for according to each answered in the answer information information and Similarity between the content information determines each predicate answered corresponding to information；Second determining module, being used for will The most predicate of the corresponding quantity for answering information is determined as the target predicate corresponding to the target problem template in the predicate.

Another aspect according to the ... of the embodiment of the present invention additionally provides a kind of storage medium, which is characterized in that the storage is situated between Computer program is stored in matter, wherein the computer program is arranged to execute described in any of the above-described when operation Method.

Another aspect according to the ... of the embodiment of the present invention additionally provides a kind of electronic device, including memory and processor, It is characterized in that, computer program is stored in the memory, and the processor is arranged to hold by the computer program Method described in row any of the above-described.

In embodiments of the present invention, it using the target problem template and answer information obtained with correspondence, and obtains Take predicate and content information with correspondence；According to each answered in the answer information information and the content information it Between similarity determine it is described it is each answer information corresponding to predicate；The corresponding quantity for answering information in the predicate is most Predicate be determined as the mode of the target predicate corresponding to the target problem template, obtain the target problem with correspondence Template and answer information and the predicate with correspondence and content information, further according to each answers information and content information it Between similarity establish each correspondence answered between information and predicate, by the most meaning of the corresponding quantity for answering information Word is determined as the corresponding position of target problem template, to by the way of ballot, be pressed automatically for each target problem template It votes predicate according to similarity, and establishes the correspondence between each target problem template and the highest predicate of poll, To be that question template determines its corresponding predicate automatically, improved when determining the predicate corresponding to question template to realize The technique effect of efficiency, and then solve efficiency lower technology when determining the predicate corresponding to question template in the related technology and ask Topic.

Description of the drawings

Attached drawing described herein is used to provide further understanding of the present invention, and is constituted part of this application, this hair Bright illustrative embodiments and their description are not constituted improper limitations of the present invention for explaining the present invention.In the accompanying drawings：

Fig. 1 is the signal of the determination method of the predicate corresponding to a kind of optional question template according to the ... of the embodiment of the present invention Figure；

Fig. 2 is the application of the determination method of the predicate corresponding to a kind of optional question template according to the ... of the embodiment of the present invention Environment schematic；

Fig. 3 is the determination side of the predicate corresponding to a kind of optional question template according to optional embodiment of the invention The schematic diagram of method；

Fig. 4 is the determination of the predicate corresponding to the optional question template of another kind according to optional embodiment of the invention The schematic diagram of method；

Fig. 5 is the signal of the determining device of the predicate corresponding to a kind of optional question template according to the ... of the embodiment of the present invention Figure；

Fig. 6 is the application of the determination method of the predicate corresponding to a kind of optional question template according to the ... of the embodiment of the present invention Schematic diagram of a scenario；And

Fig. 7 is a kind of schematic diagram of optional electronic device according to the ... of the embodiment of the present invention.

Specific implementation mode

In order to enable those skilled in the art to better understand the solution of the present invention, below in conjunction in the embodiment of the present invention Attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is only The embodiment of a part of the invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill people The every other embodiment that member is obtained without making creative work should all belong to the model that the present invention protects It encloses.

It should be noted that term " first " in description and claims of this specification and above-mentioned attached drawing, " Two " etc. be for distinguishing similar object, without being used to describe specific sequence or precedence.It should be appreciated that using in this way Data can be interchanged in the appropriate case, so as to the embodiment of the present invention described herein can in addition to illustrating herein or Sequence other than those of description is implemented.In addition, term " comprising " and " having " and their any deformation, it is intended that cover It includes to be not necessarily limited to for example, containing the process of series of steps or unit, method, system, product or equipment to cover non-exclusive Those of clearly list step or unit, but may include not listing clearly or for these processes, method, product Or the other steps or unit that equipment is intrinsic.

One side according to the ... of the embodiment of the present invention provides a kind of determination method of the predicate corresponding to question template, As shown in Figure 1, this method includes：

S102 obtains the target problem template with correspondence and answers information, and obtains with correspondence Predicate and content information；

S104 determines each answer information according to the similarity each answered in answer information between information and content information Corresponding predicate；

The most predicate of the corresponding quantity for answering information in predicate is determined as the mesh corresponding to target problem template by S106 Mark predicate.

Optionally, in the present embodiment, the determination method of the predicate corresponding to above problem template can be applied to such as Fig. 2 Shown in the hardware environment that is constituted of equipment 202.As shown in Fig. 2, equipment 202 obtains the target problem mould with correspondence Plate and answer information, and obtain predicate and content information with correspondence；Information is each answered according to answering in information Similarity between content information determines each predicate answered corresponding to information；By the corresponding quantity for answering information in predicate Most predicates is determined as the target predicate corresponding to target problem template.

Optionally, in the present embodiment, the determination method of the predicate corresponding to above problem template can be, but not limited to answer For determining between question template and predicate in the scene of correspondence.Wherein, the predicate corresponding to above problem template is really The method of determining can be, but not limited to be applied to various types of applications, for example, online education application, instant messaging application, community's sky Between apply, game application, shopping application, browser application, financial application, multimedia application, live streaming application, medical applications, with And interactive application of Intelligent hardware etc. (wherein, Intelligent hardware can be, but not limited to include：Smart home device, intelligence wearing are set Standby, intelligent transportation tool etc.).Specifically, can be, but not limited to be applied in above-mentioned browser application determine question template with Between predicate in the scene of correspondence, or can with but be not limited to be applied to above-mentioned game application determine question template with meaning Between word in the scene of correspondence, the efficiency when predicate corresponding to question template is determined to improve.Above-mentioned is only that one kind is shown , do not do any restriction to this in the present embodiment.

Optionally, in the present embodiment, the determination method of the predicate corresponding to above problem template can be, but not limited to answer For in the scene of natural language processing.Such as：Knowledge base question and answer (KnowledgeBase Question Answering, letter Referred to as KB-QA), natural language problem (Query) is given, by carrying out semantic understanding and parsing to problem, and then utilizes knowledge Library is inquired, reasoning obtains answer (Answer).

Such as：As shown in figure 3, there is the knowledge mapping of part Liu, asked now there are one user：The wife of Liu Whom is.This Query by template generation processing after template be：Whom the wife of [singer] is, if the template reflected The predicate penetrated is wife, so whom the wife of [singer], which is, corresponds to wife.Knowledge base (knowledge base, referred to as KB the entity that singer is Liu is found in), predicate is wife, and it is Zhu that can find result, finally returns to result Zhu So-and-so.

Optionally, in the present embodiment, information triple is typically expressed as in knowledge mapping：Subject (subject), predicate (predicate), object (object), subject and object are typically all entity, and what predicate indicated is the relationship between two entities Or the attribute of subject.Wherein, entity refers to the base unit for indicating a concept.Such as：From the point of view of data processing, reality Objective things in the world can be referred to as entity, it be it is any in real world distinguish, identifiable things.Entity can be with Refer to people, such as teacher, student, object, such as book, warehouse can also be referred to.It can not only refer to the objective objects that can be touched, can be with Refer to abstract event, such as performance, football match.

Optionally, in the present embodiment, template can be a kind of general clause with extension sample, and question template is exactly A kind of general problem clause with extension sample.Such as：For problem 1：Whom the wife of Zhang San is, problem 2：Li Si's is old Whom mother-in-law is, problem 3：Whom the wife of king five is, problem 4：The wife of Zhao six is for whom, is all to ask although subject is different It asks that whom whose wife is to so-and-so, the same question template is arrived it is possible to which problem 1 to problem 4 is concluded：[person] wife is Who.

Optionally, in the present embodiment, the target problem template with correspondence and answer information can be obtained from advance It is obtained in the question and answer pair taken, such as：The question and answer pair for obtaining 6,000,000 in advance, by 6,000,000 question and answer to the problems in be converted to ask one by one Inscribe template.Transformed question template is integrated, then one or more may have been corresponded under the same question template This corresponding answer of one or more problem is determined as the corresponding answer information of the template by problem, to obtain have pair The target problem template and answer information that should be related to.

Optionally, in the present embodiment, the predicate with correspondence and content information can be, but not limited to from knowledge base Middle acquisition, with correspondence predicate and content information can be, but not limited to be integrated according to subject, can also but It is not limited to be integrated according to predicate.

Such as：By taking disease knowledge as an example, the relevant knowledge of 9600 diseases is got as knowledge base, from this 9600 diseases The predicate of this knowledge is extracted in each of the relevant knowledge of disease, and is believed this knowledge as the corresponding content of the predicate Breath, a plurality of knowledge may extract identical predicate, be integrated to predicate, and it is one or more corresponding to obtain each predicate A content information, to obtain the predicate with correspondence and content information.Alternatively, can also be according to subject to relevant knowledge It is integrated, then extracts predicate under each subject, and establish the correspondence between predicate and content information, will had and correspond to Subject, predicate and the content information of relationship are determined as above-mentioned predicate and content information with correspondence.

In an optional embodiment, as shown in figure 4, by taking disease areas as an example, 6,000,000 question and answer pair are obtained, from Target problem template and answer information of the 6000000 question and answer centering extraction with correspondence.Obtain the correlation of 9600 diseases Knowledge is as knowledge base, predicate and content information of the extraction with correspondence from knowledge base.It is each in information according to answering It answers the similarity between information and content information and determines each predicate answered corresponding to information, then count each predicate and correspond to Answer information quantity, the most predicate of the corresponding quantity for answering information in predicate is determined as corresponding to target problem template Target predicate.

As it can be seen that through the above steps, obtain target problem template with correspondence and answer information and with pair The predicate and content information that should be related to are established each answer further according to the similarity between each answer information and content information and are believed The most predicate of the corresponding quantity for answering information is determined as target problem template and corresponded to by the correspondence between breath and predicate Position, to using vote by the way of, vote automatically predicate according to similarity for each target problem template, and The correspondence between each target problem template and the highest predicate of poll is established, to be that question template determines that its is right automatically The predicate answered to realize the technique effect for improving efficiency when determining the predicate corresponding to question template, and then solves Efficiency lower technical problem when determining the predicate corresponding to question template in the related technology.

As a kind of optional scheme, obtains the target problem template with correspondence and answer information includes：

Each problem of question and answer centering is converted to question template, wherein the problem of question and answer are to for correspondence by S1 And answer；

S2 obtains mutually different question template from question template, obtains target problem template；

Answer corresponding the problem of belonging to target problem template is determined as the corresponding answer of target problem template and believed by S3 Breath obtains the target problem template with correspondence and answers information.

Optionally, in the present embodiment, during problem is converted into question template, it is possible that multiple problems The case where being converted to the same question template, such as：How scapulohumeral periarthritis treats, which the therapeutic modality of cervical spondylosis has, how The problems such as treatment varicella etc., can be converted to the same question template：How [sickname] treats.That is, for M A problem can be converted to Q mutually different question templates, and wherein Q is less than or equal to M, different by this Q The problem of template be determined as target problem template.

Optionally, in the present embodiment, the corresponding answer of all the problem of being converted into same problem template is determined as The corresponding answer information of the problem.Such as：Problem：How scapulohumeral periarthritis treats, which the therapeutic modality of cervical spondylosis has, how to treat Varicella is converted to target problem template：How [sickname] treats.Problem：How scapulohumeral periarthritis prevent, the prevention side of cervical spondylosis Which formula, which has, how to prevent varicella is converted to target problem template：How [sickname] prevents.So can by scapulohumeral periarthritis why Treat corresponding answer 1, which corresponding answer 2 therapeutic modality of cervical spondylosis has, how treating varicella, corresponding answer 3 is true It is set to target problem template：How [sickname] treats corresponding answer information, and how scapulohumeral periarthritis is prevented corresponding answer 4, which corresponding answer 5 precautionary approach of cervical spondylosis has, how to prevent the corresponding answer of varicella 6 and be determined as target problem mould Plate：How [sickname] prevents corresponding answer information.Obtain the target problem template with correspondence：[sickname] How to treat and answer information：Answer 1, answer 2, answer 3, and the target problem template with correspondence： How [sickname] prevents and answers information：Answer 4, answer 5, answer 6.

As a kind of optional scheme, mutually different question template is obtained from question template, obtains target problem mould Plate includes：

S1 obtains mutually different question template from question template；

S2 obtains template the problem of question template type in mutually different question template belongs to binary fact type problem；

S3, by template the problem of the most destination number of corresponding problematic amount in template the problem of binary fact type problem It is determined as target problem template.

Optionally, in the present embodiment, binary fact type problem (Binary Factoid Question, referred to as BFQ), the problem of generally referring to the attribute of inquiry entity in a certain respect.

Optionally, in the present embodiment, frequency highest can be selected from magnanimity target problem template and belongs to BFQ problems 200 templates.

As a kind of optional scheme, obtaining predicate and content information with correspondence includes：

S1 obtains the knowledge data of target domain；

S2 establishes the knowledge mapping of target domain according to knowledge data；

S3 obtains subject, predicate and object with correspondence from the information triple of knowledge mapping；

S4, by with correspondence predicate and object be determined as the predicate with correspondence and content information, alternatively, By with correspondence under subject predicate and object be determined as corresponding predicate and the content letter with correspondence of subject Breath.

Optionally, in the present embodiment, knowledge mapping is also known as mapping knowledge domains, is known as knowledge domain in books and information group Visualization or ken map map, are explicit knowledge's development process and a series of a variety of different figures of structural relation, With visualization technique Description of Knowledge resource and its carrier, excavation, analysis, structure, drafting and explicit knowledge and the phase between them Mutually contact.

Such as：9600 relevant knowledge data for obtaining disease areas, the content of 9600 initial data is also closed Keyword extracts.Such as：The predicate of scapulohumeral periarthritis has：Treatment, common sympton, complication etc., wherein the pass of " treatment " this predicate Keyword extracts result：Analgesic, adhesion, drug, operation, Chinese medicine；The keyword extraction result of " common sympton " this predicate is： Chronic, pain, diffusion, shoulder mobility is limited, shoulder ache, and head movement is limited, shoulder pain.

Optionally, in the present embodiment, the predicate with correspondence and content information can be, but not limited to be according to master What language was classified, i.e., every knowledge data extracts subject, predicate and content information, by each knowledge data according to subject into Row classification, the knowledge data of identical subject are integrated together, and each subject has corresponded to predicate, and each predicate has corresponded to content letter again Breath.Alternatively, what predicate and content information with correspondence can also but be not limited to not distinguish according to subject, also It is to say, every knowledge data extracts a predicate and content information with correspondence.

Optionally, in the present embodiment, it can be, but not limited to determine that each to answer information institute right one of in the following manner The predicate answered：

The predicate for having correspondence and object are being determined as the predicate with correspondence and content information by mode one In the case of, determine each the first similarity for answering information content information corresponding with each first predicate, wherein the first meaning Word is the predicate in predicate and content information with correspondence；

Highest first predicate of corresponding first similarity is determined as the predicate corresponding to each answer information.

Mode two, by with correspondence under subject predicate and object be determined as that subject is corresponding to close with corresponding In the case of the predicate and content information of system, the second predicate and content information with correspondence are obtained, wherein the second predicate Identical subject corresponding with each answer information；

Determine each the second similarity for answering information content information corresponding with each second predicate；

Highest second predicate of corresponding second similarity is determined as the predicate corresponding to each answer information.

It is true according to the similarity each answered in answer information between information and content information as a kind of optional scheme Determining each predicate answered corresponding to information includes：

S1 extracts keyword from each answer information, and obtaining, there is each of correspondence, which to answer information and first, closes Keyword set；

S2 extracts keyword from the corresponding content information of each predicate, obtain having correspondence content information and Second keyword set；

S3 obtains each corresponding first keyword set of information corresponding with each content information second of answering and closes respectively Similarity between keyword set, and the similarity between the first keyword set and the second keyword set is determined as each Answer the similarity between information and content information；

Predicate corresponding each highest object content information of similarity answered between information is determined as each by S4 Answer the predicate corresponding to information.

Optionally, in the present embodiment, determine each similarity for answering information and content information mode can with but not It is limited to include the similarity determined between each keyword and the keyword of content information for answering information.

Optionally, in the present embodiment, determine that the mode of the similarity between keyword and keyword can be, but not limited to Including：Obtain the outstanding card similarity between keyword and keyword.For two set x and y, outstanding card similarity calculation Formula isI.e. outstanding card similarity J is two intersection of sets divided by two unions of sets.

It should be noted that other similarity calculating methods can also be used when above-mentioned calculating similarity, such as The computational methods based on term vector may be used.

As a kind of optional scheme, asked the most predicate of the corresponding quantity for answering information in predicate is determined as target After inscribing the target predicate corresponding to template, further include：

S1, using with correspondence target problem template and target predicate to obtain first problem input by user institute right First answered is answered；

S2 is answered first and is returned to user as the response message of first problem.

Optionally, in the present embodiment, the target problem template and target with correspondence got can be utilized Predicate carries out further natural language processing, such as：Obtain the answer etc. that user is asked a question.

As a kind of optional scheme, using with correspondence target problem template and target predicate to obtain user defeated Corresponding to the first problem entered first answer include：

S1 obtains first problem input by user, and the first subject is extracted from first problem；

First problem is converted to first problem template by S2；

S3, from correspondence target problem template and target predicate in obtain first problem template corresponding first Predicate；

S4 obtains the corresponding knowledge mapping of the first subject from the knowledge base of first problem fields；

S5, obtains the corresponding answer information of the first predicate from the corresponding knowledge mapping of the first subject, and by the first predicate Corresponding answer information is determined as the first answer.

Optionally, in the present embodiment, it gets after target problem template with correspondence and target predicate just Quick predicate can be carried out to problem input by user to predict, first problem input by user is converted into first problem mould Plate, then from correspondence target problem template and target predicate in predict the first problem template it is corresponding first meaning The corresponding answer information of first predicate is obtained as first time in word, then knowledge mapping corresponding to the subject of first problem It answers and returns to user.

It should be noted that for each method embodiment above-mentioned, for simple description, therefore it is all expressed as a series of Combination of actions, but those skilled in the art should understand that, the present invention is not limited by the described action sequence because According to the present invention, certain steps can be performed in other orders or simultaneously.Secondly, those skilled in the art should also know It knows, embodiment described in this description belongs to preferred embodiment, and involved action and module are not necessarily of the invention It is necessary.

Through the above description of the embodiments, those skilled in the art can be understood that according to above-mentioned implementation The method of example can add the mode of required general hardware platform to realize by software, naturally it is also possible to by hardware, but it is very much In the case of the former be more preferably embodiment.Based on this understanding, technical scheme of the present invention is substantially in other words to existing The part that technology contributes can be expressed in the form of software products, which is stored in a storage In medium (such as ROM/RAM, magnetic disc, CD), including some instructions are used so that a station terminal equipment (can be mobile phone, calculate Machine, server or network equipment etc.) execute method described in each embodiment of the present invention.

Other side according to the ... of the embodiment of the present invention additionally provides a kind of for implementing corresponding to above problem template The determining device of predicate corresponding to the problem of determination method of predicate template, as shown in figure 5, the device includes：

First acquisition module 52, for obtaining target problem template and answer information with correspondence, and acquisition Predicate with correspondence and content information；

First determining module 54, for true according to the similarity each answered in answer information between information and content information Fixed each predicate answered corresponding to information；

Second determining module 56, for the most predicate of the corresponding quantity for answering information in predicate to be determined as target problem Target predicate corresponding to template.

Optionally, in the present embodiment, the determining device of the predicate corresponding to above problem template can be applied to such as Fig. 2 Shown in the hardware environment that is constituted of equipment 202.As shown in Fig. 2, equipment 202 obtains the target problem mould with correspondence Plate and answer information, and obtain predicate and content information with correspondence；Information is each answered according to answering in information Similarity between content information determines each predicate answered corresponding to information；By the corresponding quantity for answering information in predicate Most predicates is determined as the target predicate corresponding to target problem template.

Optionally, in the present embodiment, the determining device of the predicate corresponding to above problem template can be, but not limited to answer For determining between question template and predicate in the scene of correspondence.Wherein, the predicate corresponding to above problem template is really The method of determining can be, but not limited to be applied to various types of applications, for example, online education application, instant messaging application, community's sky Between apply, game application, shopping application, browser application, financial application, multimedia application, live streaming application, medical applications, with And interactive application of Intelligent hardware etc. (wherein, Intelligent hardware can be, but not limited to include：Smart home device, intelligence wearing are set Standby, intelligent transportation tool etc.).Specifically, can be, but not limited to be applied in above-mentioned browser application determine question template with Between predicate in the scene of correspondence, or can with but be not limited to be applied to above-mentioned game application determine question template with meaning Between word in the scene of correspondence, the efficiency when predicate corresponding to question template is determined to improve.Above-mentioned is only that one kind is shown , do not do any restriction to this in the present embodiment.

Optionally, in the present embodiment, the determining device of the predicate corresponding to above problem template can be, but not limited to answer For in the scene of natural language processing.Such as：Knowledge base question and answer (Knowledge Base Question Answering, letter Referred to as KB-QA), natural language problem (Query) is given, by carrying out semantic understanding and parsing to problem, and then utilizes knowledge Library is inquired, reasoning obtains answer (Answer).

In an optional embodiment, as shown in figure 4, by taking disease areas as an example, 6,000,000 question and answer pair are obtained, from Target problem template and answer information of the 6000000 question and answer centering extraction with correspondence.Obtain the correlation of 9600 diseases Knowledge is as knowledge base, predicate and content information of the extraction with correspondence from knowledge base.It is each in information according to answering It answers the similarity between information and content information and determines each predicate answered corresponding to information, corresponded to counting each predicate Answer information quantity, the most predicate of the corresponding quantity for answering information in predicate is determined as corresponding to target problem template Target predicate.

As it can be seen that by above-mentioned apparatus, obtain target problem template with correspondence and answer information and with pair The predicate and content information that should be related to are established each answer further according to the similarity between each answer information and content information and are believed The most predicate of the corresponding quantity for answering information is determined as target problem template and corresponded to by the correspondence between breath and predicate Position, to using vote by the way of, vote automatically predicate according to similarity for each target problem template, and The correspondence between each target problem template and the highest predicate of poll is established, to be that question template determines that its is right automatically The predicate answered to realize the technique effect for improving efficiency when determining the predicate corresponding to question template, and then solves Efficiency lower technical problem when determining the predicate corresponding to question template in the related technology.

As a kind of optional scheme, the first acquisition module includes：

First converting unit, for each problem of question and answer centering to be converted to question template, wherein question and answer to for The problem of correspondence and answer；

First acquisition unit obtains target problem template for obtaining mutually different question template from question template；

First determination unit, for answer corresponding the problem of belonging to target problem template to be determined as target problem mould The corresponding answer information of plate obtains the target problem template with correspondence and answers information.

As a kind of optional scheme, first acquisition unit includes：

First obtains subelement, for obtaining mutually different question template from question template；

Second obtains subelement, belongs to binary fact type for obtaining question template type in mutually different question template The problem of problem template；

Determination subelement is used for the most number of targets of corresponding problematic amount in template the problem of binary fact type problem The problem of amount, template was determined as target problem template.

As a kind of optional scheme, the first acquisition module includes：

Second acquisition unit, the knowledge data for obtaining target domain；

Unit is established, the knowledge mapping for establishing target domain according to knowledge data；

Third acquiring unit, for obtaining subject, predicate with correspondence from the information triple of knowledge mapping And object；

Second determination unit, predicate for the predicate with correspondence and object to be determined as having correspondence and Content information, alternatively, by with correspondence under subject predicate and object to be determined as subject corresponding with correspondence Predicate and content information.

Optionally, in the present embodiment, the second determining module is used for：It is determined in the predicate and object that will have correspondence In the case of predicate and content information with correspondence, determine that each answer information is corresponding with each first predicate interior Hold the first similarity of information, wherein the first predicate is the predicate in predicate and content information with correspondence；It will correspond to Highest first predicate of the first similarity be determined as it is each answer information corresponding to predicate.

Optionally, in the present embodiment, the second determining module is used for：By with correspondence under subject predicate and In the case that object is determined as subject corresponding predicate and content information with correspondence, the with correspondence is obtained Two predicates and content information, wherein the second predicate identical subject corresponding with each answer information；Determine it is each answer information with Second similarity of the corresponding content information of each second predicate；Highest second predicate of corresponding second similarity is determined as Each predicate answered corresponding to information.

As a kind of optional scheme, the second determining module includes：

First extraction unit obtains that there is each of correspondence to return for extracting keyword from each answer information Answer information and the first keyword set；

Second extraction unit obtains having corresponding close for extracting keyword from the corresponding content information of each predicate The content information of system and the second keyword set；

4th acquiring unit is believed for obtaining each corresponding first keyword set of information of answering respectively with each content Cease the similarity between corresponding second keyword set, and by the phase between the first keyword set and the second keyword set It is determined as the similarity between each answer information and content information like degree；

Third determination unit, for the highest object content information of similarity between each answer information is corresponding Predicate is determined as the predicate corresponding to each answer information.

As a kind of optional scheme, above-mentioned apparatus further includes：

Second acquisition module is inputted for being obtained user using the target problem template and target predicate with correspondence First problem corresponding to first answer；

Module being returned, user is returned to as the response message of first problem for answering first.

As a kind of optional scheme, the second acquisition module includes：

5th acquiring unit for obtaining first problem input by user, and extracts the first subject from first problem；

Second converting unit, for first problem to be converted to first problem template；

6th acquiring unit, for from correspondence target problem template and target predicate in obtain first problem Corresponding first predicate of template；

7th acquiring unit, for obtaining the corresponding knowledge graph of the first subject from the knowledge base of first problem fields Spectrum；

8th acquiring unit, for obtaining the corresponding answer letter of the first predicate from the corresponding knowledge mapping of the first subject Breath, and the corresponding answer information of the first predicate is determined as the first answer.

The application environment of the embodiment of the present invention can be, but not limited to reference to the application environment in above-described embodiment, the present embodiment In this is repeated no more.An embodiment of the present invention provides the optional tools of one kind of the connection method for implementing above-mentioned real-time Communication for Power Body application example.

As a kind of optional embodiment, the determination method of the predicate corresponding to above problem template can be, but not limited to answer In scene for determining the corresponding answer of the problem of receiving in natural language processing as shown in FIG. 6.In this scene, carry The template for having gone out a kind of knowledge based question and answer library is intended to method for digging, finds the mapping relations between template and knowledge base predicate, The true intention (prediction predicate) for understanding user, then finds answer according to predicate in the database.This method generates each Then the corresponding template of problem is intended to method for digging using the template of knowledge based collection of illustrative plates to template and carries out between template and predicate Mapping, the algorithm of keyword extraction, outstanding card similarity calculation and ballot, final choice poll are used in mapping method Predicate of the highest predicate as the template can finally integrate the mapping relations between all templates and predicate.

Optionally, in the present embodiment, the existing knowledge for using knowledge based collection of illustrative plates, in conjunction with the related field of excavation Query and answer carry out the prediction of predicate using the method for ballot, finally obtain and vote between most template and predicate A kind of mapping relations, the supplement as existing knowledge base.Arbitrarily carry out a query after handling in this way, it can be by finding it Mapping relations of the template in knowledge base find its corresponding predicate, finally find answer in the database.Obvious this method More intelligence copes with the magnanimity way to put questions of user with effectively in the case of data volume abundance.

Optionally, in this scene, it is proposed that a kind of knowledge based collection of illustrative plates comes the mapping method of prediction module intention, user Query to directly generate template, and the method for being then based on knowledge mapping passes through the mapping process of template to predicate, so that it may to look for To the corresponding predicate of the template, knot of the predicate result of corresponding entity as the query is finally found in knowledge mapping again Fruit.

Optionally, in this scene, the template that the above method can be applied to the every field of dobby intelligence assistants is dug Pick.When creating some field, a large amount of language materials in the field and a large amount of related entities in the field may be collected, But the way to put questions of user is ever-changing, limited sentence can not possibly include all ways to put questions.Therefore, template is for field Effect seems particularly significant, and good template can include the ever-changing way to put questions in the field, therefore pass through language material and relevant entity It is suitble to the template in the field to seem particularly significant to excavate.For already existing field, if it find that some ways to put questions can't It supports, template can also be excavated by these ways to put questions and related entities, promote the recall rate and semantic understanding ability in the field. Therefore, the mining ability of template all seems particularly significant for having field and new field.

This method mainly applies to the answer at the chat end of product and relevant knowledge question, to necks such as health, diseases Domain is extremely important, in addition, being also very important for the knowledge question of general field.After the query in any field has come Corresponding template can be generated, then the predicate of corresponding templates is found by trained mapping relations, passes through correspondence Entity and predicate can easily find answer, return to user.

In order to illustrate being how to carry out template Intention Anticipation using knowledge mapping in this scene, illustrate herein It is bright, in the present embodiment, excavates the 6000000 and relevant qa of disease and the relevant knowledge of (question and answer to) and 9600 diseases has been made For KB, it is therefore an objective to obtain the mapping relations between the template of disease areas and corresponding predicate.As shown in fig. 6, whole process packet Include following steps：

Step 1, template is excavated：The template for first having to carry out 6,000,000 query is excavated, and (entity therein is exactly 9600 diseases Disease), that is, complete the conversion from query to template.(for example obtain：How [sick_name], [sick_name] control The template for the treatment of ... ..., etc..)

Step 2, the BFQ templates of top200 are found out：The problem of that knowledge question is supported is BFQ, therefore can be from magnanimity mould Frequency highest is selected in plate and belongs to 200 templates of BFQ problems.(such as selection：How [sick_name] treats, Which the symptom of [sick_name] has ... ...)

Step 3, all answer of each template are recalled and extract answer keyword：Find all of each template Answer, so as to the basis as ballot.And interminable answer is subjected to keyword extraction.(such as：" how is [sick_name] The answer of this query for the treatment of " has many hundreds, is respectively：1. scapulohumeral periarthritis obtained after should western medical treatment scapulohumeral periarthritis it is general Using drug therapy with operative treatment, drug therapy is mainly the drug for allowing patient to take orally anti-inflammatory analgetic class, but most of trouble Person can recur after being discontinued, and be treated with operation method, adhesion be easily caused, so general treatment scapulohumeral periarthritis All compare and recommends Chinese traditional treatment method.2…….Then to each answer carry out keyword extraction, such as first answer Keyword extraction result is as follows：Scapulohumeral periarthritis, drug are performed the operation, and are taken orally, and are relieved pain, recurrence, adhesion, Chinese medicine).Keyword herein carries It takes and Textrank4 algorithms may be used, extract 50 keywords in every section of words.

Step 4,9600 initial data carry out predicate and keyword extraction, as KB：In order to facilitate processing, by 9600 The content of initial data also carries out keyword extraction.(such as：The predicate of scapulohumeral periarthritis has：Treatment, common sympton, complication ... The keyword extraction result for wherein " treating " this predicate is：Analgesic, adhesion, drug, operation, Chinese medicine；" common sympton " this The keyword extraction result of predicate is：Chronic, pain, diffusion, shoulder mobility is limited, shoulder ache, and head movement is limited, shoulder Bitterly).

Step 5, template is intended to ballot：Each of template, which is answered, can only vote to some intention, finally for a certain Highest intention of winning the vote for a template is exactly its prediction intention or true intention, such as：How [sick_name] treats { predicate " treatment "：30, predicate " Diet "：20, predicate " CommonCause "：10 ... }, then for this template For how [sick_name] treats, final voting results are exactly " treatment ", that is to say, that this template is asked just It is treatment-related information.Each template has hundreds of thousands of to answer, each answer needs to carry out each intention of KB Outstanding card similarity calculation, and voted according to result of calculation and give similarity highest predicate, it is counted after final all answer ballots The highest predicate of ticket is intended to result as the prediction of template.(such as：First template for example, perform the operation by scapulohumeral periarthritis, drug, It is oral, relieve pain, recurrence, adhesion, Chinese medicine these keywords need and KB in corresponding entity " scapulohumeral periarthritis " be possible to call The outstanding card similarity of keyword progress of word is calculated, and (for two set x and y, outstanding card calculating formula of similarity isThat is two intersection of sets divided by two unions of sets), select highest scoring, such as first answer and predicate The highest scoring of " treatment " is answered for final hundreds of thousands of and is voted respectively, " treatment "：500 tickets, " common sympton "：100 tickets, " simultaneously Send out disease "：The final prediction that 50 tickets ... at this moment can select first template is intended to result and is：" treatment ").

Another aspect according to the ... of the embodiment of the present invention additionally provides a kind of for implementing corresponding to above problem template The electronic device of the determination of predicate, as shown in fig. 7, the electronic device includes：One or more (one is only shown in figure) processing Device 702, memory 704, sensor 706, encoder 708 and transmitting device 710 are stored with computer journey in the memory Sequence, the processor are arranged to execute the step in any of the above-described embodiment of the method by computer program.

Optionally, in the present embodiment, above-mentioned electronic device can be located in multiple network equipments of computer network At least one network equipment.

Optionally, in the present embodiment, above-mentioned processor can be set to execute following steps by computer program：

S1 obtains the target problem template with correspondence and answers information, and obtains the meaning with correspondence Word and content information；

S2 determines each answer information institute according to the similarity each answered in answer information between information and content information Corresponding predicate；

The most predicate of the corresponding quantity for answering information in predicate is determined as the target corresponding to target problem template by S3 Predicate.

Optionally, it will appreciated by the skilled person that structure shown in Fig. 7 is only to illustrate, electronic device also may be used To be smart mobile phone (such as Android phone, iOS mobile phones), tablet computer, palm PC and mobile internet device The terminal devices such as (Mobile Internet Devices, MID), PAD.Fig. 7 it does not cause the structure of above-mentioned electronic device It limits.For example, electronic device may also include more than shown in Fig. 7 or less component (such as network interface, display device Deng), or with the configuration different from shown in Fig. 7.

Wherein, memory 702 can be used for storing software program and module, such as the problems in embodiment of the present invention template institute Corresponding program instruction/the module of determination method and apparatus of corresponding predicate, processor 704 are stored in memory by operation Software program in 702 and module realize above-mentioned target element to perform various functions application and data processing Control method.Memory 702 may include high speed random access memory, can also include nonvolatile memory, such as one or Multiple magnetic storage devices, flash memory or other non-volatile solid state memories.In some instances, memory 702 can be into one Step includes the memory remotely located relative to processor 704, these remote memories can pass through network connection to terminal.On The example for stating network includes but not limited to internet, intranet, LAN, mobile radio communication and combinations thereof.

Above-mentioned transmitting device 710 is used to receive via a network or transmission data.Above-mentioned network specific example It may include cable network and wireless network.In an example, transmitting device 710 includes a network adapter (Network Interface Controller, NIC), can be connected with other network equipments with router by cable so as to interconnection Net or LAN are communicated.In an example, transmitting device 710 is radio frequency (Radio Frequency, RF) module, For wirelessly being communicated with internet.

Wherein, specifically, memory 702 is for storing application program.

The embodiments of the present invention also provide a kind of storage medium, computer program is stored in the storage medium, wherein The computer program is arranged to execute the step in any of the above-described embodiment of the method when operation.

Optionally, in the present embodiment, above-mentioned storage medium can be set to store by executing based on following steps Calculation machine program：

Optionally, storage medium is also configured to store for executing step included in the method in above-described embodiment Computer program, this is repeated no more in the present embodiment.

Optionally, in the present embodiment, one of ordinary skill in the art will appreciate that in the various methods of above-described embodiment All or part of step be that can be completed come command terminal device-dependent hardware by program, which can be stored in In one computer readable storage medium, storage medium may include：Flash disk, read-only memory (Read-Only Memory, ROM), random access device (Random Access Memory, RAM), disk or CD etc..

The embodiments of the present invention are for illustration only, can not represent the quality of embodiment.

If the integrated unit in above-described embodiment is realized in the form of SFU software functional unit and as independent product Sale in use, can be stored in the storage medium that above computer can be read.Based on this understanding, skill of the invention Substantially all or part of the part that contributes to existing technology or the technical solution can be with soft in other words for art scheme The form of part product embodies, which is stored in a storage medium, including some instructions are used so that one Platform or multiple stage computers equipment (can be personal computer, server or network equipment etc.) execute each embodiment institute of the present invention State all or part of step of method.

In the above embodiment of the present invention, all emphasizes particularly on different fields to the description of each embodiment, do not have in some embodiment The part of detailed description may refer to the associated description of other embodiment.

In several embodiments provided herein, it should be understood that disclosed client, it can be by others side Formula is realized.Wherein, the apparatus embodiments described above are merely exemplary, for example, the unit division, only one Kind of division of logic function, formula that in actual implementation, there may be another division manner, such as multiple units or component can combine or It is desirably integrated into another system, or some features can be ignored or not executed.Another point, it is shown or discussed it is mutual it Between coupling, direct-coupling or communication connection can be INDIRECT COUPLING or communication link by some interfaces, unit or module It connects, can be electrical or other forms.

The unit illustrated as separating component may or may not be physically separated, aobvious as unit The component shown may or may not be physical unit, you can be located at a place, or may be distributed over multiple In network element.Some or all of unit therein can be selected according to the actual needs to realize the mesh of this embodiment scheme 's.

In addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, it can also It is that each unit physically exists alone, it can also be during two or more units be integrated in one unit.Above-mentioned integrated list The form that hardware had both may be used in member is realized, can also be realized in the form of SFU software functional unit.

The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art For member, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also answered It is considered as protection scope of the present invention.

Claims

1. a kind of determination method of the predicate corresponding to question template, which is characterized in that including：

It obtains the target problem template with correspondence and answers information, and obtain predicate and content with correspondence Information；

Each answer is determined according to the similarity between information and the content information is each answered in the answer information Predicate corresponding to information；

The most predicate of the corresponding quantity for answering information in the predicate is determined as the mesh corresponding to the target problem template Mark predicate.

2. according to the method described in claim 1, it is characterized in that, obtaining the target problem template with correspondence and answer Information includes：

Each problem of question and answer centering is converted into question template, wherein the problem of question and answer are to for correspondence and It answers；

Mutually different question template is obtained from described problem template, obtains the target problem template；

It is corresponding described that answer corresponding the problem of belonging to the target problem template is determined as the target problem template Information is answered, the target problem template with correspondence and the answer information are obtained.

3. according to the method described in claim 2, it is characterized in that, obtaining different problem mould from described problem template Plate, obtaining the target problem template includes：

The mutually different question template is obtained from described problem template；

Obtain template the problem of question template type in the mutually different question template belongs to binary fact type problem；

Template the problem of the most destination number of corresponding problematic amount in template the problem of the binary fact type problem is true It is set to the target problem template.

4. according to the method described in claim 1, it is characterized in that, obtaining the predicate and content information packet with correspondence It includes：

Obtain the knowledge data of target domain；

The knowledge mapping of the target domain is established according to the knowledge data；

Subject, predicate and object with correspondence are obtained from the information triple of the knowledge mapping；

By with correspondence predicate and object be determined as the predicate and content information with correspondence, alternatively, general Predicate with correspondence and object under the subject are determined as the corresponding meaning with correspondence of the subject Word and content information.

5. according to the method described in claim 4, it is characterized in that, according to information and described is each answered in the answer information Similarity between content information determines that each predicate answered corresponding to information includes：

The case where the predicate with correspondence and object are determined as the predicate and content information with correspondence Under, determine each first similarity for answering information content information corresponding with each first predicate, wherein described first Predicate is the predicate in the predicate and content information with correspondence；

Corresponding highest first predicate of first similarity is determined as each meaning answered corresponding to information Word.

6. according to the method described in claim 4, it is characterized in that, according to information and described is each answered in the answer information Similarity between content information determines that each predicate answered corresponding to information includes：

By with correspondence under the subject predicate and object be determined as that the subject is corresponding described to have correspondence In the case of the predicate and content information of relationship, the second predicate and content information with correspondence are obtained, wherein described the Two predicates identical subject corresponding with each answer information；

Determine each second similarity for answering information content information corresponding with each second predicate；

Corresponding highest second predicate of second similarity is determined as each meaning answered corresponding to information Word.

7. according to the method described in claim 1, it is characterized in that, according to information and described is each answered in the answer information Similarity between content information determines that each predicate answered corresponding to information includes：

Keyword is extracted from each answer information, each answer information with correspondence and first is obtained and closes Keyword set；

Extract keyword from the corresponding content information of each predicate, obtain having correspondence the content information and Second keyword set；

Corresponding first keyword set of each answer information and each content information corresponding second are obtained respectively Similarity between keyword set, and by the similarity between first keyword set and second keyword set It is determined as each similarity answered between information and the content information；

It will be determined as each corresponding predicate of the highest object content information of similarity answered between information described every Predicate corresponding to a answer information.

8. method according to any one of claim 1 to 6, which is characterized in that answering letter by corresponding in the predicate The predicate that the quantity of breath is most is determined as after the target predicate corresponding to the target problem template, and the method further includes：

Using with correspondence the target problem template and the target predicate obtain first problem institute input by user Corresponding first answers；

Described first is answered and returns to the user as the response message of the first problem.

9. according to the method described in claim 8, it is characterized in that, using with correspondence the target problem template and The target predicate obtains the first answer corresponding to first problem input by user：

First problem input by user is obtained, and extracts the first subject from the first problem；

The first problem is converted into first problem template；

From with correspondence the target problem template and the target predicate in obtain the first problem template and correspond to The first predicate；

The corresponding knowledge mapping of first subject is obtained from the knowledge base of the first problem fields；

Obtain the corresponding answer information of first predicate from the corresponding knowledge mapping of first subject, and by described first The corresponding answer information of predicate is determined as described first and answers.

10. a kind of determining device of the predicate corresponding to question template, which is characterized in that including：

First acquisition module, for obtains with correspondence target problem template and answer information, and obtain with pair The predicate and content information that should be related to；

First determining module, for according to the similarity each answered in the answer information between information and the content information Determine each predicate answered corresponding to information；

Second determining module is asked for the most predicate of the corresponding quantity for answering information in the predicate to be determined as the target Inscribe the target predicate corresponding to template.

11. device according to claim 10, which is characterized in that first acquisition module includes：

First converting unit, for each problem of question and answer centering to be converted to question template, wherein the question and answer to for The problem of correspondence and answer；

First acquisition unit obtains the target problem for obtaining mutually different question template from described problem template Template；

First determination unit is asked for answer corresponding the problem of belonging to the target problem template to be determined as the target The corresponding answer information of template is inscribed, the target problem template with correspondence and the answer information are obtained.

12. device according to claim 10, which is characterized in that first acquisition module includes：

Second acquisition unit, the knowledge data for obtaining target domain；

Unit is established, the knowledge mapping for establishing the target domain according to the knowledge data；

Third acquiring unit, for obtaining subject, predicate with correspondence from the information triple of the knowledge mapping And object；

Second determination unit, for by the predicate with correspondence and object be determined as the predicate with correspondence and Content information, alternatively, by with correspondence under the subject predicate and object to be determined as the subject corresponding described Predicate with correspondence and content information.

13. device according to claim 10, which is characterized in that second determining module includes：

First extraction unit obtains having the described every of correspondence for extracting keyword from each answer information A answer information and the first keyword set；

Second extraction unit obtains having corresponding close for extracting keyword from the corresponding content information of each predicate The content information and the second keyword set of system；

4th acquiring unit, for obtain respectively corresponding first keyword set of each answer information with it is each it is described in Hold the similarity between corresponding second keyword set of information, and by first keyword set and second keyword Similarity between set is determined as each similarity answered between information and the content information；

Third determination unit, being used for will be corresponding with each highest object content information of similarity answered between information Predicate is determined as each predicate answered corresponding to information.

14. a kind of storage medium, which is characterized in that be stored with computer program in the storage medium, wherein the computer Program is arranged to execute the method described in any one of claim 1 to 9 when operation.

15. a kind of electronic device, including memory and processor, which is characterized in that be stored with computer journey in the memory Sequence, the processor are arranged to execute the side described in any one of claim 1 to 9 by the computer program Method.