CN106919655A

CN106919655A - A kind of answer provides method and apparatus

Info

Publication number: CN106919655A
Application number: CN201710060256.5A
Authority: CN
Inventors: 朱臻
Original assignee: Netease Hangzhou Network Co Ltd
Current assignee: Netease Hangzhou Network Co Ltd
Priority date: 2017-01-24
Filing date: 2017-01-24
Publication date: 2017-07-04
Anticipated expiration: 2037-01-24
Also published as: CN106919655B

Abstract

Embodiments of the present invention provide a kind of answer and provide method.The method includes：Receive the problem that user submits to；According to the entity node and attribute node that are extracted from described problem, the corresponding inquiry chain of described problem is built；According to the corresponding inquiry chain of described problem, the answer of described problem is retrieved in the advance knowledge mapping for building, the answer of the described problem obtained as the retrieval mode retrieval of knowledge based collection of illustrative plates；The answer feedback of the described problem that the retrieval mode retrieval of knowledge based collection of illustrative plates is obtained is to the user.Embodiments of the present invention can to a certain extent improve the possibility of the answer for retrieving the problem that the answer for obtaining is submitted to for user, so as to improve the accuracy of the answer for feeding back to user to a certain extent.Additionally, embodiments of the present invention provide a kind of answer offer device.

Description

A kind of answer provides method and apparatus

Technical field

Embodiments of the present invention are related to automatic question answering technical field, more specifically, embodiments of the present invention are related to one Plant answer and method and apparatus are provided.

Background technology

This part is it is intended that the embodiments of the present invention stated in claims provide background or context.Herein Description in being included in this part because just do not recognize it is prior art.

At present, existing some user's intelligent Answer Systems, user input problem, intelligent Answer System is using certain retrieval side The answer of the problem of formula retrieval user input.

The retrieval mode that conventional user's intelligent Answer System is used is for the retrieval mode based on search engine, based on depth Spend retrieval mode, the retrieval mode based on local problem base, the retrieval mode based on Ask-Answer Community knowledge base of study.

But these current retrieval modes, will not be understood and be analyzed to the problem that user submits to, but directly According to default template or keyword search and the immediate answer of the problem, this may result in the standard of the answer for feeding back to user The relatively low problem of true property, Consumer's Experience is poor.

The content of the invention

In view of present in prior art according to be with the problem raised the price be user feedback answer accuracy is low, user's body Poor problem is tested, the present invention provides a set of answer and provides method and apparatus, so as to improve accuracy and the user of answer feedback Experience.

In the present context, embodiments of the present invention expect that providing a kind of answer provides method and apparatus.

In the first aspect of embodiment of the present invention, there is provided a kind of answer provides method, including：

Receive the problem that user submits to；

According to the entity node and attribute node that are extracted from described problem, the corresponding inquiry chain of described problem is built；

According to the corresponding inquiry chain of described problem, the answer of described problem is retrieved in the advance knowledge mapping for building, made It is the answer of the described problem that the retrieval mode retrieval of knowledge based collection of illustrative plates is obtained；

The answer feedback of the described problem that the retrieval mode retrieval of knowledge based collection of illustrative plates is obtained is to the user.

Preferably, in methods described, according to the entity node and attribute node that are extracted from described problem, build described The corresponding inquiry chain of problem, including：

Interdependent syntactic analysis is carried out to described problem, the corresponding dependency tree of described problem is obtained；

Entity node, attribute node, interrogative node and core verb from the dependency tree in extraction described problem Node；

According to the dependence between the dependency tree interior joint, by entity node to the attribute between core verb node Node, as first kind attribute node；And,

Determine interrogative node to the attribute node between core verb node, and according to the interrogative node for determining to core Attribute node between heart verb node, determines Equations of The Second Kind attribute node；

Entity node, first kind attribute node and Equations of The Second Kind attribute node are spliced successively, is constituted described problem Corresponding inquiry chain, wherein, each inquiry chain includes an entity node and at least one attribute node, described at least one Attribute node sorts in chain is inquired about according to dependence.

Preferably, in methods described, according to the corresponding inquiry chain of described problem, retrieved in the advance knowledge mapping for building The answer of described problem, including：

Entity node and adjacent attribute node are extracted as an entity attribute key from the corresponding inquiry chain of described problem Value is right, and the adjacent attribute node is that described problem is corresponding inquires about attribute node adjacent with the entity node in chain；

Synonym extension is carried out respectively to the entity node and attribute node in the key-value pair；

Entity attribute key-value pair is rebuild according to the entity node and attribute node after synonym extension；

For the entity attribute key-value pair that each rebuilds, in the advance knowledge mapping for building retrieving this rebuilds The corresponding result of entity attribute key-value pair, and be saved in result set；

Judge whether the adjacent attribute node is corresponding last attribute node inquired about in chain of described problem；

If so, according to the result set, determining the answer of described problem；

If it is not, in the corresponding result of entity attribute key-value pair that this is rebuild and the corresponding inquiry chain of described problem The attribute node not extracted constitutes the corresponding new inquiry chain of described problem, and execution is taken out from the corresponding inquiry chain of described problem The step of entity node and adjacent attribute node are taken as an entity attribute key-value pair, wherein, the result is used as new inquiry Entity node in chain.

Preferably, in methods described, rebuild according to the entity attribute node and attribute node after synonym extension Entity attribute key-value pair, including：

Each entity node of entity node concentration is obtained after being extended for synonym, the entity node and synonym are expanded Each attribute node that attribute node after exhibition is concentrated constitutes entity attribute key-value pair, the entity attribute key assignments for being rebuild To collection.

Preferably, in methods described, according to the result set, the answer of described problem is determined, including：

Determine the total number of word included in described problem；

For each Different Results in the result set, it is determined that in retrieving the dependency tree used during the result The number of words of entity node and attribute node and, as the corresponding number of words of the result and；

The corresponding number of words of the result and the ratio with the total number of word are calculated, as the corresponding ratio of the result；

The maximum result of corresponding ratio in the result set is defined as the answer of described problem.

Preferably, in methods described, according to the interrogative node for determining to the attribute node between core verb node, really Determine Equations of The Second Kind attribute node, including：

Semantic analysis is carried out to described problem, the interrogative node to the attribute node between core verb node is determined Whether need to carry out taking antonym operation；

If so, the interrogative node to the attribute node between core verb node is carried out to take antonym operation, and The attribute node after antonym operation will be taken and be defined as Equations of The Second Kind attribute node；

If it is not, the interrogative node to the attribute node between core verb node is defined as into the second generic attribute section Point.

Preferably, in methods described, the answer of the described problem that the retrieval mode retrieval of knowledge based collection of illustrative plates is obtained is anti- Feed the user：

The answer of described problem is not retrieved according to the retrieval mode based on local problem base, by knowledge based collection of illustrative plates The answer feedback of the described problem that retrieval mode retrieval is obtained is to the user.

Preferably, answer provided in an embodiment of the present invention provides method, also includes：

The way of search based on search engine, the retrieval mode based on Ask-Answer Community knowledge base is respectively adopted, based on depth In the retrieval mode of study, at least one retrieval mode retrieves the answer of described problem, wherein, different retrieval mode correspondences are not Same priority；

If the retrieval mode of knowledge based collection of illustrative plates does not retrieve the answer of the described problem for obtaining, answering for described problem will be returned to The answer feedback of the described problem that the retrieval mode retrieval of highest priority is obtained is to described at least one retrieval mode of case User.

Preferably, in methods described, the priority of at least one retrieval mode order from high to low is：

Retrieval mode based on Ask-Answer Community knowledge base, the retrieval mode based on deep learning, the inspection based on search engine Rope mode.

The answer of described problem is not retrieved according to the retrieval mode based on local problem base, the user is will be feedbacked to The answer of described problem be saved in local problem base.

In the second aspect of embodiment of the present invention, there is provided a kind of answer offer device, including：

Receiver module, the problem for receiving user's submission；

Module is built, the entity node and attribute node extracted from described problem for basis build described problem Corresponding inquiry chain；

First retrieval module, for according to the corresponding inquiry chain of described problem, being retrieved in the advance knowledge mapping for building The answer of described problem, the answer of the described problem obtained as the retrieval mode retrieval of knowledge based collection of illustrative plates；

Feedback module, for the retrieval mode of knowledge based collection of illustrative plates to be retrieved the answer feedback of the described problem for obtaining to institute State user.

Preferably, the structure module, including：

Analytic unit, for carrying out interdependent syntactic analysis to described problem, obtains the corresponding dependency tree of described problem；

Extraction unit, for extracting entity node, attribute node, interrogative section in described problem from the dependency tree Point and core verb node；

First determining unit, for according to the dependence between the dependency tree interior joint, by entity node to core Attribute node with dependence between verb node, as first kind attribute node；

Second determining unit, for determining interrogative node to the attribute node between core verb node, and according to true Fixed interrogative node determines Equations of The Second Kind attribute node to the attribute node between core verb node；

Concatenation unit, for entity node, first kind attribute node and Equations of The Second Kind attribute node to be spliced successively, The corresponding inquiry chain of described problem is constituted, wherein, each inquiry chain includes an entity node and at least one attribute node, At least one attribute node sorts in chain is inquired about according to dependence.

Preferably, the first retrieval module, including：

Extracting unit, for extracting entity node and adjacent attribute node conduct from the corresponding inquiry chain of described problem One entity attribute key-value pair, the adjacent attribute node is that described problem is corresponding inquires about adjacent with the entity node in chain Attribute node；

Expanding element, for carrying out synonym extension respectively to the entity node and attribute node in the key-value pair；

First reconfiguration unit, for being extended according to synonym after entity node and attribute node rebuild entity category Property key-value pair；

Retrieval unit, for the entity attribute key-value pair rebuild for each, in the advance knowledge mapping for building The corresponding result of entity attribute key-value pair for rebuilding is retrieved, and is saved in result set；

Judging unit, for judge the adjacent attribute node whether be described problem it is corresponding inquiry chain in last Individual attribute node；

3rd determining unit, in the case of judging to be in the judging unit, according to the result set, it is determined that described The answer of problem；

Second reconfiguration unit, for the judging unit judge it is no in the case of, the entity attribute that this is rebuild It is corresponding that the attribute node not extracted in the corresponding result of key-value pair and the corresponding inquiry chain of described problem constitutes described problem New inquiry chain, and it is real as one to perform extraction entity node and adjacent attribute node from described problem corresponding inquiry chain The step of body attribute key-value pair, wherein, the result is used as the entity node in new inquiry chain.

Preferably, first reconfiguration unit specifically for：

Preferably, the 3rd determining unit specifically for：

Determine the total number of word included in described problem；

Preferably, second determining unit specifically for：

Preferably, the feedback module specifically for：

Preferably, answer offer device provided in an embodiment of the present invention, also includes：

Second retrieval module, for the way of search based on search engine to be respectively adopted, based on Ask-Answer Community knowledge base In retrieval mode, the retrieval mode based on deep learning, at least one retrieval mode retrieves the answer of described problem, wherein, no Same retrieval mode correspondence different priorities；

The feedback module, if the retrieval mode for being additionally operable to knowledge based collection of illustrative plates does not retrieve answering for the described problem that obtains Case, the retrieval mode of highest priority at least one retrieval mode for returning to the answer of described problem is retrieved described in obtaining The answer feedback of problem gives the user.

Preferably, the priority of at least one retrieval mode order from high to low is：Based on Ask-Answer Community knowledge The retrieval mode in storehouse, the retrieval mode based on deep learning, the retrieval mode based on search engine.

Preserving module, the answer for not retrieving described problem according to the retrieval mode based on local problem base will The answer for feeding back to the described problem of the user is saved in local problem base.

In the third aspect of embodiment of the present invention, there is provided a kind of answer provides equipment, for example, storage can be included Device and processor, wherein, processor can be used for reading the program in memory, perform following process：

Receive the problem that user submits to；

In the fourth aspect of embodiment of the present invention, there is provided a kind of program product, it includes program code, when described When program product runs, described program code is used to perform procedure below：

Receive the problem that user submits to；

Answer according to embodiment of the present invention provides method and apparatus, and the problem that user submits to is understood and divided The entity and attribute being extracted in problem in analysis, and the problem submitted to from user constitute inquiry chain, and in the knowledge for pre-setting The corresponding result of inquiry chain is retrieved in collection of illustrative plates such that it is able to improve retrieve the answer for obtaining for user carries to a certain extent The possibility of the answer of the problem of friendship, and then the accuracy of the answer for feeding back to user is improve to a certain extent.

Brief description of the drawings

Detailed description below, above-mentioned and other mesh of exemplary embodiment of the invention are read by reference to accompanying drawing , feature and advantage will become prone to understand.In the accompanying drawings, if showing by way of example, and not by way of limitation of the invention Dry implementation method, wherein：

Fig. 1 schematically shows the application scenarios schematic diagram according to embodiment of the present invention；

Fig. 2 schematically shows and provides the embodiment schematic flow sheet of method one according to the answer of embodiment of the present invention；

The answer that Fig. 3 schematically shows embodiment of the present invention provides another schematic flow sheet of method；

Fig. 4 schematically shows the schematic flow sheet of the corresponding inquiry chain of Construct question in embodiment of the present invention；

Fig. 5 schematically shows the answer of the search problem in the advance knowledge mapping for building in embodiment of the present invention Schematic flow sheet；

Fig. 6 schematically shows the schematic flow sheet of the answer for determining problem in embodiment of the present invention according to result set；

Fig. 7 schematically shows the structural representation of answer offer device according to an embodiment of the invention；

Fig. 8 schematically shows the structural representation of the answer offer device according to further embodiment of this invention；

The program product for providing method for answer that Fig. 9 schematically shows according to yet another embodiment of the invention is illustrated Figure.

In the accompanying drawings, identical or corresponding label represents identical or corresponding part.

Specific embodiment

Principle of the invention and spirit are described below with reference to some illustrative embodiments.It should be appreciated that providing this A little implementation methods are used for the purpose of better understood when those skilled in the art and then realizing the present invention, and not with any Mode limits the scope of the present invention.Conversely, these embodiments are provided so that the disclosure is more thorough and complete, and energy It is enough that the scope of the present disclosure is intactly conveyed into those skilled in the art.

One skilled in the art will appreciate that embodiments of the present invention can be implemented as a kind of system, device, equipment, method Or computer program product.Therefore, the disclosure can be implemented as following form, i.e.,：Complete hardware, complete software (including firmware, resident software, microcode etc.), or the form that hardware and software is combined.

According to the embodiment of the present invention, it is proposed that a kind of answer treating method and apparatus.

Herein, it is to be understood that any number of elements in accompanying drawing is used to example and unrestricted and any Name is only used for distinguishing, without any limitation.

Below with reference to some representative embodiments of the invention, principle of the invention and spirit are explained in detail.

Summary of the invention

The inventors discovered that, in current retrieval mode, the problem that user submits to will not be understood and be analyzed, and It is that this may result in and feeds back to answering for user directly according to default template or keyword search and the immediate answer of the problem The relatively low problem of the accuracy of case, Consumer's Experience is poor.

Therefore, providing method and apparatus the invention provides a kind of answer, method can include：

Receive the problem that user submits to；

After general principle of the invention is described, lower mask body introduces various non-limiting embodiment party of the invention Formula.

Application scenarios overview

With reference first to Fig. 1, as shown in figure 1, for the application scenarios that answer provided in an embodiment of the present invention provides method are illustrated Figure, including user 10, user terminal 101 and server 102, wherein, the client that user 10 can install in user terminal 101 Input problem is simultaneously submitted to, and server 102 receives the problem that user submits to；According to from described problem extract entity node and Attribute node, builds the corresponding inquiry chain of described problem；According to the corresponding inquiry chain of described problem, in the advance knowledge graph for building The answer of described problem is retrieved in spectrum, the answer of the described problem obtained as the retrieval mode retrieval of knowledge based collection of illustrative plates；Will The answer feedback of the described problem that the retrieval mode retrieval of knowledge based collection of illustrative plates is obtained is to the user 10.Wherein, user terminal Can be communicated by internet with server.User terminal can be mobile phone, panel computer or desktop computer etc..

Illustrative methods

Describe to provide method according to the answer of exemplary embodiment of the invention below with reference to Fig. 2~Fig. 6.Should be noted , embodiments of the present invention can apply to applicable any scene.

The schematic flow sheet of one embodiment of a kind of answer offer method that Fig. 2 is provided for the present invention, mainly including basis The problem that user submits to, to the flow of the corresponding answer of user feedback, the executive agent of the embodiment of the present invention can for server or User terminal, it is also possible to which a part performs a part and performed by user terminal by server, does not limit here.As shown in Fig. 2 A kind of answer provided in an embodiment of the present invention provides method, comprises the following steps：

Step 201, receives the problem that user submits to.

In this step, the problem that user submits to interface to submit to by problem is received.During specific implementation, problem submits interface bag to Two options of text input problem and phonetic entry problem are included, user may be selected any one mode and be input into problem and submit to, when connecing When the problem for receiving user's submission is voice document, Text region is carried out to voice document, corresponding text is obtained, so as to follow-up The problem is analyzed.Prior art specifically is referred to the process that voice carries out Text region, is not detailed here.

Step 202, according to the entity node and attribute node that are extracted from described problem, builds described problem corresponding Inquiry chain.

In this step, word level can be carried out to the corresponding text of problem that user submits to according to existing word level treatment technology Pretreatment, including participle, part-of-speech tagging and name Entity recognition, wherein, name Entity recognition include name, mechanism name, Name and other all entity nodes with entitled mark recognize, the also identification including attribute entities, in obtaining the problem Entity node and attribute node, entity node and attribute node in problem, in the advance special template system of collection of illustrative plates for building Template matches are carried out in system, the corresponding inquiry chain of problem of user's submission is obtained, in the special template system of collection of illustrative plates for building in advance Saving various types of inquiry chain ATLs such as includes personage's ATL, time template storehouse, situation template storehouse, organization ATL etc..Save the inquiry chain template relevant with personage in such as personage's ATL, inquiry chain template can include personage- Date of birth, personage-sex, personage-occupation etc., the entity node in problem is detected are personage and attribute node and birth When date is related, it may be determined that the corresponding inquiry chain template of the problem is personage-date of birth, in the problem submitted to further according to user Specific task builds the corresponding inquiry chain of the problem, will personage replace with specific name.Wherein, for same answer The different same inquiry chain templates of question formulation correspondence, such as in March, 2016 was the date of birth of Zhang San, for the date of birth of Zhang San This answer of phase, can corresponding question formulation include Zhang San be born when, birthdate of Zhang San be when or Zhang San birthday Be when, after the advance collection of illustrative plates specialized system for building, it may be determined that the corresponding inquiry chain of these three different question formulations is equal It is Zhang San-date of birth.

Wherein, the entity node and attribute node in problem, enters in the advance special template system of collection of illustrative plates for building The process of row template matches is existing process, is not detailed here.

Illustrate, it is assumed that user submit to problem be：When the birthday of Zhang San isWord level treatment is carried out to the problem Afterwards, Zhang San is obtained for people entities, birthday are attribute entities, then according to people entities and attribute entities in the advance collection of illustrative plates for building Template matches are carried out in special ATL system, finally determines to match with the date of birth template in personage's ATL, and really The corresponding template query chain of date of birth template is determined for personage-date of birth, now, extracts Zhang San this personage in the problem Entity simultaneously inquires about chain with date of birth composition, finally gives the corresponding inquiry chain of the problem, i.e. Zhang San-date of birth.

Also the entity node and attribute node in the problem that user submits to can be extracted using other manner, described problem is built Corresponding inquiry chain, does not limit here.

Wherein, the entity node of extraction is theoretically the entity node in the knowledge mapping built in advance in step 203, takes out Attribute node in the knowledge mapping built in advance in the attribute node for taking theoretically step 203.

Step 203, according to the corresponding inquiry chain of described problem, described problem is retrieved in the advance knowledge mapping for building Answer, the answer of the described problem obtained as the retrieval mode retrieval of knowledge based collection of illustrative plates.

During specific implementation, the corresponding inquiry chain of problem that user submits to includes entity node and attribute section in the problem Point, the entity node in retrieval and inquisition chain in the advance knowledge mapping for building；It is determined that being examined in the advance knowledge mapping for building Rope inquires about the category in the inquiry chain after entity node to after the entity node in inquiry chain in the advance knowledge mapping for building Property, i.e., the side being connected with the entity node for inquiring；Further, it is determined that the side being connected with the entity node for inquiring is another The entity node of side connection is the corresponding answer of inquiry chain, such as, the inquiry chain is that (corresponding user submits to Zhang San-wife Problem be：Whom the wife of Zhang San is), then this entity node of Zhang San, Zhi Hou are inquired about in the advance knowledge mapping for building Searched in the knowledge mapping for building in advance be connected with Zhang San this entity node for embodying the side of wife, and determine for body The entity node Li Si of the opposite side on the side of existing wife is Lee for the inquiry chain Zhang San-corresponding answer of wife, i.e. wife of Zhang San Four.

It should be noted that knowledge mapping is regarded as a huge figure, the node in figure is entity node, in figure The side being connected with any entity node represents the entity attributes, the opposite side connection on the side being connected with any entity node Entity is the corresponding specific object value of any entity node, such as, any entity node is Zhang San, with any entity section The side of point connection represents wife, then the entity of the opposite side connection on the side being connected with the Zhang San is the sub- wife of Zhang San, i.e., specific category Property value be Li Si.Specifically, the entity in knowledge mapping can be comprising concept, formula, name, place name, organization etc., knowledge Attribute can be the spy that relation or the entity between two entities of the corresponding side connection of the attribute possess in itself in collection of illustrative plates Property.The embodiment of the present invention is applied in the case of education of middle and primary schools question answering system, and knowledge graph can be in advance built in the following manner Spectrum：

First, obtain structural data.

Wherein, the mode for obtaining structural data includes：1st, from the database for prestoring in acquisition and K12 teaching materials The related structural data of knowledge；2nd, by reptile obtain in Baidupedia, wikipedia and interactive encyclopaedia with K12 teaching material phases The structural data of pass；3rd, extracted in answering questions record from the teacher for prestoring by way of machine learning and template are combined The structural data answerred questions in record, wherein, the teacher for prestoring answers questions in record, and to save teacher right in course of education The manual answering of the problem of student question；4th, obtained from the unstructured data in some Educational websites or education Ask-Answer Community Education relevant knowledge, and therefrom extract structural data.

Second, knowledge, concept and formula in the structural data of acquisition etc. as the entity node in knowledge mapping, For each entity node, the specific object value of the entity node is connected by attribute side, so as to obtain education sector knowledge graph Spectrum.

Wherein, the attribute of entity node is embodied by the side being connected with the entity node, the side being connected with the entity node Another side connection entity node be the entity node specific object value, also be understood as between any two entity node Relation embodied by the side between the two entity nodes, in knowledge mapping building process, such as run into for same entity The different expression ways of node, it is necessary to carry out duplicate removal treatment to entity node, if two title phases of different entity nodes Together, then entity node is made a distinction by entity node specific attribute, and is embodied in knowledge mapping respectively.

It should be noted that the real knowledge mapping for building in advance is to an entity node list and an attribute node row Table, entity node list is used to store the entity node in the knowledge mapping of advance structure, and attribute node list is used to store pre- Attribute node in the knowledge mapping for first building.

Step 204, the answer feedback of the described problem that the retrieval mode retrieval of knowledge based collection of illustrative plates is obtained is to the use Family.

During specific implementation, other retrieval sides higher than the retrieval mode priority of knowledge based collection of illustrative plates can determined with Formula, in the case of the answer of the problem for not retrieving user's submission, the institute that the retrieval mode retrieval of knowledge based collection of illustrative plates is obtained The answer feedback of problem is stated to the user.Or, it is determined that being submitted to only with the retrieval mode retrieval user of knowledge based collection of illustrative plates Problem in the case of, the answer feedback of the described problem for directly obtaining the retrieval of the retrieval mode of knowledge based collection of illustrative plates is to described User.

The embodiment that Fig. 2 is provided, is understood and has been analyzed to the problem that user submits to, and in the problem submitted to from user The entity and attribute being extracted in problem constitute inquiry chain, and it is corresponding that the inquiry chain is retrieved in the knowledge mapping for pre-setting As a result such that it is able to improve the possibility of the answer for retrieving the problem that the answer for obtaining is submitted to for user to a certain extent, And then the accuracy of the answer for feeding back to user is improve to a certain extent.

It should be noted that the answer of the problem that can be submitted to using various retrieval modes retrieval user simultaneously, wherein, can be pre- It is first every kind of retrieval mode setting priority, and in the retrieval mode of the answer that will retrieve the problem of user's submission, priority The answer feedback of the problem that the user that the retrieval of highest retrieval mode is obtained submits to is to user.

Preferably, the answer of the described problem for as follows obtaining the retrieval mode retrieval of knowledge based collection of illustrative plates is anti- Feed the user：

Wherein, while using the retrieval mode based on local problem base and the retrieval mode of knowledge based collection of illustrative plates, retrieval The answer of the problem that user submits to, wherein, the priority of the retrieval mode based on local problem base is higher than knowledge based spectrogram The priority of retrieval mode, i.e. the preferential problem for submitting the user that the retrieval mode retrieval based on local problem base is obtained to Answer feedback is to user.

If the mode of retrieval mode and knowledge based collection of illustrative plates based on local problem base does not retrieve asking for user's submission The answer of topic, it is possible that situation of the search less than the answer of the problem that user submits to, can be use to be further ensured that Family furnishes an answer, and answer provided in an embodiment of the present invention provides method, can also comprise the following steps, as shown in Figure 3：

Step 301, be respectively adopted the way of search based on search engine, the retrieval mode based on Ask-Answer Community knowledge base, In retrieval mode based on deep learning, at least one retrieval mode retrieves the answer of described problem, wherein, different retrieval sides Formula correspondence different priorities.

During specific implementation, the embodiment of the present invention is except using retrieval mode and knowledge based spectrogram based on local problem base Retrieval mode retrieval user submit to problem answer, while the way of search based on search engine being also respectively adopted, being based on At least one retrieval mode retrieval in the retrieval mode of Ask-Answer Community knowledge base and the retrieval mode based on deep learning is used The answer of the problem that family is submitted to, it is preferable that while the way of search based on search engine being also respectively adopted, being known based on Ask-Answer Community Know the answer of the problem of retrieval mode retrieval user's submission in the retrieval mode and the retrieval mode three based on deep learning in storehouse. Wherein, the way of search based on search engine, the retrieval mode based on Ask-Answer Community knowledge base and the inspection based on deep learning In rope mode, the different priority of every kind of retrieval mode correspondence is preferentially retrieved to user feedback priority retrieval mode high Answer.

It should be noted that the way of search based on search engine is：Directly invoke the search engine retrieving of current main flow The answer of the problem that user submits to, and obtain in the answer of search engine return in the first order answer, submitted to as user Problem answer；Retrieval mode based on Ask-Answer Community knowledge base is：Using substantial amounts of in reptile instrument periodic collection network Question and answer resource, and by question and answer resource conservation to Ask-Answer Community knowledge base in the way of key-value pair；Retrieval side based on deep learning Formula is：In advance build deep learning model, using user submit to problem as deep learning model input, so as to obtain depth The output of learning model, the answer of the problem that the output is submitted to as user.Wherein, the way of search based on search engine, base Retrieval mode in Ask-Answer Community knowledge base, the process that implements of the retrieval mode based on deep learning refer to existing skill Art, is not described in detail here.

Step 302, it is described by returning if the retrieval mode of knowledge based collection of illustrative plates does not retrieve the answer of the described problem for obtaining The answer of the described problem that the retrieval mode retrieval of highest priority is obtained is anti-at least one retrieval mode of the answer of problem Feed the user.

In this step, the priority of the retrieval mode of knowledge based collection of illustrative plates is higher than way of search, base based on search engine Any retrieval in retrieval mode in Ask-Answer Community knowledge base and the retrieval mode based on deep learning these three retrieval modes The priority of mode, in the case where the retrieval mode of knowledge based collection of illustrative plates does not retrieve the answer of the described problem for obtaining, will return The described problem that the retrieval mode retrieval of highest priority is obtained at least one retrieval mode of the answer for returning described problem Answer feedback gives the user.Preferably, the priority of at least one retrieval mode order from high to low is：Based on asking Answer retrieval mode, the retrieval mode based on deep learning, the retrieval mode based on search engine of community's knowledge base.

Illustrate, it is assumed that in step 301, the way of search based on search engine is respectively adopted, known based on Ask-Answer Community The answer that the retrieval mode in storehouse, retrieval mode based on deep learning retrieve described problem is known, in determining these three retrieval modes Return to the retrieval mode of the answer of the problem that user submits to；If the retrieval mode of knowledge based collection of illustrative plates is not retrieved obtain described and is asked The answer of topic, then priority retrieval higher in the retrieval mode of the answer of the problem submitted return user in these three modes to The answer feedback that mode is retrieved is to user, it is assumed that the retrieval of the answer of the problem that user submits to is returned in these three retrieval modes Mode is the retrieval mode based on Ask-Answer Community knowledge base, the retrieval mode based on deep learning, then will be known based on Ask-Answer Community The answer that the retrieval mode retrieval in knowledge storehouse is obtained returns to user.

It should be noted that the high to Low order of the priority of various retrieval modes can be：Based on local problem base Retrieval mode, the retrieval mode of knowledge based spectrogram, the retrieval mode based on Ask-Answer Community knowledge base, the inspection based on deep learning Rope mode, the retrieval mode based on search engine.Wherein, when the retrieval mode of higher priority retrieves the problem of user's submission Answer when, indicate the relatively low retrieval mode of priority to stop the operation of the answer of the problem that retrieval user submits to, and will be higher The answer feedback that the retrieval mode of priority is retrieved is to user.That is, not needing simultaneously all retrieval modes to return to use The answer of the problem that family is submitted to, if the retrieval mode of high priority has returned to the answer of the problem of user's submission, and now also The operation of the answer of the problem submitted in the retrieval mode retrieval user using low priority, then may indicate that stopping using low preferential The retrieval mode of level retrieves the operation of answer, so can to a certain extent save amount of calculation.

Preferably, the answer of described problem is not retrieved according to the retrieval mode based on local problem base, is will be feedbacked to The answer of the described problem of the user is saved in local problem base.

Wherein, the retrieval mode highest priority based on local problem base, the answer that will can finally feed back to user is preserved To local problem base, during specific implementation, the corresponding inquiry chain of problem that can be submitted user to is corresponding with the answer for feeding back to user Local problem base is stored, or, the text storage value corresponding with the answer for feeding back to user of the problem that user is submitted to is local Problem base, does not limit here.

Preferably, the content for being provided according to Fig. 4, according to the entity node and attribute node that are extracted from described problem, Build the corresponding inquiry chain of described problem：

Step 401, interdependent syntactic analysis is carried out to described problem, obtains the corresponding dependency tree of described problem.

In this step, the text of the problem submitted to user carries out interdependent syntactic analysis, so as to obtain asking for user's submission Inscribe corresponding dependency tree.After the problem that user submits to carries out interdependent syntactic analysis, the part of speech and each word of each word are can obtain Between dependence, wherein, dependency tree can also be called interdependent syntax tree, the node in dependency tree be sentence in word, it is interdependent Set for embodying the dependence between the word in sentence.

Step 402, from the dependency tree extract described problem in entity node, attribute node, interrogative node and Core verb node.

Specifically, for user submit to problem set up entity node list, attribute node list, interrogative node listing with And core verb node list, in the corresponding node listing of node storage value that will be extracted from dependency tree, specifically, can use Name the entity node and attribute node of the mode from dependency tree in extraction problem of Entity recognition and store corresponding node row Table, can also extract the entity node and attribute node in dependency tree using other manner, not limit here.Retrieving user After the answer of the problem of submission, delete and entity node list, attribute node list, interrogative are set up by the problem that user submits to Node listing and core verb node list.

Step 403, according to the dependence between the dependency tree interior joint, by entity node to core verb node it Between attribute node, as first kind attribute node.

Step 404, determines interrogative node to the attribute node between core verb node, and according to the interrogative for determining Node determines Equations of The Second Kind attribute node to the attribute node between core verb node.

During specific implementation, step 403 and step 404 execution sequence can not be limited.

According to the dependence between dependency tree interior joint, by entity node in the problem of user's submission to core verb section Attribute node is used as first kind attribute node, the attribute node nearest apart from entity node in first kind attribute node between point There is dependence with the entity node, there is dependence between first kind attribute node.According between dependency tree interior joint Dependence determine interrogative node to the attribute node between core verb node, it is and dynamic to core according to interrogative node Attribute node between word node, determines Equations of The Second Kind attribute node, and Equations of The Second Kind attribute node is dynamic to core to interrogative node Attribute node between word node processed after attribute node, such as to interrogative node between core verb node have After the attribute node for having dependence carries out taking synonym or antonym operation, Equations of The Second Kind attribute node is obtained.

Preferably, in such a way according to the interrogative node for determining to the attribute node between core verb node, Determine Equations of The Second Kind attribute node：Semantic analysis is carried out to described problem, determine the interrogative node to core verb node it Between attribute node whether need to carry out to take antonym and operate；If so, by the interrogative node between core verb node Attribute node take antonym operation, and the attribute node that will be taken after antonym operation is defined as Equations of The Second Kind attribute node； If it is not, the interrogative node is defined as into Equations of The Second Kind attribute node to the attribute node between core verb node.

Illustrate, it is assumed that the problem that user submits to is " whose husband the younger brother of the wife of Zhang San is ", then entity node Be " Zhang San ", core verb be "Yes", entity node to the attribute node between core verb node be " wife ", " younger brother ", The attribute node that interrogative node " who " is arrived between core verb node "Yes" is " husband ", now can be according to the language of the problem Justice, determines that attribute node " husband " can be exchanged into attribute node " wife ", accordingly, it is determined that Equations of The Second Kind attribute node be " wife Son ", i.e. interrogative node is converted to " wife " to the attribute node between core verb node.Wherein, by the antisense of " husband " Word is defined as " wife ".

Step 405, is successively spliced entity node, first kind attribute node and Equations of The Second Kind attribute node, is constituted The corresponding inquiry chain of described problem, wherein, each inquiry chain includes an entity node and at least one attribute node, described At least one attribute node sorts in chain is inquired about according to dependence.

During specific implementation, the dependence according to attribute node in dependency tree, successively by entity node, the first generic attribute Node and the second generic attribute are spliced, and obtain the corresponding inquiry chain of problem of user's submission, such as the problem that user submits to " whose husband the younger brother of the wife of Zhang San is " corresponding inquiry chain is " Zhang San-wife-younger brother-wife ", wherein, in inquiry chain First wife depends on entity node Zhang San, younger brother to depend on the first wife in inquiry chain, second in inquiry chain Wife depends on younger brother.

The embodiment that Fig. 4 is provided, interdependent syntactic analysis is carried out by the problem submitted to user, and can reach is carried out to problem Analysis and the purpose for understanding, and the therefrom entity node and attribute node in extraction problem, build the problem submitted to user more Stick on and cut, reliably inquire about chain, so as to the answer according to inquiry chain search problem so that the accuracy of the answer that retrieval is obtained is more It is high.

Preferably, according to Fig. 5 offer content, according to the corresponding inquiry chain of described problem, in the advance knowledge for building The answer of described problem is retrieved in collection of illustrative plates：

Step 501, extraction entity node and adjacent attribute node are real as one from described problem corresponding inquiry chain Body attribute key-value pair, the adjacent attribute node is that described problem is corresponding inquires about attribute adjacent with the entity node in chain Node.

During specific implementation, user submit to problem it is corresponding inquiry chain may have one be likely to it is a plurality of, for each Inquiry chain, the inquiry chain includes an entity node and at least one attribute node, entity node and each attribute section Press impinge upon user submission problem in dependence sequence, for user submit to the corresponding every inquiry chain of problem, from Entity node and the attribute node adjacent with the entity node are extracted in the inquiry chain as an entity attribute key-value pair.

Step 502, synonym extension is carried out to the entity node and attribute node in the key-value pair respectively.

During specific implementation, different users may state the entity in entity attribute key-value pair with different expression ways Entity node " Zhang San " in node and attribute node, such as " whose husband the younger brother of the wife of Zhang San is " this problem and Can be expressed as " Xiao Ming " etc., attribute node " wife " can be expressed as " wife " etc. again, in order to ensure in the advance knowledge for building Entity node and attribute node are retrieved in collection of illustrative plates, by the entity node and attribute node in entity attribute key-value pair in this step Synonym extension is carried out respectively.

In this step, the entity node and attribute node in entity attribute key-value pair can be entered by general synonym table Row synonym extends, and, by the advance corresponding collection of illustrative plates special attribute expansion table of knowledge mapping for building to entity attribute key The attribute node of value centering carries out synonym extension, and, by the advance corresponding collection of illustrative plates application-specific entity of knowledge mapping for building Expansion table carries out synonym extension to the entity node in entity attribute key-value pair.Wherein, during knowledge mapping is built, For each entity node in knowledge mapping, by the form of presentation and the entity node in knowledge mapping of the entity node Corresponding synonym correspondence is preserved into collection of illustrative plates application-specific entity expansion table, and, for each attribute node in knowledge mapping, By the synonym corresponding with the attribute node of the form of presentation in knowledge mapping of the attribute node it is corresponding preserve it is special to collection of illustrative plates With in attribute extension table.

Step 503, entity attribute key-value pair is rebuild according to the entity node and attribute node after synonym extension.

During specific implementation, each form of presentation composition entity node collection of the entity node after synonym extension, synonym Each expression way composition attribute node collection of the attribute node after extension, it is preferable that entity node is concentrated to be included rebuilding The entity node in entity attribute key-value pair before, attribute node concentrates the entity attribute key-value pair before including rebuilding In entity node.

Preferably, entity attribute key-value pair can in such a way be rebuild：

Such as, the entity node " Zhang San " in " whose husband the younger brother of the wife of Zhang San is " this problem is carried out synonymous The entity node obtained after word extension is concentrated and includes " Zhang San ", " Xiao Ming ", after carrying out synonym extension to attribute node " wife " The entity node for obtaining is concentrated includes " wife ", " wife ", and the entity attribute key-value pair for rebuilding is concentrated includes " Zhang San-wife Son ", " Xiao Ming-wife ", " Zhang San-wife ", " Xiao Ming-wife " this four entity attribute key-value pairs.

Step 504, for the entity attribute key-value pair that each rebuilds, retrieval should in the advance knowledge mapping for building The corresponding result of entity attribute key-value pair for rebuilding, and it is saved in result set.

In this step, for each key-value pair for rebuilding the entity attribute key-value pair concentration for obtaining, built in advance Knowledge mapping in retrieve the corresponding result of entity attribute key-value pair for rebuilding, and be saved in result set.

Step 505, judges whether the adjacent attribute node is corresponding last category inquired about in chain of described problem Property node, if so, perform step 506, otherwise, perform step 507.

During specific implementation, multiple attribute nodes are potentially included in the corresponding inquiry chain of problem that user submits to, in this step, Judge the adjacent attribute node in step 501 whether be user submit to problem it is corresponding inquiry chain in last attribute section Point.

Step 506, according to the result set, determines the answer of described problem.

Multiple results are saved in this step, in result set, in the corresponding inquiry chain of the problem that will can be submitted to user most The result that the corresponding entity attribute key-value pair of latter attribute node is obtained after being retrieved is defined as the problem of user's submission Answer, when the corresponding inquiry chain of the problem that user submits to has a plurality of, can be by last attribute node pair in each inquiry chain The result that the entity attribute key-value pair answered is obtained after being retrieved carries out duplicate removal treatment, and result after duplicate removal is processed is defined as The answer of the problem that user submits to.

Step 507, the corresponding result of entity attribute key-value pair that this is rebuild and the corresponding inquiry of described problem The attribute node not extracted in chain constitutes the corresponding new inquiry chain of described problem, and continues executing with step 501.

Wherein, the corresponding result of entity attribute key-value pair that this rebuilds is used as the entity node in new inquiry chain.

In this step, if the adjacent attribute node in step 501 is not last attribute node inquired about in chain, will The category not being extracted in the problem that the corresponding result of entity attribute key-value pair and user that this rebuilds in step 504 are submitted to Property node constitute new inquiry chain, and step 501 is continued executing with, until the adjacent attribute node in step 501 is in inquiry chain Untill last attribute node.

Illustrate, it is assumed that the problem that user submits to is " whose husband the younger brother of the wife of Zhang San is ", then the problem pair The inquiry chain answered is " Zhang San-wife-younger brother-wife ", extract first entity node " Zhang San " in the inquiry chain and with reality The adjacent attribute node of body node " wife ", obtains entity attribute key-value pair " Zhang San-wife ", and to the entity in the key-value pair After node and attribute node carry out synonym extension, for each weight that the entity attribute key-value pair for rebuilding for obtaining is concentrated The new entity attribute key-value pair for building, the entity attribute key-value pair pair for rebuilding is retrieved in the advance instruction collection of illustrative plates for building The result " Li Si " answered；" wife " behind " Zhang San " is not last attribute node inquired about in chain, therefore, by " Lee Four " this result is configured to new inquiry chain with remaining attribute node " younger brother-wife " in inquiry chain, and looks into new chain from new Middle extraction " Li Si-younger brother " is carried out together as entity attribute key-value pair to the entity node and attribute node in the key-value pair After adopted word extension, for each entity attribute key assignments for rebuilding that the entity attribute key-value pair for rebuilding for obtaining is concentrated It is right, retrieve the corresponding result of the entity attribute key-value pair " Zhang San " for rebuilding in the advance instruction collection of illustrative plates for building；" Lee Four " " younger brother " behind is not last attribute node inquired about in chain, therefore, by " Zhang San " this result and inquiry chain In remaining attribute node " wife " be configured to new inquiry chain, and extract " Zhang San-wife " as real from new looking into new chain Body attribute key-value pair, and after carrying out synonym extension to the entity node and attribute node in the key-value pair, for the weight for obtaining Each entity attribute key-value pair for rebuilding that the new entity attribute key-value pair for building is concentrated, in the advance instruction collection of illustrative plates for building It is middle to retrieve the corresponding result of entity attribute key-value pair " king five " for rebuilding.

The embodiment that Fig. 5 is provided, can be in the case where the corresponding inquiry chain of problem includes multiple attribute nodes, can be with The multilevel query of the problem for user's submission is realized, is most accurately answered so as to be chosen from the result set obtained by multilevel query Case feeds back to user.

Preferably, the content that can be provided according to Fig. 6, according to the result set, determines the answer of described problem：

Step 601, determines the total number of word included in described problem.

During specific implementation, the total number of word in the corresponding text of problem that user submits to is determined.Such as problem " the wife of Zhang San Younger brother be whose husband " in total number of word be 12.

Step 602, for each Different Results in the result set, it is determined that retrieve used during the result it is described according to Deposit entity node in tree and attribute node number of words and, as the corresponding number of words of the result and.

In this step, for each result in result set, it is determined that in retrieving the dependency tree used during the result The number of words of entity node and attribute node and, retrieve result such as problem " whose husband the younger brother of the wife of Zhang San is " The number of words of entity node " Zhang San " and attribute node " wife " in the dependency tree used when " Li Si " and be 4, then " Li Si " This and the corresponding number of words of result and be 4, retrieve used during result " king five " (i.e. the wife of the younger brother of Li Si) interdependent Entity node " Zhang San " and attribute node " wife ", " younger brother ", the number of words of " wife " in tree and be 8, then " king five " this knot Really corresponding number of words and be 8.

Step 603, calculates the corresponding number of words of the result and the ratio with the total number of word, used as the corresponding ratio of the result Value.

Step 604, the maximum result of corresponding ratio in the result set is defined as the answer of described problem.

During specific implementation, the corresponding number of words of each result and the ratio with total number of word are calculated, and by the maximum result of ratio It is defined as the answer of the problem of user's submission, continues to use the example above, " Li Si " this and corresponding number of words of result and be 4, then " Lee Four " the corresponding ratio of this result is 4/12, the corresponding number of words of " king five " this result and be 8, then " king five " this result pair The ratio answered is 8/12, now, " king five " is defined as into problem " whose husband the younger brother of the wife of Zhang San is " this problem Answer.

Example devices

After the answer offer method for describing exemplary embodiment of the invention, next, describing this hair with reference to Fig. 7 Bright illustrative embodiments answer offer device.

Fig. 7 is a kind of one of structural representation of answer offer device provided in an embodiment of the present invention, as shown in fig. 7, can With including such as lower module：

Receiver module 701, the problem for receiving user's submission；

Build module 702, the entity node and attribute node extracted from described problem for basis, structure described in ask Inscribe corresponding inquiry chain；

First retrieval module 703, for according to the corresponding inquiry chain of described problem, being examined in the advance knowledge mapping for building The answer of rope described problem, the answer of the described problem obtained as the retrieval mode retrieval of knowledge based collection of illustrative plates；

Feedback module 704, the answer feedback of the described problem for the retrieval mode retrieval of knowledge based collection of illustrative plates to be obtained To the user.

Preferably, the structure module 702, including：

Analytic unit 7021, for carrying out interdependent syntactic analysis to described problem, obtains the corresponding dependency tree of described problem；

Extraction unit 7022, for extracting entity node, attribute node, query in described problem from the dependency tree Word node and core verb node；

First determining unit 7023, for according to the dependence between the dependency tree interior joint, entity node being arrived Attribute node with dependence between core verb node, as first kind attribute node；

Second determining unit 7024, for determining interrogative node to the attribute node between core verb node, and root According to the interrogative node for determining to the attribute node between core verb node, Equations of The Second Kind attribute node is determined；

Concatenation unit 7025, for successively carrying out entity node, first kind attribute node and Equations of The Second Kind attribute node Splicing, constitutes the corresponding inquiry chain of described problem, wherein, each inquiry chain includes an entity node and at least one attribute Node, at least one attribute node sorts in chain is inquired about according to dependence.

Preferably, the first retrieval module 703, including：

Extracting unit 7031, for extracting entity node and adjacent attribute node from the corresponding inquiry chain of described problem As an entity attribute key-value pair, the adjacent attribute node be in the corresponding inquiry chain of described problem with the entity node phase Adjacent attribute node；

Expanding element 7032, for carrying out synonym expansion respectively to the entity node and attribute node in the key-value pair Exhibition；

First reconfiguration unit 7033, for being extended according to synonym after entity node and attribute node rebuild reality Body attribute key-value pair；

Retrieval unit 7034, for the entity attribute key-value pair rebuild for each, in the advance knowledge graph for building The corresponding result of entity attribute key-value pair for rebuilding is retrieved in spectrum, and is saved in result set；

Judging unit 7035, inquires about in chain most for judging whether the adjacent attribute node is that described problem is corresponding Latter attribute node；

3rd determining unit 7036, in the case of judging to be in the judging unit 7035, according to the result set, Determine the answer of described problem；

Second reconfiguration unit 7037, for the judging unit 7035 judge it is no in the case of, this is rebuild The attribute node not extracted in the corresponding result of entity attribute key-value pair and the corresponding inquiry chain of described problem is asked described in constituting Corresponding new inquiry chain is inscribed, and performs extraction entity node and adjacent attribute node from described problem corresponding inquiry chain The step of as an entity attribute key-value pair, wherein, the result is used as the entity node in new inquiry chain.

Preferably, first reconfiguration unit 7033 specifically for：

Preferably, the 3rd determining unit 7036 specifically for：

Determine the total number of word included in described problem；

Preferably, second determining unit 7024 specifically for：

Preferably, the feedback module 704 specifically for：

Second retrieval module 705, for the way of search based on search engine to be respectively adopted, based on Ask-Answer Community knowledge base Retrieval mode, the retrieval mode based on deep learning in, at least one retrieval mode retrieves the answer of described problem, wherein, Different retrieval mode correspondence different priorities；

The feedback module 704, if the retrieval mode for being additionally operable to knowledge based collection of illustrative plates does not retrieve the described problem that obtains Answer, the institute that the retrieval mode retrieval of highest priority at least one retrieval mode for returning to the answer of described problem is obtained The answer feedback of problem is stated to the user.

Preserving module 706, the answer for not retrieving described problem according to the retrieval mode based on local problem base, The answer that will be feedbacked to the described problem of the user is saved in local problem base.

Example devices

After the answer for describing exemplary embodiment of the invention provides method and apparatus, next, introducing basis The answer offer device of another exemplary implementation method of the invention.

Person of ordinary skill in the field it is understood that various aspects of the invention can be implemented as system, method or Program product.Therefore, various aspects of the invention can be implemented as following form, i.e.,：It is complete hardware embodiment, complete Full Software Implementation (including firmware, microcode etc.), or the implementation method that hardware and software aspect is combined, can unite here Referred to as " circuit ", " module " or " system ".

In some possible implementation methods, answer offer device of the invention at least can include at least one Reason unit and at least one memory cell.Wherein, the memory cell has program stored therein code, when described program code quilt When the processing unit is performed so that the processing unit performs the root described in above-mentioned " illustrative methods " part of this specification According to the step in the answer offer method of various illustrative embodiments of the invention.For example, the processing unit can be performed such as The step of shown in Fig. 2 201, receive the problem that user submits to, step 202, according to the entity node extracted from described problem And attribute node, build the corresponding inquiry chain of described problem, step 203, according to the corresponding inquiry chain of described problem, advance The answer of described problem is retrieved in the knowledge mapping of structure, is retrieved as the retrieval mode of knowledge based collection of illustrative plates and is asked described in obtaining The answer of topic and step 204, the answer feedback of the described problem that the retrieval mode retrieval of knowledge based collection of illustrative plates is obtained is to institute State user.

The answer offer device 80 of this implementation method of the invention is described referring to Fig. 8.What Fig. 8 showed answers Case offer device 80 is only an example, should not carry out any limitation to the function of the embodiment of the present invention and using range band.

As shown in figure 8, answer offer device 80 is showed in the form of universal computing device.The component of answer offer device 80 Can include but is not limited to：Above-mentioned at least one processing unit 801, above-mentioned at least one memory cell 802, connection different system The bus 803 of component (including processing unit 801 and memory cell 802).

Bus 803 represents one or more in a few class bus structures, including memory bus or Memory Controller, Peripheral bus, processor or the local bus using any bus structures in various bus structures.

Memory cell 802 can include the computer-readable recording medium of form of volatile memory, such as random access memory (RAM) 8021 and/or cache memory 8022, can further include read-only storage (ROM) 8023.

Memory cell 802 can also include thering is one group of program/utility of (at least one) program module 8024 8025, such program module 8024 is included but is not limited to：Operating system, one or more application program, other program moulds Block and routine data, potentially include the realization of network environment in each or certain combination in these examples.

Answer offer device 80 can also communicate with one or more external equipments 804 (such as keyboard, sensing equipment etc.), The equipment communication that is interacted with answer offer device 80 can also be enabled a user to one or more, and/or with cause the answer Any equipment (such as router, modem that offer device 80 can be communicated with one or more of the other computing device Etc.) communication.This communication can be carried out by input/output (I/O) interface 805.Also, answer offer device 80 can be with By network adapter 806 and one or more network (such as LAN (LAN), wide area network (WAN) and/or public network, Such as internet) communication.As shown in figure 8, network adapter 806 by bus 803 with for answer provide device 80 its Its module communicates.It will be appreciated that though not shown in figure, answer offer device 80 can be combined and use other hardware and/or soft Part module, including but not limited to：Microcode, device driver, redundant processing unit, external disk drive array, RAID system, Tape drive and data backup storage system etc..

Exemplary process product

In some possible implementation methods, the various aspects that the answer that the present invention is provided provides method are also implemented as A kind of form of program product, it includes program code, when described program product runs on a computing device, described program Code is each according to the present invention described in above-mentioned " illustrative methods " part of this specification for performing the computer equipment The step in the answer offer method of illustrative embodiments is planted, for example, the computer equipment can be performed as shown in Figure 2 The step of 201, receive user submit to problem, step 202, according to from described problem extract entity node and attribute section Point, builds the corresponding inquiry chain of described problem, step 203, according to the corresponding inquiry chain of described problem, in the advance knowledge for building The answer of described problem is retrieved in collection of illustrative plates, the answer of described problem obtained as the retrieval mode retrieval of knowledge based collection of illustrative plates with And step 204, the answer feedback of the described problem that the retrieval mode retrieval of knowledge based collection of illustrative plates is obtained is to the user.

Described program product can be using any combination of one or more computer-readable recording mediums.Computer-readable recording medium can be readable letter Number medium or readable storage medium storing program for executing.Readable storage medium storing program for executing for example may be-but not limited to-electricity, magnetic, optical, electromagnetic, red The system of outside line or semiconductor, device or device, or it is any more than combination.The more specifically example of readable storage medium storing program for executing (non exhaustive list) includes：Electrical connection, portable disc with one or more wires, hard disk, random access memory (RAM), read-only storage (ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disc Read memory (CD-ROM), light storage device, magnetic memory device or above-mentioned any appropriate combination.

As shown in figure 9, describing the program product 90 provided for answer according to the embodiment of the present invention, it can be with Using portable compact disc read only memory (CD-ROM) and including program code, it is possible to run on the terminal device.However, Program product not limited to this of the invention, in this document, readable storage medium storing program for executing can be it is any comprising or storage program have Shape medium, the program can be commanded execution system, device or device and use or in connection.

Readable signal medium can include the data-signal propagated in a base band or as a carrier wave part, wherein carrying Readable program code.The data-signal of this propagation can take various forms, including --- but being not limited to --- electromagnetism letter Number, optical signal or above-mentioned any appropriate combination.Readable signal medium can also be beyond readable storage medium storing program for executing it is any can Read medium, the computer-readable recording medium can send, propagate or transmit for by instruction execution system, device or device use or Program in connection.

The program code included on computer-readable recording medium can be transmitted with any appropriate medium, including --- but being not limited to --- Wirelessly, wired, optical cable, RF etc., or above-mentioned any appropriate combination.

Program for performing present invention operation can be write with one or more any combination of programming language Code, described program design language includes object oriented program language-Java, C++ etc., also including routine Procedural programming language-such as " C " language or similar programming language.Program code can fully in user Performed on computing device, partly performed on a user device, calculated as an independent software kit execution, partly in user Its upper side point is performed or performed in remote computing device or server completely on a remote computing.It is being related to far In the situation of journey computing device, remote computing device can be by the network of any kind --- including LAN (LAN) or wide Domain net (WAN)-be connected to user calculating equipment, or, it may be connected to external computing device is (such as using Internet service Provider comes by Internet connection).

Although it should be noted that being referred to some units or subelement of device, this stroke in above-detailed What is point be merely exemplary is not enforceable.In fact, according to the embodiment of the present invention, it is above-described two or more The feature and function of unit can embody in a unit.Conversely, the feature and function of an above-described unit can To be further divided into being embodied by multiple units.

Additionally, although the operation of the inventive method is described with particular order in the accompanying drawings, this do not require that or Hint must be performed according to the particular order these operation, or the operation having to carry out shown in whole could realize it is desired As a result.Additionally or alternatively, it is convenient to omit some steps, multiple steps are merged into a step to perform, and/or by one Step is decomposed into execution of multiple steps.

Although describing spirit and principles of the present invention by reference to some specific embodiments, it should be appreciated that, this Invention is not limited to disclosed specific embodiment, and the division to each side does not mean that the feature in these aspects can not yet Combination is this to divide merely to the convenience of statement to be benefited.It is contemplated that cover appended claims spirit and In the range of included various modifications and equivalent arrangements.

Claims

1. a kind of answer provides method, including：

Receive the problem that user submits to；

According to the corresponding inquiry chain of described problem, the answer of described problem is retrieved in the advance knowledge mapping for building, as base The answer of the described problem obtained in the retrieval mode retrieval of knowledge mapping；

2. method according to claim 1, wherein, according to the entity node and attribute section that are extracted from described problem Point, builds the corresponding inquiry chain of described problem, including：

Entity node, attribute node, interrogative node and core verb node from the dependency tree in extraction described problem；

According to the dependence between the dependency tree interior joint, by entity node to the attribute section between core verb node Point, as first kind attribute node；And,

Determine interrogative node to the attribute node between core verb node, and it is dynamic to core according to the interrogative node for determining Attribute node between word node, determines Equations of The Second Kind attribute node；

Entity node, first kind attribute node and Equations of The Second Kind attribute node are spliced successively, is constituted described problem correspondence Inquiry chain, wherein, each inquiry chain include an entity node and at least one attribute node, at least one attribute Node sorts in chain is inquired about according to dependence.

3. method according to claim 2, wherein, according to the corresponding inquiry chain of described problem, in the advance knowledge for building The answer of described problem is retrieved in collection of illustrative plates, including：

Entity node and adjacent attribute node are extracted as an entity attribute key-value pair from the corresponding inquiry chain of described problem, The adjacent attribute node is that described problem is corresponding inquires about attribute node adjacent with the entity node in chain；

For the entity attribute key-value pair that each rebuilds, the reality for rebuilding is retrieved in the advance knowledge mapping for building The corresponding result of body attribute key-value pair, and it is saved in result set；

If it is not, the corresponding result of entity attribute key-value pair that this is rebuild and not taken out in the corresponding inquiry chain of described problem The attribute node for taking constitutes the corresponding new inquiry chain of described problem, and it is real to perform the extraction from described problem corresponding inquiry chain The step of body node and adjacent attribute node are as an entity attribute key-value pair, wherein, the result is used as in new inquiry chain Entity node.

4. method according to claim 3, wherein, according to entity attribute node and attribute node after synonym extension Entity attribute key-value pair is rebuild, including：

Each entity node of entity node concentration is obtained after being extended for synonym, after the entity node and synonym extension Attribute node concentrate each attribute node constitute entity attribute key-value pair, the entity attribute key-value pair for being rebuild Collection.

5. method according to claim 3, wherein, according to the result set, determine the answer of described problem, including：

Determine the total number of word included in described problem；

For each Different Results in the result set, it is determined that retrieving the entity in the dependency tree used during the result The number of words of node and attribute node and, as the corresponding number of words of the result and；

6. method according to claim 2, wherein, according to the interrogative node for determining to the category between core verb node Property node, determines Equations of The Second Kind attribute node, including：

Semantic analysis is carried out to described problem, determine the interrogative node to the attribute node between core verb node whether Needs carry out taking antonym operation；

If so, the interrogative node to the attribute node between core verb node is carried out to take antonym operation, and will take Attribute node after antonym operation is defined as Equations of The Second Kind attribute node；

If it is not, the interrogative node is defined as into Equations of The Second Kind attribute node to the attribute node between core verb node.

7. method according to claim 1, described problem that the retrieval mode retrieval of knowledge based collection of illustrative plates is obtained is answered Case feeds back to the user：

The answer of described problem is not retrieved according to the retrieval mode based on local problem base, by the retrieval of knowledge based collection of illustrative plates Mode retrieves the answer feedback of the described problem for obtaining to the user.

8. method according to claim 7, also includes：

The way of search based on search engine, the retrieval mode based on Ask-Answer Community knowledge base is respectively adopted, based on deep learning Retrieval mode in, at least one retrieval mode retrieves the answer of described problem, wherein, different retrieval mode correspondences is different excellent First level；

If the retrieval mode of knowledge based collection of illustrative plates does not retrieve the answer of the described problem for obtaining, the answer of described problem will be returned to The answer feedback of the described problem that the retrieval mode retrieval of highest priority is obtained is to the user at least one retrieval mode.

9. method according to claim 8, wherein, the priority of at least one retrieval mode order from high to low For：

Retrieval mode based on Ask-Answer Community knowledge base, the retrieval mode based on deep learning, the retrieval side based on search engine Formula.

10., according to any described methods of claim 7-9, also include：

The answer of described problem is not retrieved according to the retrieval mode based on local problem base, the institute of the user is will be feedbacked to The answer for stating problem is saved in local problem base.

A kind of 11. answer offer devices, including：

Receiver module, the problem for receiving user's submission；

Module is built, the entity node and attribute node extracted from described problem for basis build described problem correspondence Inquiry chain；

First retrieval module, for according to the corresponding inquiry chain of described problem, retrieval to be described in the advance knowledge mapping for building The answer of problem, the answer of the described problem obtained as the retrieval mode retrieval of knowledge based collection of illustrative plates；

Feedback module, for the retrieval mode of knowledge based collection of illustrative plates to be retrieved the answer feedback of the described problem for obtaining to the use Family.