CN103577556B - Device and method for obtaining association degree of question and answer pair - Google Patents

Abstract

The invention discloses a device and a method for obtaining the association degree of a question and answer pair. The method comprises the following steps: carrying out word extraction on question content and answer content of a question and answer pair to be analyzed to obtain at least one question word to be analyzed and at least one answer word to be analyzed; selecting at least one question and answer knowledge record from a question and answer knowledge library including a plurality of question and answer knowledge records according to the question words to be analyzed and the answer words to be analyzed, and calculating the association degree of the question and answer pair to be analyzed according to the selected question and answer knowledge records. According to the device and the method, the quality of the question and answer pair can be semantically evaluated, and the evaluation effect is good; in addition, the method is easy to implement and good in universality.

Description

A kind of apparatus and method of the associated degree obtaining question and answer pair

Technical field

The present invention relates to network data communication field and in particular to a kind of device of associated degree of acquisition question and answer pair and Method.

Background technology

Ask-Answer Community is the network application that a kind of user produces content, and primitive form is to be carried according to the demand of oneself by user Go wrong, and to provide answer by other users.This form obtains information on network for user and provides new channel. Content can optionally be created yet with any user, the information quality difference that result in Ask-Answer Community is very big, with As for occurring in that substantial amounts of low quality question and answer pair in Ask-Answer Community.This not only searches information to user and brings inconvenience, with When also reduce the quality of Ask-Answer Community.Meanwhile, the method for prior art, the non-textual feature relying more heavily on question and answer pair is come Evaluate question and answer to quality, its versatility can be affected.

Content of the invention

In view of the above problems it is proposed that the present invention so as to provide one kind overcome the problems referred to above or at least in part solve on State a kind of device of associated degree of acquisition question and answer pair of problem and the method for the corresponding associated degree obtaining question and answer pair.

According to one aspect of the present invention, there is provided a kind of device of the associated degree of acquisition question and answer pair, this device bag Include:

Question and answer knowledge base, is suitable to store a plurality of question and answer knowledge record；

Word extraction unit, is suitable to the problem content to question and answer pair to be analyzed and answer content carries out word and extracts behaviour Make, obtain at least one problem word to be analyzed and at least one answer word to be analyzed；

Associated degree computing unit, is suitable to according to problem word to be analyzed and answer word to be analyzed, from question and answer knowledge Storehouse selects at least one question and answer knowledge record, calculates question and answer to be analyzed to being associated according to selected question and answer knowledge record Degree.

Alternatively, this device further includes question and answer construction of knowledge base unit, described question and answer construction of knowledge base unit, is suitable to Extract multiple question and answer pair from the webpage containing question and answer pair in advance, include a plurality of question and answer knowledge note according to the question and answer extracted to building The question and answer knowledge base of record；Described question and answer construction of knowledge base unit, be further adapted for extract from the webpage containing question and answer pair many During individual question and answer pair, crawl is with described question and answer to corresponding classification；Described question and answer construction of knowledge base unit, is further adapted in basis Extract question and answer to build question and answer knowledge base when, according to question and answer to and with described question and answer to corresponding classification build question and answer knowledge note Record；Each question and answer knowledge record corresponds to a classification, includes a problem word, an answer word respectively, and described Semantic relevancy between problem word and described answer word.

Alternatively, described associated degree computing unit, is suitable to choose problem word and the problem word to be analyzed that it includes Language coupling and the question and answer knowledge record of the answer word of inclusion and answer word match to be analyzed；Known according to the question and answer of described selection Corresponding to the question and answer knowledge record of identical category in memorize record, obtain this question and answer to be analyzed and be associated to for each classification Degree；Choose the maximum to the associated degree for each classification for this question and answer to be analyzed above-mentioned, using this maximum as The associated degree of question and answer pair to be analyzed.

Alternatively, described associated degree computing unit, is suitable in the question and answer knowledge record that will choose corresponding to mutually similar The semantic relevancy weighting summation of other question and answer knowledge record, obtains this question and answer to be analyzed to the phase being respectively directed to each classification Correlation degree.

Alternatively, described word extraction unit, is suitable to the problem content to question and answer pair to be analyzed and answer content is carried out Participle, removal stop words, word merge, and extract the operation of entity word.

Alternatively, described question and answer construction of knowledge base unit, is suitable to operate following to execution to each question and answer: to this question and answer pair Problem content and answer content carry out word extract operation, obtain problem set of words and answer set of words；Make problem word Each problem word in language set and each the answer word in answer set of words respectively with this question and answer to corresponding every One information record is formed on individual classification；Described question and answer construction of knowledge base unit, is suitable to each information record, execution is following Operation: calculate the probability that this answer word belongs to the category, calculate this answer word solution to this problem word in the category The single-minded degree released, calculates the intensity that this problem word is explained in the category with this answer word；By above-mentioned probability, specially One degree is multiplied with intensity, and obtained product is the semantic relevancy of this answer word and this problem word；Make this problem word Language, this answer word and its semantic relevancy form a question and answer knowledge record corresponding to the category.

Alternatively, described question and answer construction of knowledge base unit, is suitable to calculate this answer word as follows and belongs to this The probability of classification:

$p\left(\mathrm{ck}\right|\mathrm{awj})=p\left(\mathrm{awj}\right|\mathrm{ck})*\frac{p\left(\mathrm{ck}\right)}{p\left(\mathrm{awj}\right)};$

Described question and answer construction of knowledge base unit, is suitable to calculate each answer word pair in the category as follows The single-minded degree of the explanation of this problem word:

$\mathrm{apecific}(\mathrm{qwi},\mathrm{awi}|c=\mathrm{ck})=p\left(\mathrm{qwi}\right|\mathrm{awj},c=\mathrm{ck})=\frac{\#(\mathrm{qwi},\mathrm{awj})}{\#\left(\mathrm{awj}\right)}|c=\mathrm{ck};$

Described question and answer construction of knowledge base unit, is suitable to calculate as follows in the category this problem word with each The intensity that individual answer word explains:

$\mathrm{interpret}(\mathrm{qwi},\mathrm{awj}|c=\mathrm{ck})=p\left(\mathrm{awj}\right|\mathrm{qwi},c=\mathrm{ck})=\frac{\#(\mathrm{qwi},\mathrm{awj})}{{\mathrm{\σ}}_{j=1}^x\#(\mathrm{qwi},\mathrm{awj})}|c=\mathrm{ck};$

Described question and answer construction of knowledge base unit, is suitable to as follows by above-mentioned probability, single-minded degree and intensity phase Take advantage of:

Weight(qwi, awj | c=ck)=p(ck | awj) * specific(qwi, awj | c=ck) * interpret (qwi, awj | c=ck)；

Wherein, p(ck) represent the probability that classification ck occurs；P(awj) represent the probability that answer is awj；P(awj │ ck) table Show that ck classification belongs to the probability of awj；

#(qwi, awj) problem of representation word is qwi and answer word is the number of times of awj；

#(awj) represent the number of times that answer word is awj.

According to a further aspect in the invention, there is provided a kind of method of the associated degree obtaining question and answer pair, the method bag Include following steps:

Problem content to question and answer pair to be analyzed and answer content carry out word and extract operation, obtain at least one and treat point Analysis problem word and at least one answer word to be analyzed；

According to problem word to be analyzed and answer word to be analyzed, from the question and answer knowledge base including a plurality of question and answer knowledge record Select at least one question and answer knowledge record, calculate the associated journey of question and answer pair to be analyzed according to selected question and answer knowledge record Degree.

Alternatively, the method further includes: extracts multiple question and answer pair from the webpage containing question and answer pair in advance, according to carrying The question and answer taking include the question and answer knowledge base of a plurality of question and answer knowledge record to building；Multiple extracting from the webpage containing question and answer pair During question and answer pair, crawl is with described question and answer to corresponding classification；When according to the question and answer extracted to building question and answer knowledge base, according to asking Answer questions and with described question and answer, question and answer knowledge record is built to corresponding classification；Each question and answer knowledge record corresponds to a classification, Include the semantic phase between a problem word, an answer word, and described problem word and described answer word respectively Guan Du.

Alternatively, described according to problem word to be analyzed with answer word to be analyzed, select at least one from question and answer knowledge base Bar question and answer knowledge record, calculates the associated degree of question and answer pair to be analyzed according to selected question and answer knowledge record, concrete bag Include: choose the problem word that it includes with problem word match to be analyzed and the answer word that includes and answer word to be analyzed The question and answer knowledge record joined；According to the question and answer knowledge record corresponding to identical category in the question and answer knowledge record of described selection, obtain To this question and answer to be analyzed to the associated degree for each classification；Choose this question and answer to be analyzed above-mentioned to for each class The maximum of other associated degree, using this maximum as the associated degree of question and answer pair to be analyzed.

Alternatively, according to the question and answer knowledge record corresponding to identical category in the question and answer knowledge record of described selection, obtain This question and answer to be analyzed, to the associated degree being respectively directed to each classification, specifically includes: in the question and answer knowledge record that will choose Corresponding to the semantic relevancy weighting summation of the question and answer knowledge record of identical category, obtain this question and answer to be analyzed to being respectively directed to The associated degree of each classification.

Alternatively, the described problem content to described question and answer pair to be analyzed and answer content carry out word and extract operation, Specifically include: the problem content to question and answer pair to be analyzed and answer content carry out participle, remove stop words, word merging, and carry The operation for the treatment of excess syndrome pronouns, general term for nouns, numerals and measure words.

Alternatively, described according to question and answer to and with described question and answer to corresponding classification build question and answer knowledge base, specifically include: To each question and answer pair, the problem content to this question and answer pair and answer content carry out word and extract operation, obtain problem set of words With answer set of words；Each the problem word in problem set of words is made to divide with each the answer word in answer set of words On with this question and answer to each classification corresponding, do not form an information record；To each information record, execution is following to be operated: Calculate the probability that this answer word belongs to the category, calculate in the category this answer word special to the explanation of this problem word One degree, calculates the intensity that this problem word is explained in the category with this answer word；By above-mentioned probability, single-minded degree It is multiplied with intensity, obtained product is the semantic relevancy of this answer word and this problem word；Make this problem word, this answers Case word and its semantic relevancy form a question and answer knowledge record corresponding to the category.

Alternatively, described this answer word of calculating belongs to the probability of the category, specifically includes:

$p\left(\mathrm{ck}\right|\mathrm{awj})=p\left(\mathrm{awj}\right|\mathrm{ck})*\frac{p\left(\mathrm{ck}\right)}{p\left(\mathrm{awj}\right)};$

The described single-minded degree calculating the explanation to this problem word for each answer word in the category, specifically includes:

$\mathrm{apecific}(\mathrm{qwi},\mathrm{awi}|c=\mathrm{ck})=p\left(\mathrm{qwi}\right|\mathrm{awj},c=\mathrm{ck})=\frac{\#(\mathrm{qwi},\mathrm{awj})}{\#\left(\mathrm{awj}\right)}|c=\mathrm{ck};$

The described calculating intensity that this problem word is explained with each answer word in the category, specifically includes:

$\mathrm{interpret}(\mathrm{qwi},\mathrm{awj}|c=\mathrm{ck})=p\left(\mathrm{awj}\right|\mathrm{qwi},c=\mathrm{ck})=\frac{\#(\mathrm{qwi},\mathrm{awj})}{{\mathrm{\σ}}_{j=1}^x\#(\mathrm{qwi},\mathrm{awj})}|c=\mathrm{ck};$

Above-mentioned probability, single-minded degree are multiplied with intensity, specifically include:

Weight(qwi, awj | c=ck)=p(ck | awj) * specific(qwi, awj | c=ck) * interpret (qwi, awj | c=ck)；

Wherein, p(ck) represent the probability that classification ck occurs；P(awj) represent the probability that answer is awj；P(awj │ ck) table Show that ck classification belongs to the probability of awj；

#(qwi, awj) problem of representation word is qwi and answer word is the number of times of awj；

#(awj) represent the number of times that answer word is awj.

Technology according to the present invention scheme, from the webpage containing question and answer pair extract multiple question and answer to and asked according to extraction Answer questions and build question and answer knowledge base, the problem content to question and answer pair to be analyzed and the answer content including a plurality of question and answer knowledge record Carry out word to extract operation and obtain at least one problem word to be analyzed and at least one answer word to be analyzed, and then according to Problem word to be analyzed and answer word to be analyzed select at least one question and answer knowledge record and according to selected from question and answer knowledge base The question and answer knowledge record selected calculates the associated degree of question and answer pair to be analyzed, can evaluate the matter of question and answer pair in terms of semantic Amount, solves the problems, such as that the evaluation effect that prior art is evaluated the quality of question and answer pair only in morphology aspect and led to is not good, and And easy realize, highly versatile.

Brief description

By reading the detailed description of hereafter preferred implementation, various other advantages and benefit are common for this area Technical staff will be clear from understanding.Accompanying drawing is only used for illustrating the purpose of preferred implementation, and is not considered as to the present invention Restriction.And in whole accompanying drawing, it is denoted by the same reference numerals identical part.In the accompanying drawings:

The flow chart that Fig. 1 shows the method for associated degree obtaining question and answer pair according to an embodiment of the invention；

Fig. 2 shows the detailed flow chart building question and answer knowledge base；

Fig. 3 shows an interpretation model schematic diagram using question and answer knowledge base obtained from step as shown in Figure 2；

Fig. 4 shows the detailed flow chart of step s200 in Fig. 1；And

Fig. 5 shows the block diagram of the device of associated degree obtaining question and answer pair according to an embodiment of the invention；

Fig. 6 shows the block diagram of the device of associated degree obtaining question and answer pair in accordance with another embodiment of the present invention.

Specific embodiment

The method of the existing associated degree obtaining question and answer pair, is to be described and asked using text feature and non-textual feature The problem answered questions and answer.Text feature mainly includes textual visual feature, and (such as punctuation mark density, average word is long, text Entropy etc.) and content of text feature (such as content of text word ratio, interrogative density, related term covering etc.), and extract Chinese certainly The wrong widely used feature (such as individual character density feature etc.) of moment；Non-textual feature comprises the technorati authority index of user, answer Problem state, answer response time, customer relationship interaction feature etc..After respectively feature is extracted to problem and answer, in instruction Practice and on collection, learn a problem quality forecast model and answer quality prediction model respectively, and the output knot using two models Fruit is evaluating question and answer to quality.However, being entered for answer quality using the method for the existing associated degree obtaining question and answer pair When row is evaluated, simply use related term Cover Characteristics to describe the semantic matching degree between problem and answer, this is only not only Rest in morphology aspect, and do not account for the semantic matching degree between problem and answer.But the language between problem and answer The adopted matching degree exactly core to quality for the question and answer, such as problem are that " where the capital of China is？", answer 1 is " Beijing ", answers Case 2 is " capital of China is Shanghai ".So problem through participle and abandons after stop words processes, for " the Chinese capital is where ", Answer 1 word segmentation result is " Beijing ", and answer 2 word segmentation result is " the Chinese capital Shanghai ".In prior art, semantic matching degree is permissible It is defined as: the word number jointly occurring in problem and answer is divided by the number of all words in problem and answer.Then problem and The semantic matching degree of answer 1 is: 0/4=0.The semantic matching degree of problem and answer 2 is: 2/4=0.5.Using prior art, will Think that answer 2 and problem are more mated.And it is understood that this clearly improperly.

It is more fully described the exemplary embodiment of the disclosure below with reference to accompanying drawings.Although showing the disclosure in accompanying drawing Exemplary embodiment it being understood, however, that may be realized in various forms the disclosure and should not be by embodiments set forth here Limited.On the contrary, these embodiments are provided to be able to be best understood from the disclosure, and can be by the scope of the present disclosure Complete conveys to those skilled in the art.

The flow chart that Fig. 1 shows the method for associated degree obtaining question and answer pair according to an embodiment of the invention. According to a further aspect in the invention, there is provided a kind of method of the associated degree obtaining question and answer pair, the method includes walking as follows Rapid s100 and step s200:

S100, the problem content to question and answer pair to be analyzed and answer content carry out word and extract operation, obtain at least one Individual problem word to be analyzed and at least one answer word to be analyzed.

In one embodiment of the invention, the problem content to question and answer pair to be analyzed and answer content carry out word and carry Extract operation specifically includes: the problem content to question and answer pair to be analyzed and answer content carry out participle, remove stop words, word merging (word join), and extract the operation of entity word (such as noun, verb etc.).Then by the problem content of question and answer pair to be analyzed Obtain at least one problem word to be analyzed, at least one answer word to be analyzed is obtained by the answer content of question and answer pair to be analyzed Language.

S200, according to problem word to be analyzed and answer word to be analyzed, from the question and answer including a plurality of question and answer knowledge record Knowledge base selects at least one question and answer knowledge record, calculates the phase of question and answer pair to be analyzed according to selected question and answer knowledge record Correlation degree.

Question and answer pair to be analyzed can be asked in terms of semantic by step s200 of the present embodiment by using question and answer knowledge base Topic content and answer content are analyzed to obtain the associated degree of question and answer pair to be analyzed, and evaluation effect is more preferably and easily real Existing.

Further, the described question and answer knowledge base including a plurality of question and answer knowledge record, is by advance from containing question and answer pair Webpage extract multiple question and answer pair, according to extract question and answer to build obtained from.In one embodiment of the invention, exist When the webpage containing question and answer pair extracts multiple question and answer pair, crawl is with described question and answer to corresponding classification.Then according to extraction Question and answer to build question and answer knowledge base when, according to question and answer to and with described question and answer to corresponding classification build question and answer knowledge record. Each question and answer knowledge record among the question and answer knowledge base obtaining corresponds to a classification, includes a problem word respectively (qw), the semantic relevancy between an answer word (aw), and described problem word and described answer word.

Include a plurality of question and answer knowledge record by using magnanimity, the high-quality question and answer extracted by webpage to structure to ask Answer knowledge base, can be obtained based on the study to magnanimity information the problem word of a plurality of question and answer knowledge record and answer word it Between semantic relevancy；And extract, by using from webpage, the information architecture question and answer knowledge base obtaining, applicable scope is wider, The versatility of method is higher.

Fig. 2 shows the detailed flow chart building question and answer knowledge base.Specifically include following steps s310, step s320 and Step s330:

S310, extract multiple question and answer pair from the webpage containing question and answer pair in advance, crawl is with described question and answer to corresponding class Not.

In the present embodiment, can be by using web crawlers, the webpage capture containing high-quality question and answer pair from the Internet Data simultaneously extracts question and answer pair, to ensure the quality of extracted question and answer pair；The described webpage containing high-quality question and answer pair includes Cqa community, each big specialty forum etc., then can use floor technology of identification, be asked a question according to building-owner, be for 1st floor 2nd floors etc. answer Mode, to extract question and answer pair.Include the classification corresponding to each question and answer pair due to the described webpage containing high-quality question and answer pair Information it is possible to capture question and answer to while capture in the lump with described question and answer to corresponding classification.

S320, to each question and answer pair, the problem content to this question and answer pair and answer content carry out word and extract operation, obtain Problem set of words and answer set of words；Make every in each problem word in problem set of words and answer set of words Individual answer word forms an information record respectively on this question and answer to each classification corresponding.

In one embodiment of the invention, to extract in step s310 the described question and answer obtaining to each of question and answer To problem content and answer content carry out word and extract operation, specifically include, the problem content to question and answer pair and answer content Carry out participle, remove stop words, word merging, and the operation extracting entity word.

Then at least one problem word is obtained by the problem content of each question and answer pair, by the answer of each question and answer pair Hold and obtain at least one answer word, then can obtain the category set < c for this question and answer pair₁..., c_k..., c_p>, problem word Language set < qw₁..., qw_i..., qw_m>and answer set of words<aw₁..., aw_j..., aw_n>.

By making each problem word (qw in problem set of words_i) with answer set of words in each answer word (aw_j) respectively with this question and answer to corresponding each classification (c_k) one information record of upper formation, such as < qw_i, aw_j, c_k>, then may be used To form m*n*p bar information record.

S330, to each information record, execution is following to be operated: calculates the probability that this answer word belongs to the category, meter Calculate the single-minded degree of the explanation to this problem word for this answer word in the category, calculate this problem word in the category and use The intensity that this answer word explains；Above-mentioned probability, single-minded degree are multiplied with intensity, obtained product is this answer word Language and the semantic relevancy of this problem word；Make this problem word, this answer word corresponding with its semantic relevancy formation one Question and answer knowledge record < qw in the category_i, aw_j, weight(qw_i, aw_j)>or<qw_i, aw_j, c_k, weight(qw_i, aw_j) >.This Step s330 in embodiment, can be to the word having carried out as described in step s320 in the question and answer to the magnanimity from webpage capture After the information record that language extracts operation and obtains magnanimity, the information record based on described magnanimity is carried out, then the letter based on magnanimity The semantic relevancy that breath records and obtains is more accurate.

It is preferred that described calculate the probability that this answer word belongs to the category, specifically include:

$p\left(\mathrm{ck}\right|\mathrm{awj})=p\left(\mathrm{awj}\right|\mathrm{ck})*\frac{p\left(\mathrm{ck}\right)}{p\left(\mathrm{awj}\right)};$

The described single-minded degree calculating the explanation to this problem word for each answer word in the category, specifically includes:

$\mathrm{apecific}(\mathrm{qwi},\mathrm{awi}|c=\mathrm{ck})=p\left(\mathrm{qwi}\right|\mathrm{awj},c=\mathrm{ck})=\frac{\#(\mathrm{qwi},\mathrm{awj})}{\#\left(\mathrm{awj}\right)}|c=\mathrm{ck};$

The described calculating intensity that this problem word is explained with each answer word in the category, specifically includes:

$\mathrm{interpret}(\mathrm{qwi},\mathrm{awj}|c=\mathrm{ck})=p\left(\mathrm{awj}\right|\mathrm{qwi},c=\mathrm{ck})=\frac{\#(\mathrm{qwi},\mathrm{awj})}{{\mathrm{\&sigma;}}_{j=1}^x\#(\mathrm{qwi},\mathrm{awj})}|c=\mathrm{ck};$

Above-mentioned probability, single-minded degree are multiplied with intensity, specifically include:

Weight(qwi, awj | c=ck)=p(ck | awj) * specific(qwi, awj | c=ck) * interpret (qwi, awj | c=ck)；

Wherein, p(ck) represent the probability that classification ck occurs；P(awj) represent the probability that answer is awj；P(awj │ ck) table Show that ck classification belongs to the probability of awj；

#(qwi, awj) problem of representation word is qwi and answer word is the number of times of awj；

#(awj) represent the number of times that answer word is awj.

By step s310, step s320 and step s330, question and answer knowledge record can be obtained and build question and answer knowledge base.Figure 3 show an interpretation model schematic diagram using question and answer knowledge base obtained from step as shown in Figure 2.Understand, for every One problem word qw_i, category set < c can be directed to₁..., c_k..., c_p> in each classification, obtain n bar question and answer knowledge note Record.Certainly, if those skilled in the art are it will be appreciated that calculated semantic relevancy is 0, can delete corresponding Question and answer knowledge record；Furthermore, if in question and answer knowledge base the quantity of question and answer knowledge record excessive and make storage question and answer knowledge note The expense of record and the associated degree calculating question and answer pair to be analyzed is excessive, can preset a threshold value, semantic relevancy is less than The question and answer knowledge record of threshold value is deleted to reduce expense.

Fig. 4 shows the detailed flow chart of step s200 in Fig. 1.It is to be analyzed at least one is obtained by step s100 After problem word and at least one answer word to be analyzed, step s200 specifically includes following steps s210, step s220 and step Rapid s230:

S210, choose the problem word that it includes with problem word match to be analyzed and the answer word that includes with to be analyzed The question and answer knowledge record of answer word match.In the present embodiment, problem word and problem word match to be analyzed refer to be analyzed Problem word is identical with problem word or problem word to be analyzed be problem word substring；Answer word and answer word to be analyzed Language coupling refer to that answer word to be analyzed is identical with answer word or answer word to be analyzed be answer word substring, this enforcement Example pass through step s210, using the method for fields match or field searches, select from question and answer knowledge base part with to be analyzed Question and answer are to related question and answer knowledge record.

S220, according in the question and answer knowledge record of described selection correspond to identical category question and answer knowledge record, be somebody's turn to do Question and answer to be analyzed, to the associated degree being respectively directed to each classification, specifically include: right in the question and answer knowledge record that will choose This question and answer to be analyzed should be obtained in the semantic relevancy weighting summation of the question and answer knowledge record of identical category each to being respectively directed to The associated degree of individual classification.

The present embodiment, the question and answer knowledge record selected by step s210 is grouped according to its corresponding classification, Question and answer knowledge record corresponding to identical category is one group；The semantic relevancy of each group of question and answer knowledge record is weighted (example If weights are 1 or 100) it is added, obtain this question and answer to be analyzed to the associated degree for the category；Thus obtain at least One (number of the associated degree in the present embodiment is the number to corresponding classification for the question and answer to be analyzed) is associated degree.

S230, the selection maximum to the associated degree for each classification for this question and answer to be analyzed above-mentioned, with this Big value is as the associated degree of question and answer pair to be analyzed.

Fig. 5 shows the block diagram of the device of associated degree obtaining question and answer pair according to an embodiment of the invention.Should Device includes question and answer knowledge base 100, word extraction unit 200 and associated degree computing unit 300.

Question and answer knowledge base 100, is suitable to store a plurality of question and answer knowledge record；The question and answer knowledge base 100 of the present embodiment can be led to The magnanimity question and answer crossed in crawl webpage obtain to structure.

Word extraction unit 200, is suitable to the problem content to question and answer pair to be analyzed and answer content carries out word extraction Operation, obtains at least one problem word to be analyzed and at least one answer word to be analyzed.

In one embodiment of the invention, word extraction unit 200, is suitable to the problem content to question and answer pair to be analyzed Carry out participle, remove stop words, word merging (word join) with answer content, and extract entity word (such as noun, verb etc.) Operation, to obtain at least one problem word to be analyzed and at least one answer word to be analyzed.

Associated degree computing unit 300, is suitable to, according to problem word to be analyzed and answer word to be analyzed, know from question and answer Know storehouse and select at least one question and answer knowledge record, calculate the correlation of question and answer pair to be analyzed according to selected question and answer knowledge record Connection degree.

In one embodiment of the invention, associated degree computing unit 300, is suitable to choose the problem word that it includes With problem word match to be analyzed and the question and answer knowledge record of answer word and answer word match to be analyzed that includes.This enforcement In example, problem word and problem word match to be analyzed refer to that problem word to be analyzed is identical with problem word or problem to be analyzed Word is the substring of problem word；Answer word and answer word match to be analyzed refer to answer word to be analyzed and answer word Answer word identical or to be analyzed is the substring of answer word；Mutually similar according to corresponding in the question and answer knowledge record of described selection Other question and answer knowledge record, obtains this question and answer to be analyzed to the associated degree for each classification, more specifically, being to select In the question and answer knowledge record taking correspond to identical category question and answer knowledge record semantic relevancy weighting (for example, weights be 1 or 100) it is added and obtains this question and answer to be analyzed to the associated degree being respectively directed to each classification, thus obtain at least one (number of the associated degree in the present embodiment is the number to corresponding classification for the question and answer to be analyzed) is associated degree；In selection State the maximum to the associated degree for each classification for this question and answer to be analyzed, using this maximum as question and answer to be analyzed To associated degree.

Using question and answer knowledge base 100, word extraction unit 200 and associated degree computing unit 300, by using treating point Analysis problem word and answer word to be analyzed, select at least one question and answer knowledge record from question and answer knowledge base, and according to selected The question and answer knowledge record selected calculates the associated degree of question and answer pair to be analyzed, can in terms of semantic to question and answer to be analyzed to entering Row analysis, evaluation effect is realized more preferably and easily, extracts, by using from webpage, the information architecture question and answer knowledge base obtaining, and fits Scope is wider, and versatility is higher.

Fig. 6 shows the block diagram of the device of associated degree obtaining question and answer pair in accordance with another embodiment of the present invention. In the present embodiment, this device also includes question and answer construction of knowledge base unit 400, and question and answer construction of knowledge base unit 400 is suitable in advance Extract multiple question and answer pair from the webpage containing question and answer pair, include a plurality of question and answer knowledge record according to the question and answer extracted to building Question and answer knowledge base.In the arrangement as shown in fig. 5, question and answer knowledge base is existing, and the quantity of information due to real network is continuously increased, The pace of change of information content is fast, and the content of question and answer knowledge base generally requires to update, and the present embodiment is by setting up question and answer knowledge base Construction unit 400 builds (in other words update) question and answer knowledge base it is ensured that the instantaneity of the content of question and answer knowledge base and reliability Property.

It is preferred that when extracting multiple question and answer pair from the webpage containing question and answer pair, question and answer construction of knowledge base unit 400 is grabbed Take with described question and answer to corresponding classification.In the present embodiment, by using web crawlers, high-quality can be contained from the Internet The webpage capture data of question and answer pair simultaneously extracts question and answer pair, to ensure the quality of extracted question and answer pair；Described containing high-quality The webpage of question and answer pair includes cqa community, each big specialty forum etc..Due to the described webpage containing high-quality question and answer pair include right Should in the classification information of each question and answer pair, so question and answer construction of knowledge base unit 400 can capture question and answer to while in the lump Crawl is with described question and answer to corresponding classification.

In the present embodiment, question and answer construction of knowledge base unit 400, is suitable to operate following to execution to each question and answer: to this The problem content of question and answer pair and answer content carry out word and extract operation, obtain problem set of words and answer set of words, tool Body ground, question and answer construction of knowledge base unit 400 to extract the described question and answer that obtain to each of question and answer pair problem content and Answer content carries out participle, removes stop words, word merging, and extracts the operation of entity word and obtain problem word and answer word Language；Make each answer word in each problem word in problem set of words and answer set of words respectively with this question and answer To one information record of formation in each classification corresponding.Question and answer construction of knowledge base unit 400, is suitable to each information is remembered Record, execution is following to be operated: calculates the probability that this answer word belongs to the category, calculates this answer word in the category and this is asked The single-minded degree of the explanation of epigraph language, calculates the intensity that this problem word is explained in the category with this answer word；Will Above-mentioned probability, single-minded degree are multiplied with intensity, and obtained product is the semantic relevancy of this answer word and this problem word； This problem word, this answer word and its semantic relevancy is made to form a question and answer knowledge record corresponding to the category.

More specifically, question and answer construction of knowledge base unit 400, it is suitable to calculate this answer word as follows and belongs to this The probability of classification:

$p\left(\mathrm{ck}\right|\mathrm{awj})=p\left(\mathrm{awj}\right|\mathrm{ck})*\frac{p\left(\mathrm{ck}\right)}{p\left(\mathrm{awj}\right)};$

More specifically, question and answer construction of knowledge base unit 400, it is suitable to calculate as follows that each is answered in the category The single-minded degree of the explanation to this problem word for the case word:

$\mathrm{apecific}(\mathrm{qwi},\mathrm{awi}|c=\mathrm{ck})=p\left(\mathrm{qwi}\right|\mathrm{awj},c=\mathrm{ck})=\frac{\#(\mathrm{qwi},\mathrm{awj})}{\#\left(\mathrm{awj}\right)}|c=\mathrm{ck};$

More specifically, question and answer construction of knowledge base unit 400, it is suitable to calculate this problem in the category as follows The intensity that word is explained with each answer word:

$\mathrm{interpret}(\mathrm{qwi},\mathrm{awj}|c=\mathrm{ck})=p\left(\mathrm{awj}\right|\mathrm{qwi},c=\mathrm{ck})=\frac{\#(\mathrm{qwi},\mathrm{awj})}{{\mathrm{\&sigma;}}_{j=1}^x\#(\mathrm{qwi},\mathrm{awj})}|c=\mathrm{ck};$

More specifically, question and answer construction of knowledge base unit 400, it is suitable to as follows by above-mentioned probability, single-minded degree It is multiplied with intensity:

Weight(qwi, awj | c=ck)=p(ck | awj) * specific(qwi, awj | c=ck) * interpret (qwi, awj | c=ck)；

Wherein, p(ck) represent the probability that classification ck occurs；P(awj) represent the probability that answer is awj；P(awj │ ck) table Show that ck classification belongs to the probability of awj；

#(qwi, awj) problem of representation word is qwi and answer word is the number of times of awj；

#(awj) represent the number of times that answer word is awj.

Be can achieve the effect that such as there are following question and answer using embodiments of the invention below by way of an example explanation Right, classification is " medical treatment ＆ health ":

Processed by participle technique, obtain problem word to be analyzed and answer word to be analyzed is as follows:

From word segmentation result as can be seen that not having related term to cover problem and answer, if therefore using prior art, Easily think this question and answer to associated low degree, of low quality.But actually using artificial judgment it will be apparent that this question and answer pair It is high-quality question and answer pair.

If processing above-mentioned question and answer pair using methods and apparatus of the present invention, it is possible, firstly, to transfer existing question and answer knowledge base, Or by capturing the question and answer pair of cqa community, each big specialty forum, build question and answer knowledge base；

Second step, to above-mentioned question and answer pair to be analyzed, extracts operation through word and obtains problem set of words < child to be analyzed Son, cough, nasal mucus>, answer set of words to be analyzed<symptom, medicine, treatment, antiviral, xiao'er ganmao granules, illustrate, agent Amount, cough-relieving, Chinese medicine, electuary, antibiotic, amoxicillin, amoxicillin granules, granule, be administered orally, Roxithromycin, curative effect, and The classification obtaining question and answer pair to be analyzed is " medical treatment ＆ health "；

3rd step, according to each problem word to be analyzed and the category, selects to obtain problem word from question and answer knowledge base Some question and answer knowledge records of language and problem word match to be analyzed, thus obtain following answer word and semantic relevancy (is Easy-to-read, the numerical value of the semantic relevancy in following table has been by the numerical value after suitable normalized):

4th step, according to the answer word to be analyzed in answer set of words to be analyzed, obtain selected by the 3rd step The question and answer knowledge record of the answer word that it includes and answer word match to be analyzed is filtered out on the basis of question and answer knowledge record, And then obtain the semantic relevancy of filtered out question and answer knowledge record.Through analysis understand, in this example with question and answer knowledge record in The answer word to be analyzed of answer word match include: < oral, cough with asthma, xiao'er ganmao granules, check, cough-relieving, treatment, stream Sense symptom, cold granules >.

The associated degree calculating above-mentioned question and answer pair to be analyzed again can draw, this question and answer to be analyzed is to being associated Degree has reached 0.9(under conditions of associated degree span is 0～1).

It should be understood that

Algorithm and display be not inherently related to any certain computer, virtual system or miscellaneous equipment provided herein. Various general-purpose systems can also be used together with based on teaching in this.As described above, construct required by this kind of system Structure be obvious.Additionally, the present invention is also not for any certain programmed language.It is understood that, it is possible to use various Programming language realizes the content of invention described herein, and the description above language-specific done is to disclose this Bright preferred forms.

In description mentioned herein, illustrate a large amount of details.It is to be appreciated, however, that the enforcement of the present invention Example can be put into practice in the case of not having these details.In some instances, known method, structure are not been shown in detail And technology, so as not to obscure the understanding of this description.

Similarly it will be appreciated that in order to simplify the disclosure and help understand one or more of each inventive aspect, Above in the description to the exemplary embodiment of the present invention, each feature of the present invention is grouped together into single enforcement sometimes In example, figure or descriptions thereof.However, the method for the disclosure should be construed to reflect an intention that i.e. required guarantor The application claims of shield more features than the feature being expressly recited in each claim.More precisely, it is such as following Claims reflected as, inventive aspect is all features less than single embodiment disclosed above.Therefore, The claims following specific embodiment are thus expressly incorporated in this specific embodiment, wherein each claim itself All as the separate embodiments of the present invention.

Those skilled in the art are appreciated that and the module in the equipment in embodiment can be carried out adaptively Change and they are arranged in one or more equipment different from this embodiment.Can be the module in embodiment or list Unit or assembly be combined into a module or unit or assembly, and can be divided in addition multiple submodule or subelement or Sub-component.In addition to such feature and/or at least some of process or unit exclude each other, can adopt any Combination is to all features disclosed in this specification (including adjoint claim, summary and accompanying drawing) and so disclosed Where method or all processes of equipment or unit are combined.Unless expressly stated otherwise, this specification (includes adjoint power Profit requires, summary and accompanying drawing) disclosed in each feature can carry out generation by the alternative features providing identical, equivalent or similar purpose Replace.

Although additionally, it will be appreciated by those of skill in the art that some embodiments described herein include other embodiments In included some features rather than further feature, but the combination of the feature of different embodiment means to be in the present invention's Within the scope of and form different embodiments.For example, in the following claims, embodiment required for protection appoint One of meaning can in any combination mode using.

The all parts embodiment of the present invention can be realized with hardware, or to run on one or more processor Software module realize, or with combinations thereof realize.It will be understood by those of skill in the art that can use in practice Microprocessor or digital signal processor (dsp) are realizing the associated degree of acquisition question and answer pair according to embodiments of the present invention Device in some or all parts some or all functions.The present invention is also implemented as execution institute here (for example, computer program and computer program produce for some or all equipment of method of description or program of device Product).Such program realizing the present invention can store on a computer-readable medium, or can have one or more The form of signal.Such signal can be downloaded from internet website and obtain, or provides on carrier signal, or to appoint What other forms provides.

It should be noted that above-described embodiment the present invention will be described rather than limits the invention, and ability Field technique personnel can design alternative embodiment without departing from the scope of the appended claims.In the claims, Any reference markss between bracket should not be configured to limitations on claims.Word "comprising" does not exclude the presence of not Element listed in the claims or step.Word "a" or "an" before element does not exclude the presence of multiple such Element.The present invention can come real by means of the hardware including some different elements and by means of properly programmed computer Existing.If in the unit claim listing equipment for drying, several in these devices can be by same hardware branch To embody.The use of word first, second, and third does not indicate that any order.These words can be explained and run after fame Claim.

Claims (12)

1. a kind of device of the associated degree obtaining question and answer pair, this device includes:

Question and answer knowledge base, is suitable to store a plurality of question and answer knowledge record；

Word extraction unit, is suitable to the problem content to question and answer pair to be analyzed and answer content carries out word and extracts operation, obtains To at least one problem word to be analyzed and at least one answer word to be analyzed；

Associated degree computing unit, is suitable to according to problem word to be analyzed and answer word to be analyzed, from the choosing of question and answer knowledge base Select at least one question and answer knowledge record, calculate the associated journey of question and answer pair to be analyzed according to selected question and answer knowledge record Degree；

Described associated degree computing unit, be particularly adapted to choose the problem word that it includes with problem word match to be analyzed and Including answer word and answer word match to be analyzed question and answer knowledge record；According in the question and answer knowledge record of described selection Corresponding to the question and answer knowledge record of identical category, obtain this question and answer to be analyzed to the associated degree for each classification；Choosing Take the maximum to the associated degree for each classification for this question and answer to be analyzed, using this maximum as question and answer to be analyzed To associated degree.

2. device according to claim 1, wherein, this device further includes question and answer construction of knowledge base unit,

Described question and answer construction of knowledge base unit, is suitable to extract multiple question and answer pair from the webpage containing question and answer pair in advance, according to carrying The question and answer taking include the question and answer knowledge base of a plurality of question and answer knowledge record to building；

Described question and answer construction of knowledge base unit, is further adapted for when extracting multiple question and answer pair from the webpage containing question and answer pair, Crawl is with described question and answer to corresponding classification；

Described question and answer construction of knowledge base unit, be further adapted for according to extract question and answer to build question and answer knowledge base when, according to Question and answer to and with described question and answer to corresponding classification build question and answer knowledge record；Each question and answer knowledge record corresponds to a class Not, include the semanteme between a problem word, an answer word, and described problem word and described answer word respectively Degree of association.

3. device according to claim 2, wherein,

Described question and answer construction of knowledge base unit, is suitable to the following operation to execution of each question and answer:

Problem content to this question and answer pair and answer content carry out word and extract operation, obtain problem set of words and answer word Set；Each the problem word in problem set of words is made to ask with this respectively with each the answer word in answer set of words Answer questions one information record of formation in each classification corresponding；

Described question and answer construction of knowledge base unit, is suitable to each information record, and execution is following to be operated:

Calculate the probability that this answer word belongs to the category, calculate this answer word explanation to this problem word in the category Single-minded degree, calculate this problem word is explained in the category intensity with this answer word；By above-mentioned probability, single-minded Degree is multiplied with intensity, and obtained product is the semantic relevancy of this answer word and this problem word；Make this problem word, This answer word and its semantic relevancy form a question and answer knowledge record corresponding to the category.

4. the device according to claims 1 to 3 any claim, wherein,

Described associated degree computing unit, is suitable to the question and answer knowledge corresponding to identical category in the question and answer knowledge record that will choose The semantic relevancy weighting summation of record, obtains this question and answer to be analyzed to the associated degree being respectively directed to each classification.

5. the device according to claims 1 to 3 any claim, wherein,

Described word extraction unit, is suitable to the problem content to question and answer pair to be analyzed and answer content carries out participle, removes and stop Word, word merge, and extract the operation of entity word.

6. device according to claim 3, wherein,

Described question and answer construction of knowledge base unit, is suitable to calculate the probability that this answer word belongs to the category as follows:

$p\left(ck\right|awj)=p\left(awj\right|ck)*\frac{p\left(ck\right)}{p\left(awj\right)};$

Described question and answer construction of knowledge base unit, is suitable to calculate each answer word in the category as follows and this is asked The single-minded degree of the explanation of epigraph language:

$specific(qwi,awj|c=ck)=p\left(qwi\right|awj,c=ck)=\frac{\#(qwi,awj)}{\#\left(awj\right)}|c=ck;$

Described question and answer construction of knowledge base unit, is suitable to calculate this problem word in the category as follows and is answered with each The intensity that case word explains:

$\mathrm{int}erpret\left(qwi,awj|c=ck\right)=p\left(awj|qwi,c=ck\right)=\frac{\#\left(qwi,awj\right)}{{\mathrm{\&sigma;}}_{j=1}^x\#\left(qwi,awj\right)}|c=ck;$

Described question and answer construction of knowledge base unit, is suitable to as follows above-mentioned probability, single-minded degree be multiplied with intensity:

Weight (qwi, awj | c=ck)=p (ck | awj) * specific (qwi, awj | c=ck) * interpret (qwi, Awj | c=ck)；

Wherein, p (ck) represents the probability that classification ck occurs；P (awj) represents the probability that answer is awj；P (awj │ ck) represents ck Classification belongs to the probability of awj；

# (qwi, awj) problem of representation word is qwi and answer word is the number of times of awj；

# (awj) represents the number of times that answer word is awj.

7. a kind of method of the associated degree obtaining question and answer pair, the method comprises the steps:

Problem content to question and answer pair to be analyzed and answer content carry out word and extract operation, obtain at least one and to be analyzed ask Epigraph language and at least one answer word to be analyzed；

According to problem word to be analyzed and answer word to be analyzed, select from the question and answer knowledge base including a plurality of question and answer knowledge record At least one question and answer knowledge record, calculates the associated degree of question and answer pair to be analyzed according to selected question and answer knowledge record；

Wherein, described according to problem word to be analyzed with answer word to be analyzed, select at least one question and answer from question and answer knowledge base Knowledge record, calculates the associated degree of question and answer pair to be analyzed, specifically includes according to selected question and answer knowledge record: chooses The problem word that it includes is with problem word match to be analyzed and the asking of the answer word and the answer word match to be analyzed that include Answer knowledge record；According to the question and answer knowledge record corresponding to identical category in the question and answer knowledge record of described selection, obtain this and treat The question and answer of analysis are to the associated degree for each classification；Choose this question and answer to be analyzed to be associated to for each classification The maximum of degree, using this maximum as the associated degree of question and answer pair to be analyzed.

8. method according to claim 7, wherein, the method further includes:

Extract multiple question and answer pair from the webpage containing question and answer pair in advance, according to the question and answer extracted, a plurality of question and answer are included to structure and know The question and answer knowledge base of memorize record；

When extracting multiple question and answer pair from the webpage containing question and answer pair, crawl is with described question and answer to corresponding classification；

When according to the question and answer extracted to building question and answer knowledge base, according to question and answer to and with described question and answer, corresponding classification is built Question and answer knowledge record；

Each question and answer knowledge record corresponds to a classification, includes a problem word, an answer word respectively, and described Semantic relevancy between problem word and described answer word.

9. method according to claim 8, wherein, described according to question and answer to and with described question and answer, corresponding classification is built Question and answer knowledge base, specifically includes:

To each question and answer pair, the problem content to this question and answer pair and answer content carry out word and extract operation, obtain problem word Set and answer set of words；

Each the problem word in problem set of words is made to ask with this respectively with each the answer word in answer set of words Answer questions one information record of formation in each classification corresponding；

To each information record, execution is following to be operated:

Calculate the probability that this answer word belongs to the category, calculate this answer word explanation to this problem word in the category Single-minded degree, calculate this problem word is explained in the category intensity with this answer word；

Above-mentioned probability, single-minded degree are multiplied with intensity, obtained product is the semanteme of this answer word and this problem word Degree of association；

This problem word, this answer word and its semantic relevancy is made to form a question and answer knowledge record corresponding to the category.

10. the method according to any claim in claim 7-9, wherein,

According to the question and answer knowledge record corresponding to identical category in the question and answer knowledge record of described selection, obtain this to be analyzed asking Answer questions the associated degree being respectively directed to each classification, specifically include:

Correspond to the semantic relevancy weighting summation of the question and answer knowledge record of identical category in the question and answer knowledge record that will choose, obtain To this question and answer to be analyzed to the associated degree being respectively directed to each classification.

11. methods according to claim 9, wherein,

Described this answer word of calculating belongs to the probability of the category, specifically includes:

$p\left(ck\right|awj)=p\left(awj\right|ck)*\frac{p\left(ck\right)}{p\left(awj\right)};$

The described single-minded degree calculating the explanation to this problem word for each answer word in the category, specifically includes:

$specific(qwi,awj|c=ck)=p\left(qwi\right|awj,c=ck)=\frac{\#(qwi,awj)}{\#\left(awj\right)}|c=ck;$

The described calculating intensity that this problem word is explained with each answer word in the category, specifically includes:

$\mathrm{int}erpret\left(qwi,awj|c=ck\right)=p\left(awj|qwi,c=ck\right)=\frac{\#\left(qwi,awj\right)}{{\mathrm{\&sigma;}}_{j=1}^x\#\left(qwi,awj\right)}|c=ck;$

Above-mentioned probability, single-minded degree are multiplied with intensity, specifically include:

Weight (qwi, awj | c=ck)=p (ck | awj) * specific (qwi, awj | c=ck) * interpret (qwi, Awj | c=ck)；

Wherein, p (ck) represents the probability that classification ck occurs；P (awj) represents the probability that answer is awj；P (awj │ ck) represents ck Classification belongs to the probability of awj；

# (qwi, awj) problem of representation word is qwi and answer word is the number of times of awj；

# (awj) represents the number of times that answer word is awj.

12. methods according to claim 7-9 any claim, wherein,

The described problem content to described question and answer pair to be analyzed and answer content carry out word and extract operation, specifically include: right The problem content of question and answer pair to be analyzed and answer content carry out participle, remove stop words, word merging, and the behaviour extracting entity word Make.