CN1335574A - Intelligent semantic searching method - Google Patents

Abstract

The intelligent semantic searching method of the present invention includes establishing learning computer to update the knowledge base timely through processing the clinks of the user, and sorting and re-sorting the search result according to the degree of association for the user to check information fast. Thus, the present invention makes it possible to adapt the user's requirement and the change in language habits and to update knowledge base for efficient and complete information service.

Description

Intelligent semantic searching method

Technical field

The present invention relates to the method that a kind of user of permission adopts the intelligent search that natural language retrieves.

Background technology

Internet fast development and extensive popularizing cause the network information explosive increase.The netizen that appears as of search technique finds information needed to bring Gospel fast.Based on traditional Suo Fangfa, mainly contain three type of search engines, i.e. robot searches engine (Robot), catalogue formula search engine (Directory or Catalog) and META Search Engine (Meta).Traditional searching method mainly adopts keyword to mate the retrieval that realizes information, because the user is difficult to verily express content retrieved that he really needs with keyword or keyword string simply, express difficulty and cause the retrieval difficulty, and natural language is along with the time, the change in region or field, identical concept can be expressed with different language Symbols forms, retrieval to identical concept, different users may use different keywords to inquire about, the result causes a large amount of irrelevant informations to return the user, in addition, traditional searching method is often only described the relevant website of preserving and is searched for, the essential dynamic change of website itself can not promptly advance to be reflected in the Search Results, the user can not query web in important information in dynamically, cause information dropout.

For addressing the above problem, in recent years, intelligent searching engine based on the intelligent search method has appearred, as block in the In-particular, move about unhurriedly, Askjeeves, Google etc., take semantic analysis technology based on knowledge base, utilize means such as participle technique, phrase identification, processing synonym, user's search instruction analyzed, comprise step following time:

(1) reception user's retrieval request;

(2) by the dictionary for word segmentation in the knowledge base participle is carried out in request;

(3) by the syntax rule in the knowledge base request statement is carried out grammatical analysis;

(4) find out the synonym of request by the dictionary in the knowledge base;

(5) by the index database retrieving files in the knowledge base;

(6) Search Results is submitted to the user.

Such as " Beijing weather how? " traditional searching method can with " Beijing weather how? " arranged in the article the novel of the words is all retrieved out, and draw " Beijing weather " this notion by semantic analysis, directly provide Beijing weather condition report, thereby improved recall precision, reduced garbage, the line correlation association that goes forward side by side provides and the relevant information of user search instruction, the clue of further inquiry is provided, realizes intelligent navigation inquiry.But search engine based on this intelligent search method, after its knowledge base of being made up of syntax rule table, index database etc. is built up, just can not with the exchanging of user in carry out self-adaptation, when user's user demand, when speech habits change, can not in time upgrade knowledge base.

Summary of the invention

The object of the present invention is to provide intelligent semantic searching method, can constantly adapt to the variation of user's user demand, speech habits, the knowledge base that upgrades in time realizes efficient, complete information service.

Intelligent semantic searching method of the present invention comprises following steps

(1). receive the retrieval request that the user proposes.

(2). by the semantic dictionary in the knowledge base request statement is carried out participle and related term analysis.

(3). by the syntax rule storehouse in the knowledge base minute vocabulary is carried out grammatical analysis, and carry out semantic analysis, and form the semantic search table according to relevant vocabulary.

(4). according to the semantic search table, the index database of knowledge base is retrieved, and Search Results is sorted.

(5). Search Results is resequenced.And submit to the user, and whether the user who detects in effective time have click to Search Results, if "Yes" then carry out

Step (six); If "No" then execution in step (nine).

(6). record click situation.

(7). click the intellectual analysis of situation, upgrade the semantic search table.

(8). repeated execution of steps (five)

(9). intellectual analysis is carried out in the change to clauses and subclauses in the semantic search structure, syntax rule table and semantic dictionary, flow process finishes, and finishes learning process one time.

Because intelligent semantic searching method of the present invention has been set up learning machine, by processing to user click condition, knowledge base is upgraded in time, and Search Results sorted and resequence according to the degree of association, make user's energy fast query information, thereby can constantly adapt to the variation of user's user demand, speech habits, the knowledge base that upgrades in time realizes efficient, complete information service.

Description of drawings

Below in conjunction with drawings and Examples the present invention is described in further detail:

Fig. 1 is a workflow diagram of the present invention, comprising the structure of knowledge base and the workflow of learning machine;

Fig. 2 is a semantic dictionary entry data structure of the present invention;

Fig. 3 is the interior clauses and subclauses storage node composition of index database that the index point of semantic dictionary clauses and subclauses of the present invention points to;

Fig. 4 is the structural drawing of the phrase data base in the syntax library of the present invention;

Fig. 5 is a semantic search table of the present invention---the structural drawing of actual search.

The specific embodiment

Intelligent semantic searching method of the present invention comprises following steps

(1) receive the retrieval request that the user proposes, for example, " computer application ".

(2) by the dictionary for word segmentation in the knowledge base request statement is carried out participle,, as be divided into: " computer " and " application " two words; The line correlation word analysis of going forward side by side is such as " computer " Class father's word be " machine ", synonym is " computer ", " PC ", subclass be " server ", " PC ", other related term are " keyboard ", " CPU ", " software ", separately relevant of record.

(3) to by the syntax rule storehouse in the knowledge base minute vocabulary being carried out syntactic analysis, And carry out semantic analysis according to relevant vocabulary, and form the semantic search table. Such as " Computer application With " be inverted moving guest's phrase, and be a general phrase, then it is relevant " soft The part exploitation " also should be at the row of being searched. Each clauses and subclauses of semantic search table comprise actual will the search Entry, the semantic relevancy of entry, and produce this semantic syntax rule, relevant vocabulary Clauses and subclauses. Such as this semantic interpretation of " computer application " and " software development ", wherein one Individual reason is that " computer " is related term with " software ", thereby this association has been credited to In the semantic search table in " software development " clauses and subclauses.

(4) according to the semantic search table, the index database of knowledge base is retrieved, and to searching Hitch really sorts. Putting in order of Search Results is to go out according to searched object in the Search Results The situations such as existing frequency, position, density and the calculation of relationship degree in the semantic search table.

(5) Search Results is resequenced. And submit to the user, and detect effective time In the user whether Search Results is had click, if "Yes" then execution in step (six); As The fruit "No" is execution in step (eight) then.

(6) record click situation.

(7) intellectual analysis of click situation, update semantics search list: recall the process that current click object is retrieved out, find out relevant semantic search table clause, and combination click is in the past recorded and intellectual analysis is carried out in current click, then the degree of correlation of the relevant entry in the update semantics searching structure body

(8) repeated execution of steps (five). Along with the increase of the number of times of clicking, needed Search Results will all be arranged in the front.

(9) intellectual analysis is carried out in the change of clauses and subclauses in the semantic search structure, grammer rule Then show and semantic dictionary, upgrade the degree of correlation of clauses and subclauses wherein. Finish the once process of study.

Claims (1)

1, intelligent semantic searching method is characterized in that: comprise following steps:

(1). receive the retrieval request that the user proposes.

(2). by the semantic dictionary in the knowledge base request statement is carried out participle and related term analysis.

(3). by the syntax rule storehouse in the knowledge base minute vocabulary is carried out grammatical analysis, and carry out semantic analysis, and form the semantic search table according to relevant vocabulary.

(4). according to the semantic search table, the index database of knowledge base is retrieved, and Search Results is sorted.

(5). Search Results is resequenced.And submit to the user, and whether the user who detects in effective time have click to Search Results, if "Yes" then carry out the step

Suddenly (six); If "No" then execution in step (nine).

(6). record click situation.

(7). click the intellectual analysis of situation, upgrade the semantic search table.

(8). repeated execution of steps (five)

(9). intellectual analysis is carried out in the change to clauses and subclauses in the semantic search structure, syntax rule table and semantic dictionary, flow process finishes, and finishes learning process one time.

Images