CN102096717B

CN102096717B - Search method and search engine

Info

Publication number: CN102096717B
Application number: CN 201110038433
Authority: CN
Inventors: 刘建柱
Original assignee: Beijing Baidu Netcom Science and Technology Co Ltd
Current assignee: Baidu Online Network Technology Beijing Co Ltd; Beijing Baidu Netcom Science and Technology Co Ltd
Priority date: 2011-02-15
Filing date: 2011-02-15
Publication date: 2013-01-16
Anticipated expiration: 2031-02-15
Also published as: CN102096717A

Abstract

The invention provides a search method. The search method comprises the following steps of: S1, receiving a query command; S2, performing demand intention analysis on the query command based on a knowledge base, and defining the demand intention of the query command; S3, searching the query command carrying the demand intention in a database to obtain a search result; and S4, outputting the search result. Compared with the prior art, the search method has the advantages that: on the basis of the knowledge base, the query command input by a user is understood better; the intention of the query command is analyzed; the structure of the query command is analyzed to perform semantic content expansion on the query command so as to better guide a search engine to select quality resources to meet the search requirements on the user; therefore, the search efficiency of the user is improved and the network traffic is saved.

Description

Searching method and search engine

Technical field

The present invention relates to search engine technique, relate in particular to a kind of searching method and search engine that query statement is carried out demand analysis and parsing based on knowledge base.

Background technology

Growth at full speed along with internet information has been full of increasing redundant information on the network, and for the Internet user who searches own required information at network, faces these information that extend endlessly undoubtedly as looking for a needle in a haystack.The appearance of search engine has brought convenience for user's search need undoubtedly to a certain extent.Search engine is a kind of software systems of using at network, and it is collected and discovery information at network with certain strategy, and after information being processed and organized, for the user provides information search service on the internet.Usually, this software systems provide a web interface, allow the user submit search word in client to by browser software, then return very soon one may with the relevant information list of search content of user input.This tabulation can comprise up to ten thousand clauses and subclauses usually, and each clauses and subclauses represents one piece of related web page that searches.

Since more than ten years in past, correspondingly, arise at the historic moment in numerous internet search engines and corresponding website, the outstanding person in the middle of this comprises Baidu's search (www.baidu.com) of company of Baidu and Google's search (www.google.cn) of Google.

Existing search engine is based on the understanding of query statement character mostly to the query statement of user's input, for example, the user input query instruction is " Nokia mobile phone ", can only be " Nokia " and " mobile phone " with this query statement participle based on existing search engine, and retrieve in the web database index by this word segmentation result, text is comprised the webpage Url input of " Nokia " and " mobile phone ", form Search Results, yet this search engine can not carry out to user's query statement the understanding on content and the semantic hierarchies, for example, the user input query instruction is " Nokia mobile phone ", and it can not be interpreted as this query statement that " Nokia " is a kind of brand in " mobile phone "; Certainly, more can not understand the demand intention of query statement, and the structure of query statement, can not carry out semantic content expansion etc. to query statement.Expression-form variation, demand for user's input are intended to diversified query statement, the existing demand that can not better meet the user based on the search engine of character, cause be used to searching incomplete, need to repeatedly input the Search Results that different query statements just may find to be needed, search efficiency is lower, the problem of waste Internet resources.

Summary of the invention

The object of the present invention is to provide a kind of improved searching method, it can better understand the query statement of user's input on the basis of knowledge base, the intention of analysis and consult instruction, and the structure of parsing query statement is carried out semantic content to query statement and is expanded.

The present invention also aims to provide a kind of improved search engine of realizing above-mentioned searching method.

One of for achieving the above object, first embodiment of the invention provides a kind of searching method, may further comprise the steps:

S1, reception query statement;

S2, based on knowledge base described query statement is carried out the analysis of demand intention, the demand intention of clear and definite described query statement;

S3, will in database, search for the described query statement of demand intention, obtain Search Results;

S4, export described Search Results.

As a further improvement on the present invention, described database is web page repository or the vertical search database corresponding with described demand intention.

As a further improvement on the present invention, between described S2 step and S3 step, also comprise the semantic step that expands:

Based on described knowledge base described query statement being carried out semanteme expands.

As a further improvement on the present invention, described " based on knowledge base described query statement is carried out the demand intention and analyze, the demand intention of clear and definite described query statement " specifically comprises following flow process:

S200, by each demand intention marking of user's historical behavior storehouse to each knowledge fragment in the knowledge base, make each knowledge fragment all have corresponding demand intention score;

S201, with described query statement and knowledge fragment match, at least one the knowledge fragment that obtains being complementary with described query statement;

The demand intention score of S202, the knowledge fragment that will be complementary with described query statement adds up, and obtains the first mark;

S203, the subordinate relation of knowledge fragment in described knowledge base by being complementary with described query statement are added and subtracted described the first mark, obtain knowledge base integrated demand score;

S204, judge that whether described knowledge base integrated demand score is greater than a setting threshold;

If S205 is greater than described setting threshold, then with the demand intention of the highest demand type of described knowledge base integrated demand score as described query statement;

If S206, judges then that described query statement is without obvious demand intention less than described setting threshold.

One of for achieving the above object, second embodiment of the invention provides a kind of searching method, may further comprise the steps:

S1, reception query statement;

S2, based on knowledge base described query statement is carried out the analysis of demand intention, the demand intention of clear and definite described query statement simultaneously, is carried out semanteme expansion based on described knowledge base to described query statement;

S3, will and expand semantic query statement and in database, search for demand intention, obtain Search Results;

S4, export described Search Results.

One of for achieving the above object, third embodiment of the invention provides a kind of searching method, may further comprise the steps:

S1, reception query statement;

S2, based on knowledge base and expression template storehouse described query statement is carried out the demand intention and analyze the demand intention of clear and definite described query statement;

S4, export described Search Results.

As a further improvement on the present invention, described " based on knowledge base With the expression template storehouseDescribed query statement is carried out the analysis of demand intention, the demand intention of clear and definite described query statement " specifically comprise following flow process:

S201, with described query statement and knowledge fragment and expression template coupling, at least one the knowledge fragment and the expression template that obtain being complementary with described query statement;

S204, described query statement is given a mark in the expression template aspect, obtain the expression template score;

S205, with the weighting sum of knowledge base integrated demand score and expression template score as query statement demand intensity score;

S206, judge that whether described query statement demand intensity score is greater than a setting threshold;

If S207 is greater than described setting threshold, then with the demand intention of the highest demand type of query statement demand intensity score as described query statement;

If S208, judges then that described query statement is without obvious demand intention less than described setting threshold.

As a further improvement on the present invention, the construction method in described expression template storehouse comprises following flow process:

S300, extraction comprise the query statement of knowledge fragment in user's historical behavior storehouse;

S301, described knowledge base fragment is replaced to general symbol(s), generate candidate's expression template;

The quantity of the knowledge base fragment that described candidate's expression template that S302, statistics generate meets;

S303, judge that whether described quantity is greater than setting threshold;

If S304 is greater than setting threshold, then with described candidate's expression template as expression template, and be stored in the database, generate the expression template storehouse;

If S305, then gives up described candidate's expression template less than setting threshold.

One of for achieving the above object, four embodiment of the invention provides a kind of searching method, may further comprise the steps:

S1, reception query statement;

S2, based on knowledge base and expression template storehouse described query statement is carried out the demand intention and analyze, the demand intention of clear and definite described query statement simultaneously, is carried out semanteme expansion based on described knowledge base to the query statement that receives;

S4, export described Search Results.

S303, judge that whether described quantity is greater than setting threshold;

Correspondingly, as realizing another purpose of foregoing invention, an embodiment of the present invention provides a kind of search engine, comprising:

The UI module be used for to receive query statement, and described UI module also is used for receiving the Search Results that search module returns, and exports after described Search Results is assemblied into results page;

Demand intention analysis module is used for based on knowledge base described query statement being carried out the demand intention and analyzes the demand intention of clear and definite described query statement;

Search module is used for and will searches at database with the described query statement of demand intention, obtains Search Results;

Knowledge base is used for the storage priori.

As a further improvement on the present invention, described search engine also comprises:

The web services module, be used for receiving the query statement that client is sent by procotol, and forward described query statement to described UI module, and described web services module also is used for receiving the results page that described UI module is returned, and described results page is back to described client.

User's historical behavior storehouse is used for storage user historical search record.

As a further improvement on the present invention, described user's historical search record comprises: query statement, inquiry times, and weighting clicks.

Expression template is excavated module, is used for according to the knowledge fragment of described knowledge base and the user's historical query instruction in described user's historical behavior storehouse, excavates expression template, and described expression template is stored in the expression template storehouse;

The expression template storehouse is used for storage and excavates the expression template that module is excavated by described expression template.

The textural classification module is used for based on described knowledge base described query statement being carried out semanteme and expands.

As a further improvement on the present invention, described web page repository is used for the index information of storage network page data and this web data;

Described vertical search database is used for the index information of storage particular category data and these particular category data.

Compared with prior art, the invention has the beneficial effects as follows: on the basis of knowledge base, better understand the query statement of user's input, the intention of analysis and consult instruction, resolve the structure of query statement, query statement is carried out semantic content expand, thereby better the guidance search engine selects the resource of high-quality to satisfy user's search need, so that user search efficient improves, save network traffics.

Description of drawings

Fig. 1 is that search engine of the present invention and client realize interactive fundamental diagram;

Fig. 2 is the module map of search engine the first embodiment of the present invention;

Fig. 3 is the module map of search engine the second embodiment of the present invention;

Fig. 4 is the module map of search engine the 3rd embodiment of the present invention;

Fig. 5 is the module map of search engine the 4th embodiment of the present invention;

Fig. 6 is the synoptic diagram of knowledge base framework of the present invention;

Fig. 7 is the process flow diagram of searching method the first embodiment of the present invention;

Fig. 8 is the process flow diagram of searching method the second embodiment of the present invention;

Fig. 9 is the process flow diagram of searching method the 3rd embodiment of the present invention;

Figure 10 is the process flow diagram of searching method the 4th embodiment of the present invention;

Figure 11 is the process flow diagram that the present invention " carries out the demand intention based on knowledge base to described query statement and analyze, the demand intention of clear and definite described query statement " the step 1 embodiment;

Figure 12 is the process flow diagram that the present invention " carries out the demand intention based on knowledge base to described query statement and analyze, the demand intention of clear and definite described query statement " another embodiment of step;

Figure 13 is the process flow diagram of the construction method in expression template of the present invention storehouse;

Figure 14 is the synoptic diagram of the online interface of the present invention one embodiment;

Figure 15 is the new page synoptic diagram in " jumping to new page when the user clicks vertical search result ".

Embodiment

Describe the present invention below with reference to each embodiment shown in the drawings.But these embodiments do not limit the present invention, and the conversion on the structure that those of ordinary skill in the art makes easily according to these embodiments, method or the function all is included in protection scope of the present invention.

Search engine of the present invention 10 shown in Figure 1 is realized interactive fundamental diagram with client 20.In the present embodiment, this client 20 comprises a browser 201, the client can open the online webpage of showing of search engine by this browser 201, and input inquiry instruction in the dialog box in webpage, general, the query statement of this input is text message, and certainly, this query statement can also be pictorial information, video information etc.Described search engine 10 receives the client by network and inputs to query statement in the described browser, and after this query statement searched for, Search Results is back to this browser 201 by the online displayed web page of search engine.Wherein, this search engine 10 can comprise one or more server, this client 20 can comprise one or more subscriber terminal equipments, such as personal computer, notebook computer, wireless telephone, personal digital assistant (PDA) or other department of computer science communication system of unifying.

These servers and terminal device all comprise some basic modules on framework, such as bus, disposal system, storage system, one or more input/output and communication interface etc.Bus can comprise one or more wires, is used for realizing the communication between server or each assembly of terminal device.Disposal system comprises that all types of being used for carry out processor or the microprocessor of instruction, treatment progress or thread.Storage system can comprise the dynamic storagies such as random access storage device (RAM) of storing multidate information, with the static memories such as ROM (read-only memory) (ROM) of storage static information, and the mass storage that comprises magnetic or optical record medium and respective drive.Input system arrives server or terminal device for user's input information, such as keyboard, mouse, writing pencil, sound recognition system or bioassay system etc.Output system comprises display, printer, loudspeaker of output information etc.Communication interface is used for making server or terminal device and other system or system to communicate.Can be connected in the network by wired connection, wireless connections or light between the communication interface, make search engine 10,20 of clients realize mutual communication by network.Network can comprise the combination etc. of internet, the Internet or above-mentioned these networks of Local Area Network, wide area network (WAN), telephone network such as public switch telephone network (PSTN), enterprises.

All include management of system resource on server and the terminal device, control the operating system software that other program is moved, and the application software that is used for realizing certain functional modules.As shown in Figure 2, in first embodiment of the invention, described search engine has comprised web services module 101, UI module 102 with web services module 101 interactive communications, the demand intention analysis module 103 of communicating by letter with described UI module 102, the Structural Analysis Module 104 of communicating by letter with described demand intention analysis module 103, the search module 105 of communicating by letter with described Structural Analysis Module 104, and with described demand intention analysis module 103, the knowledge base 106 of described Structural Analysis Module 104 interactive communications, user's historical behavior storehouse 107 of communicating by letter with described knowledge base 106, with described knowledge base 106, the expression template of user's historical behavior storehouse 107 communications is excavated module 108, excavate the expression template storehouse 109 that module 108 is communicated by letter with described demand intention analysis module 103 with described expression template, and the web page repository 110 of communicating by letter with described search module 105.It is worth mentioning that these modules can store and run in the same server, also can store and operate in the multiple servers.

Described web services module 101 is used for receiving the query statement that transmits from client 20 by procotol, and forward this query statement to UI module 102, in addition, this web services module 101 also is used for receiving the results page that described UI module 102 is returned, and described results page is back to client 20.

Described UI module 102 is used for receiving the query statement that described web services module 101 transmits, and this query statement is sent to described query statement analysis module 103; In addition, described UI module 102 also is used for receiving the Search Results that described search module 104 returns, and after described Search Results is assemblied into results page, returns described results page to described web services module 101.

Described demand intention analysis module 103 is used for calling described knowledge base 106, user's historical behavior storehouse 107, and described expression template storehouse 109, analyzes the demand intention of clear and definite described query statement the query statement that receives is carried out the demand intention.In the present invention, described intention analysis module 103 is at first given each demand intention marking of each knowledge fragment in the described knowledge base 106 by described user's historical behavior storehouse 107, concrete: the user is when certain class demand of inquiry, can click the result who satisfies accordingly his demand, want to obtain the relevant information of AUTO QUOTED PRICE such as the user, after search engine input inquiry instruction " bright ease ", the Url of the car website that meeting click search engine returns, such as " Netease garage ", the Url " Netease garage " that query statement " bright ease " fragment of user's input this moment and user click implies the reflection user the demand of looking for the car relevant information, based on this point, the present invention is when calculating the demand intention of each knowledge fragment, click total Url that number/certain knowledge fragment is clicked of certain class Url according to certain knowledge fragment, determine the score of this knowledge fragment demand intention, as as described in learn in user's historical behavior storehouse 107, query statement is " bright ease ", total Url number of its click is 10, wherein, commodity class Url is 5, news category Url is 3, picture category Url is 2, the demand that then can calculate the commodity class of this query statement is intended to 0.5, the demand of news category is intended to 0.3, and the demand of picture category is intended to 0.2; Secondly, behind a query statement that receives user's input, need the knowledge coupling through knowledge base 106, obtain being present in the described query statement knowledge fragment in the described knowledge base 106, and the knowledge base integrated demand intensity of the described query statement of COMPREHENSIVE CALCULATING.For example the user inputs " the bright ease quotation of Shanghai Volkswagen ", then can obtain the knowledge fragment of " Shanghai Volkswagen " " bright ease " by knowledge base.After obtaining preliminary information, at first knowledge fragment " Shanghai Volkswagen " and " bright ease " demand intention score are separately added up, obtain the first mark, secondly, by the relation of knowledge fragment " Shanghai Volkswagen " with " bright ease ", add and subtract described the first mark again, obtain knowledge base integrated demand score, in best mode for carrying out the invention, if the pass of knowledge fragment is to belong to relation together, bonus point then; If the knowledge fragment is the non-relation that belongs to together, then subtract minute; If this knowledge base integrated demand score is greater than the threshold value of setting, then with the demand intention of the highest demand type of knowledge base integrated demand score as query statement, and according to described demand intention corresponding tag information of adding in described query statement, such as " commodity ", " news ", " picture " etc.It is worth mentioning that: in preferred forms of the present invention, except calculation knowledge storehouse integrated demand gets exceptionally, when the analysis demand is intended to, also can consider the marking on the expression template aspect: behind a query statement that receives user's input, need the expression template coupling through expression template storehouse 109, obtain being present in the described query statement expression template fragment in the described expression template storehouse 109, for example user's input " the bright ease quotation of Shanghai Volkswagen " then identifies " XX quotation " template that exists in the query statement by user template.In according to said method acquire knowledge storehouse integrated demand score, query statement meets again user profile, also give a mark in the expression template aspect to described query statement in expression template storehouse 109, obtain the expression template score, then the demand intensity of whole query statement must be divided into the weighting sum of knowledge base integrated demand score and expression template score, if this weighting sum is greater than the threshold value of setting, then the demand type of weighting sum maximum is as the demand intention of query statement, and according to described demand intention adding corresponding tag information, for example " commodity " in described query statement, " news ", " picture " etc.

Described textural classification module 104 is used in conjunction with described knowledge base 106, to carrying out being sent to described search module 105 after the intelligent conversion through the query statement behind the described demand intention analysis module 103, wherein, described intelligent conversion namely is that semantic content expands, described semantic content expands the normalization that has comprised semantic content, and the expansion of semantic classes.Concrete, when described query statement has the relation of belonging to together (the knowledge fragment of the knowledge fragment of upper bit attribute+lower bit attribute), for example, described query statement is " mobile phone Nokia ", at this moment, described textural classification module 104 is when sending to described search module 105, namely can " mobile phone " on this knowledge fragment of bit attribute add the tag of " can abandon ", like this, when 105 pairs of described query statements of search module are searched for, can search for by " mobile phone Nokia ", also can search for by " Nokia ", and also can think with the webpage of " mobile phone Nokia " text message with only the same with the webpage weights of " Nokia " text message; In addition, for example: if described query statement is " Nokia ", then described textural classification module 104 is when sending to described search module 105, also can according to " Nokia " corresponding on it bit attribute expand, as expand to " mobile phone ", like this, at search module 105 to " Nokia " when searching for, can judge whether that according to result's quantity needs expand to " mobile phone " and search for, such as the fruiting quantities that searches by " Nokia " hour, can expand to " mobile phone "; And for example: if described query statement is " mobile phone ", then described textural classification module 104 can be expanded " mobile phone " corresponding its with bit attribute, as expand to " computer ", such expansion can be used for the popularization of advertisement, can carry out advertisement promotion according to " mobile phone " this query statement such as the right side at searched page, can carry out advertisement promotion according to " computer " this query statement again; For another example, if described query statement is " Nokia ", then described textural classification module 104 is when sending to described search module 105, also can expand according to " Nokia " corresponding its lower bit attribute, as expand to " N71 ", " N72 " etc., like this, at search module 105 when searching with text message webpages such as " N71 ", " N72 ", also can according to the weights of these webpages, judge whether to export as Search Results.This weights are judged the weights judgement that can join in the existing search engine, do not repeat them here.Generally speaking, when described query statement is expanded, can expand the knowledge fragment of bit attribute on it according to the strategy of search, also can expand it with the knowledge fragment of bit attribute, also can expand the knowledge fragment of its lower bit attribute.

Described search module 105 is used for receiving the query statement through after the intelligent conversion of described demand intention analysis module 103 or described textural classification module 104, and this query statement searched in web page repository 110, to obtain Search Results, simultaneously, described search module 105 also is used for described Search Results is back to described UI module 102.

Described knowledge base 106 is used for the storage priori.In best mode for carrying out the invention, described knowledge base mainly is stored as tree structure, to knowledge base tree of each class construction of knowledge base, Father's Day point identification bit attribute on it by this knowledge base tree, the right brotgher of node represents its same bit attribute, the left brotgher of node represents its lower bit attribute, and iteration like this is until leaf node.As shown in Figure 6, " masses " are its upper attribute; Its lower bit attribute is " Shanghai Volkswagen "; Have " FAW-Volkswagen " with described " Shanghai Volkswagen " coordination; In described " Shanghai Volkswagen " the next having " bright ease ", with have " way is seen " of described " bright ease " coordination ... the construction method of this knowledge base, those of ordinary skill in the art can join prior art and finish, and does not repeat them here.

User's historical behavior storehouse 107 is used for storage user historical search record.Preferably, it can comprise query statement, inquiry times, and the information such as weighting clicks.

Expression template is excavated module 108 and is used for according to the knowledge fragment of described knowledge base 106 and the user's historical query instruction in described user's historical behavior storehouse 107, excavates expression template, and described expression template is stored in the described expression template storehouse 109.The user of same requirements, similarity can appear on expression way, described expression template refers to, the general user is when having certain query demand, the query statement of its input is why, for example, and as user during in inquiry automobile relevant information, expression way has: " Sagitar how ", " horse six power how " etc. wherein can extract the expression template of commonly using when " [automobile brand/model] how ", " [automobile brand/model] power how " etc. are expressed automobile demand.In one embodiment of the present invention, be specially: the query statement that at first comprises knowledge base 106 knowledge fragments in described user's historical behavior storehouse 107 extracts, as in the query statement of " horse six how ", " Skoda how ", " Sagitar how ", extract the knowledge fragment: " horse six ", " Skoda ", " Sagitar ", secondly the knowledge base fragment is replaced to " [automobile brand/model] " symbol, namely generate candidate's expression template of " [automobile brand/model] how "; Again, the quantity of the knowledge base fragment that candidate's expression template that statistics generates meets, if this quantity greater than the threshold value of setting, then with described candidate's expression template as expression template, be stored in the described expression template storehouse 109; If this quantity is then given up described candidate's expression template less than the threshold value of setting.

Described expression template storehouse 109 is used for storage and excavates the expression template that module 108 is excavated by described expression template.

Described web page repository 110 is used for the index information of storage network page data and this web data.This database namely is general search engine database commonly used, does not repeat them here.

As shown in Figure 3, in second embodiment of the invention, described search engine has comprised web services module 101, UI module 102 with web services module 101 interactive communications, the demand intention analysis module 103 of communicating by letter with described UI module 102, the Structural Analysis Module 104 of communicating by letter with described demand intention analysis module 103, the search module 105 of communicating by letter with described Structural Analysis Module 104, and with described demand intention analysis module 103, the knowledge base 106 of described Structural Analysis Module 104 interactive communications, user's historical behavior storehouse 107 of communicating by letter with described knowledge base 106, with described knowledge base 106, the expression template of user's historical behavior storehouse 107 communications is excavated module 108, excavate the expression template storehouse 109 that module 108 is communicated by letter with described demand intention analysis module 103 with described expression template, and the web page repository 110 of communicating by letter with described search module 105, the first vertical search data 111a, the second vertical search database 111b, ... .., N vertical search database 111n.It is worth mentioning that these modules can store and run in the same server, also can store and operate in the multiple servers.

Described demand intention analysis module 103 is used for calling described knowledge base 106, user's historical behavior storehouse 107, and described expression template storehouse 109, analyzes the demand intention of clear and definite described query statement the query statement that receives is carried out the demand intention.In the present invention, described intention analysis module 103 is at first given each demand intention marking of each knowledge fragment in the described knowledge base 106 by described user's historical behavior storehouse 107, concrete: the user is when certain class demand of inquiry, can click the result who satisfies accordingly his demand, want to obtain the relevant information of AUTO QUOTED PRICE such as the user, after search engine input inquiry instruction " bright ease ", the Url of the car website that meeting click search engine returns, such as " Netease garage ", the Url " Netease garage " that query statement " bright ease " fragment of user's input this moment and user click implies the reflection user the demand of looking for the car relevant information, based on this point, the present invention is when calculating the demand intention of each knowledge fragment, click total Url that number/certain knowledge fragment is clicked of certain class Url according to certain knowledge fragment, determine the score of this knowledge fragment demand intention, as as described in learn in user's historical behavior storehouse 107, query statement is " bright ease ", total Url number of its click is 10, wherein, commodity class Url is 5, news category Url is 3, picture category Url is 2, the demand that then can calculate the commodity class of this query statement is intended to 0.5, the demand of news category is intended to 0.3, and the demand of picture category is intended to 0.2; Secondly, behind a query statement that receives user's input, need the knowledge coupling through knowledge base 106, obtain being present in the described query statement knowledge fragment in the described knowledge base 106, and the knowledge base integrated demand intensity of the described query statement of COMPREHENSIVE CALCULATING.For example the user inputs " the bright ease quotation of Shanghai Volkswagen ", then can obtain the knowledge fragment of " Shanghai Volkswagen " " bright ease " by knowledge base.After obtaining preliminary information, at first knowledge fragment " Shanghai Volkswagen " and " bright ease " demand intention score are separately added up, obtain the first mark, secondly, by the relation of knowledge fragment " Shanghai Volkswagen " with " bright ease ", add and subtract described the first mark again, obtain knowledge base integrated demand score, in best mode for carrying out the invention, if the pass of knowledge fragment is to belong to relation together, bonus point then; If the knowledge fragment is the non-relation that belongs to together, then subtract minute; If this knowledge base integrated demand score is greater than the threshold value of setting, then with the demand intention of the highest demand type of knowledge base integrated demand score as query statement, and according to described demand intention corresponding tag information of adding in described query statement, such as " commodity ", " news ", " picture " etc.It is worth mentioning that: in preferred forms of the present invention, except calculation knowledge storehouse integrated demand gets exceptionally, when the analysis demand is intended to, also can consider the marking on the expression template aspect: behind a query statement that receives user's input, need the expression template coupling through expression template storehouse 109, obtain being present in the described query statement expression template fragment in the described expression template storehouse 109, for example user's input " the bright ease quotation of Shanghai Volkswagen " then identifies " XX quotation " template that exists in the query statement by user template.In according to said method acquire knowledge storehouse integrated demand score, query statement meets again user profile, also give a mark in the expression template aspect to described query statement in expression template storehouse 109, obtain the expression template score, then the demand intensity of whole query statement must be divided into the weighting sum of knowledge base integrated demand score and expression template score, if this weighting sum is greater than the threshold value of setting, then the demand type of weighting sum maximum is as the demand intention of query statement.

Described search module 105 is used for receiving the query statement through after the intelligent conversion of described demand intention analysis module 103 or described textural classification module 104, and with this query statement a plurality of vertical search databases (the first vertical search database 111a, the second vertical search database 111b ..., N equivalence class 111n) one of them, and search in the described web page repository 1110, to obtain Search Results, simultaneously, described search module 105 also is used for described Search Results is back to described UI module 102.It is worth mentioning that: selecting certain vertical search database to carry out vertical search is to determine by the demand intention of query statement, for example: if the demand of query statement is intended to " commodity ", then search in commodity vertical search database; The demand of institute's query statement is intended to " picture ", then in picture vertical search database, search for, wherein, one or more result who in the vertical search database, searches, can be inserted among the result who in web page repository, searches, form whole Search Results.Described vertical search namely is to search under certain specific classification, and the searching method that it is concrete and system in the art more has prior art to disclose, and does not repeat them here.

Described the first vertical search database 111a, the second vertical search database 111b ..., N vertical search database 111n is used for the index information of storage particular category data and these particular category data, for example commodity data, commodity index; News data, news index; Image data, picture indices etc.

As shown in Figure 4, in third embodiment of the invention, described search engine has comprised web services module 101, UI module 102 with web services module 101 interactive communications, the demand intention analysis module 103 of communicating by letter with described UI module 102, the Structural Analysis Module 104 of communicating by letter with described UI module 102, the search module 105 of communicating by letter with described Structural Analysis Module 104, and with described demand intention analysis module 103, the knowledge base 106 of described Structural Analysis Module 104 interactive communications, user's historical behavior storehouse 107 of communicating by letter with described knowledge base 106, with described knowledge base 106, the expression template of user's historical behavior storehouse 107 communications is excavated module 108, excavate the expression template storehouse 109 that module 108 is communicated by letter with described demand intention analysis module 103 with described expression template, and the web page repository 110 of communicating by letter with described search module 105.It is worth mentioning that these modules can store and run in the same server, also can store and operate in the multiple servers.

Described textural classification module 104 is used in conjunction with described knowledge base 106, query statement to 102 inputs of UI module carries out being sent to described search module 105 after the intelligent conversion, wherein, described intelligent conversion namely is that semantic content expands, described semantic content expands the normalization that has comprised semantic content, and the expansion of semantic classes.Concrete, when described query statement has the relation of belonging to together (the knowledge fragment of the knowledge fragment of upper bit attribute+lower bit attribute), for example, described query statement is " mobile phone Nokia ", at this moment, described textural classification module 104 is when sending to described search module 105, namely can " mobile phone " on this knowledge fragment of bit attribute add the tag of " can abandon ", like this, when 105 pairs of described query statements of search module are searched for, can search for by " mobile phone Nokia ", also can search for by " Nokia ", and also can think with the webpage of " mobile phone Nokia " text message with only the same with the webpage weights of " Nokia " text message; In addition, for example: if described query statement is " Nokia ", then described textural classification module 104 is when sending to described search module 105, also can according to " Nokia " corresponding on it bit attribute expand, as expand to " mobile phone ", like this, at search module 105 to " Nokia " when searching for, can judge whether that according to result's quantity needs expand to " mobile phone " and search for, such as the fruiting quantities that searches by " Nokia " hour, can expand to " mobile phone "; And for example: if described query statement is " mobile phone ", then described textural classification module 104 can be expanded " mobile phone " corresponding its with bit attribute, as expand to " computer ", such expansion can be used for the popularization of advertisement, can carry out advertisement promotion according to " mobile phone " this query statement such as the right side at searched page, can carry out advertisement promotion according to " computer " this query statement again; For another example, if described query statement is " Nokia ", then described textural classification module 104 is when sending to described search module 105, also can expand according to " Nokia " corresponding its lower bit attribute, as expand to " N71 ", " N72 " etc., like this, at search module 105 when searching with text message webpages such as " N71 ", " N72 ", also can according to the weights of these webpages, judge whether to export as Search Results.This weights are judged the weights judgement that can join in the existing search engine, do not repeat them here.Generally speaking, when described query statement is expanded, can expand the knowledge fragment of bit attribute on it according to the strategy of search, also can expand it with the knowledge fragment of bit attribute, also can expand the knowledge fragment of its lower bit attribute.

As shown in Figure 5, in four embodiment of the invention, described search engine has comprised web services module 101, UI module 102 with web services module 101 interactive communications, the demand intention analysis module 103 of communicating by letter with described UI module 102, the Structural Analysis Module 104 of communicating by letter with described UI module 102, the search module 105 of communicating by letter with described Structural Analysis Module 104, and with described demand intention analysis module 103, the knowledge base 106 of described Structural Analysis Module 104 interactive communications, user's historical behavior storehouse 107 of communicating by letter with described knowledge base 106, with described knowledge base 106, the expression template of user's historical behavior storehouse 107 communications is excavated module 108, excavate the expression template storehouse 109 that module 108 is communicated by letter with described demand intention analysis module 103 with described expression template, and the web page repository 110 of communicating by letter with described search module 105, the first vertical search data 111a, the second vertical search database 111b, ..., N vertical search database 111n.It is worth mentioning that these modules can store and run in the same server, also can store and operate in the multiple servers.

Described textural classification module 104 is used in conjunction with described knowledge base 106,102 input inquiry instructions are carried out being sent to described search module 105 after the intelligent conversion to the UI module, wherein, described intelligent conversion namely is that semantic content expands, described semantic content expands the normalization that has comprised semantic content, and the expansion of semantic classes.Concrete, when described query statement has the relation of belonging to together (the knowledge fragment of the knowledge fragment of upper bit attribute+lower bit attribute), for example, described query statement is " mobile phone Nokia ", at this moment, described textural classification module 104 is when sending to described search module 105, namely can " mobile phone " on this knowledge fragment of bit attribute add the tag of " can abandon ", like this, when 105 pairs of described query statements of search module are searched for, can search for by " mobile phone Nokia ", also can search for by " Nokia ", and also can think with the webpage of " mobile phone Nokia " text message with only the same with the webpage weights of " Nokia " text message; In addition, for example: if described query statement is " Nokia ", then described textural classification module 104 is when sending to described search module 105, also can according to " Nokia " corresponding on it bit attribute expand, as expand to " mobile phone ", like this, at search module 105 to " Nokia " when searching for, can judge whether that according to result's quantity needs expand to " mobile phone " and search for, such as the fruiting quantities that searches by " Nokia " hour, can expand to " mobile phone "; And for example: if described query statement is " mobile phone ", then described textural classification module 104 can be expanded " mobile phone " corresponding its with bit attribute, as expand to " computer ", such expansion can be used for the popularization of advertisement, can carry out advertisement promotion according to " mobile phone " this query statement such as the right side at searched page, can carry out advertisement promotion according to " computer " this query statement again; For another example, if described query statement is " Nokia ", then described textural classification module 104 is when sending to described search module 105, also can expand according to " Nokia " corresponding its lower bit attribute, as expand to " N71 ", " N72 " etc., like this, at search module 105 when searching with text message webpages such as " N71 ", " N72 ", also can according to the weights of these webpages, judge whether to export as Search Results.This weights are judged the weights judgement that can join in the existing search engine, do not repeat them here.Generally speaking, when described query statement is expanded, can expand the knowledge fragment of bit attribute on it according to the strategy of search, also can expand it with the knowledge fragment of bit attribute, also can expand the knowledge fragment of its lower bit attribute.

As shown in Figure 7, the searching method of first embodiment of the invention may further comprise the steps:

S1, reception query statement; Preferably, this query statement be the user by the input of the browser on the client to web services module 101, this web services module 101 can forward this querying command to UI module 102 after obtaining described querying command;

S2, based on knowledge base described query statement is carried out the analysis of demand intention, the demand intention of clear and definite described query statement; Preferably, this step is finished by described demand intention analysis module 103;

S3, will in database, search for the described query statement of demand intention, obtain Search Results; Preferably, this step is finished by described search module 105;

S4, export described Search Results.Preferably, this step is finished in described UI module 102 and described web services module 101, Search Results is back to described UI module 102 from described search module 104, and after by described UI module 102 described Search Results being assemblied into results page, return described results page to described web services module 101, thereby be back to client browser by described web services module 101.

Wherein, the database in described S3 step namely can be web page repository 110, or the vertical search database corresponding with the demand intention; Certainly, also can comprise web page repository 110 and the vertical search database corresponding with described demand intention.

Between described S2 step and S3 step, also comprise the semantic step that expands:

Based on described knowledge base the query statement that receives being carried out semanteme expands; Preferably, this step is finished by Structural Analysis Module 104.

As shown in Figure 8, the searching method of second embodiment of the invention may further comprise the steps:

S1 ', reception query statement; Preferably, this query statement be the user by the input of the browser on the client to web services module 101, this web services module 101 can forward this querying command to UI module 102 after obtaining described querying command;

S2 ', based on knowledge base described query statement is carried out the analysis of demand intention, the demand intention of clear and definite described query statement simultaneously, is carried out semanteme expansion based on described knowledge base to the query statement that receives; Preferably, this step is finished by described demand intention analysis module 103 and described Structural Analysis Module 104;

S3 ', will and expand semantic query statement and in database, search for demand intention, obtain Search Results; Preferably, this step is finished by described search module 105;

S4 ', export described Search Results.Preferably, this step is finished in described UI module 102 and described web services module 101, Search Results is back to described UI module 102 from described search module 104, and after by described UI module 102 described Search Results being assemblied into results page, return described results page to described web services module 101, thereby be back to client browser by described web services module 101.

Wherein, the database in described S3 ' step namely can be web page repository 110, or the vertical search database corresponding with the demand intention; Certainly, also can comprise web page repository 110 and the vertical search database corresponding with described demand intention.

As shown in Figure 9, the searching method of third embodiment of the invention may further comprise the steps:

S10, reception query statement; Preferably, this query statement be the user by the input of the browser on the client to web services module 101, this web services module 101 can forward this querying command to UI module 102 after obtaining described querying command;

S20, based on knowledge base and expression template storehouse described query statement is carried out the demand intention and analyze the demand intention of clear and definite described query statement; Preferably, this step is finished by described demand intention analysis module 103;

S30, will in database, search for the described query statement of demand intention, obtain Search Results; Preferably, this step is finished by described search module 105;

S40, export described Search Results.Preferably, this step is finished in described UI module 102 and described web services module 101, Search Results is back to described UI module 102 from described search module 104, and after by described UI module 102 described Search Results being assemblied into results page, return described results page to described web services module 101, thereby be back to client browser by described web services module 101.

Wherein, the database in described S30 step namely can be web page repository 110, or the vertical search database corresponding with the demand intention; Certainly, also can comprise web page repository 110 and the vertical search database corresponding with described demand intention.

Between described S20 step and S30 step, also comprise the semantic step that expands:

As shown in figure 10, the searching method of four embodiment of the invention may further comprise the steps:

S10 ', reception query statement; Preferably, this query statement be the user by the input of the browser on the client to web services module 101, this web services module 101 can forward this querying command to UI module 102 after obtaining described querying command;

S20 ', based on knowledge base and expression template storehouse described query statement is carried out the demand intention and analyze, the demand intention of clear and definite described query statement simultaneously, is carried out semanteme expansion based on described knowledge base to the query statement that receives; Preferably, this step is finished by described demand intention analysis module 103 and described Structural Analysis Module 104;

S30 ', will and expand semantic query statement and in database, search for demand intention, obtain Search Results; Preferably, this step is finished by described search module 105;

S40 ', export described Search Results.Preferably, this step is finished in described UI module 102 and described web services module 101, Search Results is back to described UI module 102 from described search module 104, and after by described UI module 102 described Search Results being assemblied into results page, return described results page to described web services module 101, thereby be back to client browser by described web services module 101.

Wherein, the database in described S30 ' step namely can be web page repository 110, or the vertical search database corresponding with the demand intention; Certainly, also can comprise web page repository 110 and the vertical search database corresponding with described demand intention.

As shown in figure 11, in the searching method of first embodiment of the invention, the second embodiment, the 3rd embodiment, the 4th embodiment, one embodiment of described " based on knowledge base described query statement being carried out the demand intention analyzes; the demand intention of clear and definite described query statement " step comprises following flow process:

S200, by each demand intention marking of user's historical behavior storehouse to each knowledge fragment in the knowledge base, make each knowledge fragment all have corresponding demand intention score; Concrete: the user is when certain class demand of inquiry, can click the result who satisfies accordingly his demand, want to obtain the relevant information of AUTO QUOTED PRICE such as the user, after search engine input inquiry instruction " bright ease ", the Url of the car website that meeting click search engine returns, such as " Netease garage ", the Url " Netease garage " that query statement " bright ease " fragment of user's input this moment and user click implies the reflection user the demand of looking for the car relevant information, based on this point, the present invention is when calculating the demand intention of each knowledge fragment, click total Url that number/certain knowledge fragment is clicked of certain class Url according to certain knowledge fragment, determine the score of this knowledge fragment demand intention, as as described in learn in user's historical behavior storehouse 107, query statement is " bright ease ", total Url number of its click is 10, wherein, commodity class Url is 5, news category Url is 3, picture category Url is 2, the demand that then can calculate the commodity class of this query statement is intended to 0.5, the demand of news category is intended to 0.3, and the demand of picture category is intended to 0.2;

S201, after receiving a query statement of user input, with described query statement and knowledge fragment match, at least one the knowledge fragment that obtains being complementary with described query statement; For example the user inputs " the bright ease quotation of Shanghai Volkswagen ", then can obtain the knowledge fragment of " Shanghai Volkswagen " " bright ease " by knowledge base;

The demand intention score of S202, the knowledge fragment that will be complementary with described query statement adds up, and obtains the first mark; For example: knowledge fragment " Shanghai Volkswagen " and " bright ease " demand intention score separately adds up, and obtains the first mark;

S203, the subordinate relation of knowledge fragment in described knowledge base by being complementary with described query statement are added and subtracted described the first mark, obtain knowledge base integrated demand score; In best mode for carrying out the invention, if the pass of knowledge fragment is to belong to relation together, bonus point then; If the knowledge fragment is the non-relation that belongs to together, then subtract minute;

If S206, then judges described query statement less than described setting threshold and without obvious demand intention, searches for according to general search engine search mode, do not repeat them here.

As shown in figure 12, in the searching method of first embodiment of the invention, the second embodiment, the 3rd embodiment, the 4th embodiment, another embodiment of described " based on knowledge base and expression template storehouse described query statement being carried out the analysis of demand intention; the demand intention of clear and definite described query statement " step comprises following flow process:

S200 ', by each demand intention marking of user's historical behavior storehouse to each knowledge fragment in the knowledge base, make each knowledge fragment all have corresponding demand intention score; Concrete: the user is when certain class demand of inquiry, can click the result who satisfies accordingly his demand, want to obtain the relevant information of AUTO QUOTED PRICE such as the user, after search engine input inquiry instruction " bright ease ", the Url of the car website that meeting click search engine returns, such as " Netease garage ", the Url " Netease garage " that query statement " bright ease " fragment of user's input this moment and user click implies the reflection user the demand of looking for the car relevant information, based on this point, the present invention is when calculating the demand intention of each knowledge fragment, click total Url that number/certain knowledge fragment is clicked of certain class Url according to certain knowledge fragment, determine the score of this knowledge fragment demand intention, as as described in learn in user's historical behavior storehouse 107, query statement is " bright ease ", total Url number of its click is 10, wherein, commodity class Url is 5, news category Url is 3, picture category Url is 2, the demand that then can calculate the commodity class of this query statement is intended to 0.5, the demand of news category is intended to 0.3, and the demand of picture category is intended to 0.2;

S201 ', after receiving a query statement of user input, described query statement and knowledge fragment and the expression template that is stored in the expression template storehouse are mated at least one the knowledge fragment that obtains being complementary with described query statement and an expression template; For example the user inputs " the bright ease quotation of Shanghai Volkswagen ", then can obtain the knowledge fragment of " Shanghai Volkswagen " " bright ease " by knowledge base; Obtain the expression template of " the XX quotation " that exist in the query statement by the expression template Cook;

The demand intention score of S202 ', the knowledge fragment that will be complementary with described query statement adds up, and obtains the first mark; For example: knowledge fragment " Shanghai Volkswagen " and " bright ease " demand intention score separately adds up, and obtains the first mark;

S203 ', the subordinate relation of knowledge fragment in described knowledge base by being complementary with described query statement are added and subtracted described the first mark, obtain knowledge base integrated demand score; In best mode for carrying out the invention, if the pass of knowledge fragment is to belong to relation together, bonus point then; If the knowledge fragment is the non-relation that belongs to together, then subtract minute;

S204 ', described query statement is given a mark in the expression template aspect, obtain the expression template score;

S205 ', with the weighting sum of knowledge base integrated demand score and expression template score as query statement demand intensity score;

S206 ', judge that whether described query statement demand intensity score is greater than a setting threshold;

S207 ' is if greater than described setting threshold, then with the demand intention of the highest demand type of query statement demand intensity score as described query statement;

S208 ' judges that then described query statement is without obvious demand intention if less than described setting threshold.

As shown in figure 13, in the searching method of third embodiment of the invention, the 4th embodiment, the construction method in described expression template storehouse comprises following flow process:

S300, extraction comprise the query statement of knowledge fragment in user's historical behavior storehouse; As in the query statement of " horse six how ", " Skoda how ", " Sagitar how ", extract the knowledge fragment: " horse six ", " Skoda ", " Sagitar ";

S301, described knowledge base fragment is replaced to general symbol(s), generate candidate's expression template; For example: " [automobile brand/model] " symbol, i.e. candidate's expression template of generation " [automobile brand/model] how ";

The quantity of the knowledge base fragment that candidate's expression template that S302, statistics generate meets;

S303, judge that described quantity is whether greater than the threshold value of setting;

If S304 is greater than the threshold value of setting, then with described candidate's expression template as expression template, and be stored in the database, generate the expression template storehouse;

By above-mentioned searching method and search engine, the online interface of one embodiment of the present invention as shown in figure 14, be used at online interface that browser is opened search engine of the present invention, and input inquiry instruction in dialog box " mobile phone Nokia ", by above-mentioned searching method and search system, can judge the demand intention that this query statement has comprised the commodity classes, so in searching method of the present invention and search system, " mobile phone Nokia " this query statement can be searched in commodity vertical search database, simultaneously, insert among the result that this vertical search result searches for to the web page repository, A part such as figure, when the user clicks described vertical search result, can jump in the new page, as shown in figure 15, comprised the result for retrieval with commodity class demand intention in this new page, the B part can be found out from figure, does not comprise " mobile phone " this text message in this result for retrieval, namely is the Search Results that obtains by semantic extension of the present invention.

In sum as can be known, the present invention is on the basis of knowledge base, better understand the query statement of user's input, the intention of analysis and consult instruction, resolve the structure of query statement, query statement is carried out semantic content expand, thereby better the guidance search engine selects the resource of high-quality to satisfy user's search need, so that user search efficient improves, save network traffics.

Be to be understood that, although this instructions is described according to embodiment, but be not that each embodiment only comprises an independently technical scheme, this narrating mode of instructions only is for clarity sake, those skilled in the art should make instructions as a whole, technical scheme among each embodiment also can through appropriate combination, form other embodiments that it will be appreciated by those skilled in the art that.

Above listed a series of detailed description only is specifying for feasibility embodiment of the present invention; they are not to limit protection scope of the present invention, allly do not break away from equivalent embodiment or the change that skill spirit of the present invention does and all should be included within protection scope of the present invention.

Claims

1. a searching method is characterized in that, described searching method may further comprise the steps:

S1, reception query statement;

S4, export described Search Results;

Described " based on knowledge base described query statement is carried out the demand intention and analyze, the demand intention of clear and definite described query statement " specifically comprises following flow process:

2. searching method according to claim 1 is characterized in that, described database is web page repository or the vertical search database corresponding with described demand intention.

3. a searching method is characterized in that, described searching method may further comprise the steps:

S1, reception query statement;

S4, export described Search Results;

4. searching method according to claim 3 is characterized in that, described database is web page repository or the vertical search database corresponding with described demand intention.

5. a searching method is characterized in that, described searching method may further comprise the steps:

S1, reception query statement;

S4, export described Search Results;

Described " based on knowledge base and expression template storehouse described query statement being carried out the analysis of demand intention, the demand intention of clear and definite described query statement " specifically comprises following flow process:

6. searching method according to claim 5 is characterized in that, described database is web page repository or the vertical search database corresponding with described demand intention.

7. searching method according to claim 5 is characterized in that, the construction method in described expression template storehouse comprises following flow process:

S303, judge that whether described quantity is greater than setting threshold;

8. a searching method is characterized in that, described searching method may further comprise the steps:

S1, reception query statement;

S4, export described Search Results;

9. searching method according to claim 8 is characterized in that, described database is web page repository or the vertical search database corresponding with described demand intention.

10. searching method according to claim 8 is characterized in that, the construction method in described expression template storehouse comprises following flow process:

S303, judge that whether described quantity is greater than setting threshold;

11. a search engine is characterized in that, described search engine comprises:

Knowledge base is used for the storage priori;

Described demand intention analysis module " carries out the demand intention based on knowledge base to described query statement and analyzes, the demand intention of clear and definite described query statement " specifically comprising following flow process:

12. search engine according to claim 11 is characterized in that, described search engine also comprises:

13. search engine according to claim 11 is characterized in that, described search engine also comprises:

14. search engine according to claim 13 is characterized in that, described user's historical search record comprises: query statement, inquiry times, and weighting clicks.

15. according to claim 13 or 14 described search engines, it is characterized in that described search engine also comprises:

16. search engine according to claim 11 is characterized in that, described search engine also comprises:

17. search engine according to claim 11 is characterized in that, described database is web page repository or the vertical search database corresponding with described demand intention.

18. search engine according to claim 17 is characterized in that,

Described web page repository is used for the index information of storage network page data and this web data;