CN102915312A

CN102915312A - Method and system for issuing information on websites

Info

Publication number: CN102915312A
Application number: CN2011102213865A
Authority: CN
Inventors: 张祝玉; 黄鹏; 林锋; 冯炯
Original assignee: Alibaba Group Holding Ltd
Current assignee: Alibaba Group Holding Ltd
Priority date: 2011-08-03
Filing date: 2011-08-03
Publication date: 2013-02-06
Anticipated expiration: 2031-08-03
Also published as: HK1176431A1; CN102915312B

Abstract

The invention provides a method and a system for issuing information on websites. The method includes: a server at an information issuing end receives thematic information, input by a user, of information to be issued through a client end; the server inquires inquiring entries related to the thematic information in a storage storing inquiring entry history; the server transmits the inquired inquiring entries as keywords of information to be issued to the client end; and the server receives issued information obtained by choosing the keywords through the client end. The technical problem that how to further improve recall rate of retrieval under the circumference of not occupying extra storage space of website database is solved, and recall rate of issued information is increased without occupying extra storage space of website database.

Description

Information issuing method in the website and system

Technical field

The application relates to Internet technical field, information issuing method and system in a kind of website.

Background technology

The basic process of information retrieval is: 1) user conceives the query word of its search intention of expression and submits search engine to; 2) search engine retrieving goes out the webpage that mates with this keyword; 3) search engine sorts according to certain set strategy to the webpage that retrieves information or the relation between the webpage according to webpage itself.

Vertical search engine, such as E-commerce Search Engine, be with one of difference of universal search engine: E-commerce Search Engine is more valued the accuracy of matching result, requires the input of complete match user.Usually input the product of particular community, model or brand when for example the user inquires about, the common way of E-commerce Search Engine is that the various piece (except the standardization processing) that guarantees query word all will be retrieved in Query Result.This way has guaranteed the accuracy of result for retrieval to a great extent, but the recall rate (ratio of all relevant documentation numbers in the relevant documentation number that retrieves and the document library, what weigh is the recall ratio of searching system) have accordingly certain loss, reason is that user's the product keywords such as not comprising some specific attribute, model or brand in (offer) that release news are described, thereby finally causes the inquiry Experience Degree of inquiring user to descend.

One of method that improves the retrieval recall rate can be when the user of website releases news, requiring the user to fill in more perfect information describes, complete as the keywords such as particular community, model or brand of commodity are filled in one by one, and be uploaded to Website server, be stored in the site databases.In the retrieving of information, just can there be the searching keyword of more Query Result and inquiring user input to be complementary like this.Yet the technical matters of the maximum that this way runs into is, because increasing of the information content of filling in when the user releases news, cause the data volume of user in releasing news to increase, for a superhuge website, its database storage capacity will be challenged, and causes this website must increase more database server and stores the data message of additionally filling in when the user releases news.

For the problem of above-mentioned existence in the correlation technique, not yet propose in the situation of the storage space that does not additionally take site databases at present, how further to improve the technical scheme of the recall rate of retrieval.

Summary of the invention

The application's fundamental purpose is to provide information issuing method and the system in a kind of website, to solve at least in the prior art in the situation of the storage space that does not additionally take site databases, how further to improve the technical matters of the recall rate of retrieval.

An aspect according to the application provides the information issuing method in a kind of website, comprising: the server of information publishing side is by the subject information of the information to be released of client user input; Server is to the memory query query entries relevant with subject information, and wherein, memory stores has the query entries of historical record; Server sends to client with the query entries that inquires as the keyword of information to be released; Server by client to releasing news that keyword is selected to obtain.

Further, subject information comprises: the title of information to be released and classification.

Further, server by following query steps to the memory query query entries relevant with subject information: server is divided into M independently keyword with title, and chooses N keyword from M keyword, wherein, M and N are natural number, and M 〉=N; Server inquires about whether there is the query entries that comprises N keyword from storer; If exist, then server judges that whether the number of the query entries that belongs to described classification in the query entries that inquires is more than or equal to P, if more than or equal to, then will belong to front P query entries that inquires of classification as the query entries relevant with subject information, wherein, P is predefined natural number.

Further, if server judges that the number of the query entries that belongs to described classification in the query entries that inquires is less than P, then server also comprises to the step of the memory query query entries relevant with subject information: repeat following steps, until the number of query entries that belongs to described classification in the query entries that inquires is more than or equal to P: server makes N=N-1, and carries out query steps in storer.

Further, server by following query steps to the memory query query entries relevant with subject information: server is selected the query entries that belongs to classification from storer; Server is divided into M independently keyword with title, and chooses N keyword from M keyword, and wherein, M and N are natural number, and M 〉=N; Server searches whether there is the query entries that comprises N keyword more than Q in the query entries that belongs to classification that chooses, wherein, Q is predefined natural number; If exist, then will belong to front Q query entries that inquires of classification as the query entries relevant with subject information.

Further, if server is judged the number of the query entries that inquires that belongs to classification less than Q, then server also comprises to the step of the memory query query entries relevant with subject information: repeat following steps, until the number of the query entries that inquires that belongs to classification is more than or equal to Q: server makes N=N-1, and carries out query steps in storer.

Further, server comprises the query entries that inquires as the step that keyword sends to client: whether the on-line checking result's of each query entries in the server basis query entries that inquires number is greater than predetermined threshold; Server is recorded as the first group polling clauses and subclauses with on-line checking result's number greater than the query entries of predetermined threshold, and on-line checking result's number is recorded as the second group polling clauses and subclauses less than or equal to the query entries of predetermined threshold; Server sends to client with the first group polling clauses and subclauses and the second group polling clauses and subclauses as keyword.

Further, server comprises on-line checking result's number greater than the step that the query entries of predetermined threshold is recorded as the first group polling clauses and subclauses: calculate on-line checking result's number greater than each query entries in the query entries of predetermined threshold and the degree of correlation of title; According to degree of correlation order from big to small record queries result's in the first group polling clauses and subclauses the number query entries greater than predetermined threshold.Server comprises on-line checking result's number less than or equal to the step that the query entries of predetermined threshold is recorded as the second group polling clauses and subclauses: calculate on-line checking result's number less than or equal to each query entries in the query entries of predetermined threshold and the degree of correlation of title; According to degree of correlation order from big to small record queries result's in the second group polling clauses and subclauses the number query entries less than or equal to predetermined threshold.

Further, before the memory query query entries relevant with subject information, said method also comprises at server: server upgrades the query entries of the historical record stored in the storer.

According to the application on the other hand, information issuing system in a kind of website is provided, comprise: the server and client side of information publishing side, wherein, client is used for sending to server the subject information of the information to be released of user's input, wherein, subject information comprises title and the classification of information to be released; The server of information publishing side is used for receiving the subject information that client sends; To the memory query query entries relevant with subject information, wherein, memory stores has the query entries of historical record; The query entries that inquires is sent to client as the keyword of information to be released, and by client to releasing news that keyword is selected to obtain.

Further, server comprises: the first title processing unit is used for to the memory query query entries relevant with subject information the time, title is divided into M independently keyword, and from M keyword, chooses N keyword, wherein, M and N are natural number, and M 〉=N; The first query unit is used for whether having the query entries that comprises N keyword from the storer inquiry; The first judging unit, be used for when existence comprises the query entries of N keyword, judge that whether the number of the query entries that belongs to described classification in the query entries that inquires is more than or equal to P, if more than or equal to, then will belong to front P query entries that inquires of classification as the query entries relevant with subject information, wherein, P is predefined natural number.

Further, the number of query entries that server also is used for belonging to described classification in the query entries that the first judgment unit judges goes out to inquire is during less than P, repeat following steps, until the number of query entries that belongs to described classification in the query entries that inquires is more than or equal to P: server makes N=N-1; Notify the first title processing unit from M keyword, to choose N keyword; Notify the first query unit from storer, to inquire about and whether have the query entries that comprises N keyword; And whether the number of notifying the first judging unit to judge the query entries that belongs to described classification in the query entries that inquires when existence comprises the query entries of N keyword is individual more than or equal to P, if greater than, then will belong to front P query entries that inquires of classification as the query entries relevant with subject information.

Further, server comprises: selected cell is used for selecting the query entries that belongs to classification from storer; The second title processing unit is used for title is divided into M independently keyword, and chooses N keyword from M keyword, and wherein, M and N are natural number, and M 〉=N; The second query unit is used for searching whether there is the query entries that comprises N keyword more than Q in the query entries that belongs to classification that chooses, and wherein, Q is predefined natural number; If exist, then will belong to front Q query entries that inquires of classification as the query entries relevant with subject information.

Further, the number that server also is used for the query entries that inquires that belongs to classification that finds out in the second query unit is during less than Q, repeat following steps, until the number of the query entries that inquires that belongs to classification is more than or equal to Q: server makes N=N-1; Notify the second title processing unit from M keyword, to choose N keyword; And notify the second query unit in the query entries that belongs to classification that chooses, to search whether to exist the query entries that comprises N keyword more than Q, if exist, then will belong to front Q query entries that inquires of classification as the query entries relevant with subject information.

Further, server comprises: the second judging unit, be used for when server sends to client with the query entries that inquires as the keyword of information to be released, judge that whether on-line checking result's the number of each query entries in the query entries that inquires is greater than predetermined threshold; Record cell is used for on-line checking result's number is recorded as the first group polling clauses and subclauses greater than the query entries of predetermined threshold, and on-line checking result's number is recorded as the second group polling clauses and subclauses less than or equal to the query entries of predetermined threshold; Transmitting element is used for the first group polling clauses and subclauses and the second group polling clauses and subclauses are sent to client as keyword.

Further, record cell comprises: the first record cell is used for number by following steps record queries result greater than the query entries of predetermined threshold: calculate on-line checking result's number greater than each query entries of the query entries of predetermined threshold and the degree of correlation of title; According to degree of correlation order from big to small record queries result's in the first group polling clauses and subclauses the number query entries greater than predetermined threshold; The second record cell is used for number by following steps record queries result less than or equal to the query entries of predetermined threshold: calculate on-line checking result's number less than or equal to each query entries of the query entries of predetermined threshold and the degree of correlation of title; According to degree of correlation order from big to small record queries result's in the second group polling clauses and subclauses the number query entries less than or equal to predetermined threshold.

Further, described server comprises: updating block, be used at described server before the memory query query entries relevant with subject information, and the query entries of the historical record stored in the storer is upgraded.

The application has realized following technique effect by above technical scheme:

1) server is by sending to described client with the query entries of historical record as the keyword of described information to be released, effectively the tendentiousness of buyer user's inquiry is recommended seller user by client, so that seller user need not be filled in a large amount of information content is described, additionally do not taking in the situation in site databases server stores space, can improve the recall rate of user's release product information, finally reach the purpose that reduces by zero/few result queries word quantity, preferably, can improve the Experience Degree of buyer on e-commerce website and further promote buyer's enthusiasm of concluding the business;

2) server is selected the query entries relevant with the subject information of seller user input in historical record, and this query entries is from each buyer's difference input, avoided generating the single problem of keyword, even so that when buyer user inputs different query words, can return according to the product information of seller's issue this seller's Query Result;

3) server dynamically updates by the query entries to historical record, can solve keyword limited amount and the serious problem of homogeneity of generation, can in real time seller user be recommended in the tendentious keyword that has reacted buyer user's inquiry.

Certainly, arbitrary product of enforcement the application must not necessarily need to reach simultaneously above-described all advantages.

Description of drawings

Accompanying drawing described herein is used to provide the further understanding to the application, consists of the application's a part, and the application's illustrative examples and explanation thereof are used for explaining the application, do not consist of the improper restriction to the application.In the accompanying drawings:

Fig. 1 is a kind of preferred structure block diagram according to the information issuing system in the website of the embodiment of the present application;

Fig. 2 is a kind of preferred structure block diagram according to the server in the information issuing system in the website of the embodiment of the present application;

Fig. 3 is the another kind of preferred structure block diagram according to the server in the information issuing system in the website of the embodiment of the present application;

Fig. 4 is the another kind of structured flowchart according to the information issuing system in the website of the embodiment of the present application;

Fig. 5 is a kind of preferred flow charts according to the information issuing method in the website of the embodiment of the present application;

Fig. 6 is the another kind of preferred flow charts according to the information issuing method in the website of the embodiment of the present application;

Fig. 7 is another preferred flow charts according to the information issuing method in the website of the embodiment of the present application.

Embodiment

Hereinafter also describe in conjunction with the embodiments the application in detail with reference to accompanying drawing.The embodiment of the present application will with the product information in e-commerce website issue be retrieved as example and describe, those skilled in the art can be applied to the present techniques scheme in the websites such as video website, network forum under not needing to make the condition of creative work certainly.Need to prove that in the situation of not conflicting, embodiment and the feature among the embodiment among the application can make up mutually.

At first, the related some terms of the application are made an explanation:

1) inquiry: the i.e. query word in search procedure, inputted of user.

2) product: in e-commerce field, the commodity that businessman sells.

3) classification (classification): namely in e-commerce field, the classification (classification) under product.

4) keyword: correctly describe some words of merchandise news, be used for indexing of retrieval end merchandise news.

5) blue extra large word: e-commerce field, user search are often but the few query word of result for retrieval.

6) popular word: e-commerce field, user search are often and the many query words of result for retrieval.

7) recall rate: the ratio of all relevant documentation numbers in the relevant documentation number that retrieves and the document library, measurement be the recall ratio of searching system.

8) querylog: the user is at the inquiry log of e-commerce website.

9) product exposure rate: commodity are demonstrated the ratio to search subscriber in the historical query of e-commerce website.

Before the further details of each embodiment that describes the application, a suitable counting system structure of the principle that can be used for realizing the application is described with reference to Fig. 1.In the following description, except as otherwise noted, otherwise each embodiment of the application is described with reference to the symbolic representation of the action of being carried out by one or more computing machines and operation.Thus, be appreciated that this class action and the operation that sometimes are called as the computing machine execution comprise that the processing unit of computing machine is to representing the manipulation of the electric signal of data with structured form.This manipulation transforms data or the position in the accumulator system of computing machine safeguard it, the operation of computing machine is reshuffled or changed to this mode of all understanding with those skilled in the art.The data structure of service data is the physical location of storer with defined particular community of form of data.Yet although describe the application in above-mentioned context, it does not also mean that restrictively, and as understood by those skilled in the art, the each side of hereinafter described action and operation also available hardware realizes.

Turn to accompanying drawing, wherein identical reference number refers to identical element, and the application's principle is shown in the suitable computing environment and realizes.Below describe the embodiment based on described the application, and should not think to limit the application about the alternative embodiment clearly do not described herein.

Fig. 1 shows the synoptic diagram of an example computer architecture that can be used for these equipment.For purposes of illustration, the architecture of painting only is an example of proper environment, is not that usable range or function to the application proposes any limitation.This computing system should be interpreted as that arbitrary assembly shown in Figure 1 or its combination are had any dependence or demand yet.

The application's principle can or dispose with other universal or special calculating or communication environment and operate.The example that is applicable to the application's well-known computing system, environment and configuration includes but not limited to, personal computer, server, multicomputer system, the system based on little processing, minicomputer, mainframe computer and the distributed computing environment that comprises arbitrary said system or equipment.

In its most basic configuration, Fig. 1 has shown the information issuing system in a kind of website, and it comprises: the server 102 of information publishing side and one or more client 104.Server 102 can include but not limited to the treating apparatus of Micro-processor MCV or programmable logic device (PLD) FPGA etc., the transmitting device that is used for storage data storage device and communicates by letter with client 104; Client 104 can comprise: Micro-processor MCV, with the transmitting device of server communication, with the display device of user interactions.In the present specification and claims, " information issuing system in the website " also can be defined as can executive software, firmware or microcode come any nextport hardware component NextPort of practical function or the combination of nextport hardware component NextPort.Information issuing system in the website even can be distributed is to realize distributed function.

Embodiment 1

As shown in Figure 1, the information issuing system in the website comprises: the server 102 of interconnective information publishing side and client 104.

In the course of the work, client 104 is to the subject information of the information to be released of server 102 transmission user inputs, and in the application's preferred embodiment, subject information includes but not limited to title and the classification of information to be released; The server 102 of information publishing side is after the subject information that receives client 104 transmissions, and to the memory query query entries relevant with subject information, wherein, memory stores has the query entries of historical record; Server 102 sends to client 104 with the query entries that inquires as the keyword of information to be released, and receives releasing news that keyword is selected to obtain by client 104.Above-mentioned query entries is the Query Information that the buyer user of historical record adopts, the information such as the search custom that the user who has represented search information adopts for this information to be released and search focus.

Information to be released in the present embodiment can be the product information to be released in the e-commerce website, perhaps, and the video information to be released in the video website etc.

In above-mentioned preferred embodiment, server is by sending to client with the query entries of historical record as the keyword of information to be released, effectively will search for the tendentiousness of user's inquiry of information and recommend the user who releases news by client, thereby can improve the recall rate of the information of user's issue, finally reach the purpose that reduces by zero/few result queries word quantity.In addition, because the application effectively will search for tendentiousness information exchange that the user of information inquires about and cross client and recommend the user who releases news, so that the user who releases news need not fill in a large amount of information when releasing news content is described, the database server of storage information to be released also need not be stored a large amount of information to be released like this, namely, the application is not additionally taking in the situation in site databases server stores space, has improved the recall rate that the user that releases news releases news.Preferably, as the user who releases news during for the seller of e-commerce website, can improve the Experience Degree of buyer on e-commerce website and further promote buyer's enthusiasm of concluding the business.

In each embodiment of the application, above-mentioned storer can be arranged on the server of information publishing side, also can be arranged on other the server, and to this, the application is not construed as limiting.

In order to make server in storer, obtain the query entries relevant with subject information, the application provides two kinds of different modes, the seller to be released product information of the below on take information to be released as e-commerce website is example, is described in detail by reference to the accompanying drawings the process of obtaining the query entries relevant with subject information in storer.

(1) judges first the title of product to be released, judge again the classification of product to be released

Under this inquiry mode, server shown in Figure 1 can comprise the concrete structure among Fig. 2.As shown in Figure 2, server 202 comprises the first title processing unit 2021 that connects successively, the first query unit 2022 and the first judging unit 2023, wherein, the first title processing unit 2021 is to storer (here, storer is used for storing the query entries of historical record, it can be positioned on the server, also can be positioned on other background devices, perhaps be memory device independently) during the inquiry query entries relevant with subject information, title is divided into M independently keyword, and from M keyword, chooses N keyword, wherein, M and N are natural number, and M 〉=N; The first query unit 2022 inquires about whether there is the query entries that comprises N keyword from storer; The first judging unit 2023 is when existence comprises the query entries of N keyword, judge that whether the number of the query entries that belongs to described classification in the query entries that inquires is more than or equal to P, if greater than, then will belong to the query entries that inquires of classification as the query entries relevant with subject information, wherein, P is predefined natural number.Preferably, also can be only with the partial query clauses and subclauses as the query entries relevant with subject information, for example, if belong to the number of the query entries that inquires of classification more than or equal to P, then will belong to front P query entries that inquires of described classification as the query entries relevant with described subject information.

Wherein, the number of judging the query entries that belongs to described classification in the query entries that inquires at the first judging unit 2023 is during less than P, repeat following steps, until the number of query entries that belongs to described classification in the query entries that inquires is more than or equal to P: server 202 makes N=N-1, and notifies the first query unit 2022 to carry out query steps in storer.In this preferred embodiment, by dynamic adjustment query argument, thereby can fast, accurately obtain needed Query Result.

The below further describes above-mentioned deterministic process by way of example.Suppose that the title in the subject information is divided into 2 independent keywords: " apple " and " mobile phone ", and the classification in the subject information is " 3G network ", P is 30, when in storer, inquiring about the clauses and subclauses relevant with subject information like this, at first inquiry comprises the query entries of above-mentioned two keywords " apple " and " mobile phone " simultaneously, if find 100, judge in these 100 query entries that then which query entries belongs to " 3G network " classification, surpass 30 if judge the query entries that belongs to " 3G network " classification in 100 query entries, then with front 30 query entries that inquire as the query entries relevant with described subject information.

(2) judge first the classification of product to be released, judge again the title of product to be released

Under this inquiry mode, server shown in Figure 1 can comprise the concrete structure among Fig. 3.As shown in Figure 3, server 302 comprises selected cell 3021, the second title processing unit 3022 and the second query unit 3023 that connects successively, and wherein, selected cell 3021 is selected the query entries that belongs to classification from storer; 3022 pairs of titles of the second title processing unit are divided into M independently keyword, and choose N keyword from M keyword, and wherein, M and N are natural number, and M 〉=N; The second query unit 3023 searches whether there is the query entries that comprises N keyword more than Q in the query entries that belongs to classification that chooses, wherein, Q is predefined natural number; If exist, then will belong to the query entries that inquires of classification as the query entries relevant with subject information.Preferably, also can be only with the partial query clauses and subclauses as the query entries relevant with subject information, for example, if in the query entries that belongs to classification that chooses, there is the query entries that comprises N keyword more than Q, then will belong to front Q query entries that inquires of described classification as the query entries relevant with described subject information.

Wherein, the number of the query entries that inquires that belongs to classification that finds out in the second query unit 3023 is during less than Q, repeat following steps, until the number of the query entries that inquires that belongs to classification is more than or equal to Q: server 302 makes N=N-1, and notifies the second query unit 3023 to carry out query steps in storer.In this preferred embodiment, by dynamic adjustment query argument, thereby can fast, accurately obtain needed Query Result.

The below further describes above-mentioned deterministic process by way of example.Suppose that the title in the subject information is divided into 2 independent keywords: " apple " and " mobile phone ", and the classification in the subject information is " 3G network ", Q is 30, when in storer, inquiring about the clauses and subclauses relevant with subject information like this, at first inquiry belongs to the query entries of " 3G network " classification, if find 100, judge in these 100 query entries that then which query entries comprises above-mentioned two keywords " apple " and " mobile phone " simultaneously, surpass 30 if judge the query entries that comprises simultaneously above-mentioned two keywords " apple " and " mobile phone " in above-mentioned 100 query entries, then with front 30 query entries that inquire as the query entries relevant with described subject information.

For above-mentioned Fig. 2 and server shown in Figure 3, it selects the query entries relevant with the subject information of seller user input in historical record, and this query entries is from each buyer's difference input, avoided generating the Single-issue of keyword, so that the product information of seller's issue can satisfy the diversified characteristics of buyer user's query word.

On the basis of above-mentioned each embodiment, for the query entries that will inquire sends to client as the keyword of product to be released, server can also comprise the concrete structure among Fig. 4.As shown in Figure 4, server 402 comprises the second judging unit 4021, record cell 4022 and the transmitting element 4023 that connects successively, wherein, when the second judging unit 4021 sends to client with the query entries that inquires as the keyword of product to be released at server, judge that whether on-line checking result's the number of each query entries in the query entries that inquires is greater than predetermined threshold; Record cell 4022 is recorded as the first group polling clauses and subclauses with on-line checking result's number greater than the query entries of predetermined threshold, and on-line checking result's number is recorded as the second group polling clauses and subclauses less than or equal to the query entries of predetermined threshold; Transmitting element 4023 sends to client 404 with the first group polling clauses and subclauses and the second group polling clauses and subclauses as keyword.

For example, when above-mentioned predetermined threshold is 100, on-line checking result's number can be recorded as the second group polling clauses and subclauses less than or equal to 100 query entries, and these group polling clauses and subclauses can be considered as be blue extra large word (being worth relatively high), preferably, when sending to client, can preferentially send these group polling clauses and subclauses, like this, at first show that to seller user these are worth the tendentiousness that higher blue extra large word can more effectively reflect present user's inquiry by client; In addition, on-line checking result's number can be recorded as the first group polling clauses and subclauses greater than 100 query entries, it is popular word (being worth relatively low) that a part in these the first group polling clauses and subclauses can be considered as, preferably, after will sending to client as the second group polling clauses and subclauses of the extra large word of indigo plant, send again the first group polling clauses and subclauses as popular word.That is to say that server will reflect that with the order of popular word behind the first blue extra large word the tendentious keyword of buyer user's inquiry recommended seller user by client.By above-mentioned demonstration and record scheme, server can recommend to have reflected that the buyer inquires about tendentious keyword to seller user according to the height that is worth, thereby improve the efficient that seller user is selected.

Wherein, record cell 4022 comprises the first record cell 40221 and the second record cell 40222, wherein, the number of the first record cell 40221 by following steps record queries result is greater than the query entries of predetermined threshold: calculate on-line checking result's number greater than each query entries in the query entries of predetermined threshold and the degree of correlation of title; According to degree of correlation order from big to small record queries result's in the first group polling clauses and subclauses the number query entries greater than predetermined threshold; The number of the second record cell 40222 by following steps record queries result is less than or equal to the query entries of predetermined threshold: calculate on-line checking result's number less than or equal to each query entries in the query entries of predetermined threshold and the degree of correlation of title; According to degree of correlation order from big to small record queries result's in the second group polling clauses and subclauses the number query entries less than or equal to predetermined threshold.

On the basis of above-mentioned each embodiment, in order to realize dynamically updating the query entries in the historical record, server can also comprise the updating block 406 shown in Fig. 4, this updating block 406 links to each other with storer 405, be used at server before the storer 405 inquiries query entries relevant with subject information, the query entries of the historical record stored in the storer is upgraded.In this preferred embodiment, server dynamically updates by the query entries to historical record, can solve keyword limited amount and the serious problem of homogeneity of generation, can in real time seller user be recommended in the tendentious keyword that has reacted buyer user's inquiry.

Embodiment 2

On the basis of the information issuing system in the website of Fig. 1-shown in Figure 4, the application also provides the information issuing method in a kind of website, and as shown in Figure 5, the information issuing method in this website may further comprise the steps:

S502, the server of information publishing side is by the subject information of the information to be released of client user input;

S504, server are to the memory query query entries relevant with subject information, and wherein, memory stores has the query entries of historical record;

S506, server sends to client with the query entries that inquires as the keyword of information to be released;

S508, server by client to releasing news that keyword is selected to obtain.

Preferably, subject information comprises: the title of information to be released and classification.

Server can be by following query steps to the memory query query entries relevant with subject information: server is divided into M independently keyword with title, and chooses N keyword from M keyword, and wherein, M and N are natural number, and M 〉=N; Server inquires about whether there is the query entries that comprises N keyword from storer; If exist, then server judges that whether the number of the query entries that belongs to described classification in the query entries that inquires is more than or equal to P, if more than or equal to, then will belong to the query entries that inquires of classification as the query entries relevant with subject information, wherein, P is predefined natural number.Preferably, also can be only with the partial query clauses and subclauses as the query entries relevant with subject information, for example, if belong to the number of the query entries that inquires of classification more than or equal to P, then will belong to front P query entries that inquires of described classification as the query entries relevant with described subject information.

Wherein, if server is judged the number of the query entries that belongs to described classification in the query entries that inquires less than P, then server also comprises to the step of the memory query query entries relevant with subject information: repeat following steps, until the number of query entries that belongs to described classification in the query entries that inquires is more than or equal to P: server makes N=N-1, and carries out query steps in storer.In this preferred embodiment, by dynamic adjustment query argument, thereby can fast, accurately obtain needed Query Result.

Server can also be by following query steps to the memory query query entries relevant with subject information: server is selected the query entries that belongs to classification from storer; Server is divided into M independently keyword with title, and chooses N keyword from M keyword, and wherein, M and N are natural number, and M 〉=N; Server searches whether there is the query entries that comprises N keyword more than Q in the query entries that belongs to classification that chooses, wherein, Q is predefined natural number; If exist, then will belong to the query entries that inquires of classification as the query entries relevant with subject information.Preferably, also can be only with the partial query clauses and subclauses as the query entries relevant with subject information, for example, if in the query entries that belongs to classification that chooses, there is the query entries that comprises N keyword more than Q, then will belong to front Q query entries that inquires of described classification as the query entries relevant with described subject information.

Wherein, if server is judged the number of the query entries that inquires that belongs to classification less than Q, then server also comprises to the step of the memory query query entries relevant with subject information: repeat following steps, until the number of the query entries that inquires that belongs to classification is more than or equal to Q: server makes N=N-1, and carries out query steps in storer.In this preferred embodiment, by dynamic adjustment query argument, thereby can fast, accurately obtain needed Query Result.

For above-mentioned two kinds of inquiry modes, server is all selected the query entries relevant with the subject information of seller user input in historical record, and this query entries is from each buyer's difference input, avoided generating the Single-issue of keyword, so that the product information of seller's issue can satisfy the diversified characteristics of buyer user's query word.

On the basis of above-mentioned each embodiment, for the query entries that will inquire sends to client as the keyword of product to be released, server sends to client with the query entries that inquires as the keyword of product to be released by following steps:

S1: whether the on-line checking result's of each query entries in the query entries that the server judgement inquires number is greater than predetermined threshold;

S2: server is recorded as the first group polling clauses and subclauses with on-line checking result's number greater than the query entries of predetermined threshold, and on-line checking result's number is recorded as the second group polling clauses and subclauses less than or equal to the query entries of predetermined threshold;

Further, for the record between the query entries that belongs to blue extra large word and displaying scheme, then can carry out according to the coupling degree of correlation of this query entries and title, specifically it may further comprise the steps: server calculates on-line checking result's number less than or equal to each query entries in the query entries of predetermined threshold and the degree of correlation of title; According to degree of correlation order from big to small record queries result's in the second group polling clauses and subclauses the number query entries less than or equal to predetermined threshold.

Further, for the record between the query entries that belongs to popular word and displaying scheme, then can carry out according to the coupling degree of correlation of this query entries and title, specifically it may further comprise the steps: server calculates on-line checking result's number greater than each query entries in the query entries of predetermined threshold and the degree of correlation of title; According to degree of correlation order from big to small record queries result's in the first group polling clauses and subclauses the number query entries greater than predetermined threshold.

S3: server sends to client with the first group polling clauses and subclauses and the second group polling clauses and subclauses as keyword.

By above-mentioned demonstration and record scheme, can come to recommend to have reflected that the buyer inquires about tendentious keyword to seller user according to the height that is worth, thereby improve the efficient that seller user is selected.

On the basis of above-mentioned each embodiment, in order to realize dynamically updating the query entries in the historical record, server can also dynamically update the query entries of historical record, detailed process comprises: before the memory query query entries relevant with subject information, server upgrades the query entries of the historical record stored in the storer at server.In this preferred embodiment, server dynamically updates by the query entries to historical record, can solve keyword limited amount and the serious problem of homogeneity of generation, can in real time seller user be recommended in the tendentious keyword that has reacted buyer user's inquiry.

Describe concrete example in detail below in conjunction with Product Information Publishing System and method in accompanying drawing and the above-mentioned ecommerce.

As shown in Figure 6, when release product information, seller user is selected classification at the server of product publishing side, is filled in title and keyword and other information, then is stored in the data warehouse (also to can be understood as and be kept in the database); Then, can be added index by Build (building) index machine.Therefore, when inputting this keyword in search engine, buyer user can retrieve corresponding product.From above description as can be known, the keyword filled in the product publishing side of seller user is the key factor that can its product indexed.But reality is the seller does not understand the information such as buyer's search custom and search focus usually, therefore often can not exact matching when filling in the keyword of product to user's search query word.

To this, the application provides a kind of product information dissemination method as shown in Figure 7, it is by machine learning and data mining technology, when seller user release product information, will reflect the buyer search for the custom and the search focus keyword recommend seller user, so that the product information that provides seller user to issue can be corresponding with present buyer's search custom and search focus, thereby improve the exposure rate of this product on the electronics website, the corresponding ratio that also can reduce on the whole zero few result queries word is improved inquiring user in the search experience of e-commerce website.

Referring to Fig. 7, system is divided into back-end data and excavates module 702 and the automatic recommending module 704 of foreground keyword.It mainly is to set up incidence relation between the query word (query) by product (offer) now in each class that back-end data is excavated module 702, for example, can set up incidence relation between the query word (query) according to the click volume of product (offer) and exposure rate, wherein, two query are relevant with same offer, think that then these two query have certain incidence relation.Then by the degree of association between iterative computation candidate query and the query, the synonymy of query and query be can excavate, and combination synonym and complete synonym further from the query synonym, excavated.In addition, system also comprises backstage inquiry log (querylog) processing module, and it mainly is that query is carried out data cleansing (comprising that standardized writing, stop word are filtered, invalid word filters, spell the error correction filtration, length keywords is filtered), cat_compute (classification calculating), update_data (renewal of data day), merge_data (data merging), buildindex (building inverted index).

The treatment scheme of the automatic recommending module 704 of foreground keyword comprises:

S1: product heading message (title) and its affiliated classification to this module input carry out the central information extraction.Concrete, at first title is carried out a series of information processing, comprising: the lexical item of title (token) is changed processing (title is divided into each independently English keywords), obtains each token and part of speech; The NP (Noun Phrase, noun phrase) of title is extracted in simple syntactic analysis.Then use the technology such as statistics and machine learning to extract the information such as the center NP of title (through expansion) and centre word.

S2: excavate module 702 output ground synonym information according to back-end data, the synonym in the title is scanned and locates.

S3: each keyword is given a mark.Concrete, extract each NP, and each keyword is given a mark.

S4-S5: lose word, combination and retrieval by the mark among the step S3, obtain the candidate and recommend set of words.Concrete, according to keyword score ordering among each NP, lose successively word according to mark, and the inquiry inverted index.Here, lose successively word according to mark and refer to: make N=N-1, that is, choose score in the keyword conduct candidate keywords corresponding with the title of the product of issuing of front N-1 position.The initial value of N can preset, for example, and 10～50.

S6: result for retrieval is filtered according to classification.Particularly, judge whether in candidate keywords corresponding to the title of above-mentioned product with issuing each belongs to the classification of the product of issuing.

S7: satisfy maximum and search number if satisfy the number of the candidate keywords of classification, then go to S8, otherwise go to S4.

S8-S9: synonym is replaced and retrieval, obtains the candidate and recommends set of words.Namely, synonym to the definite candidate keywords of above-mentioned steps S5 is retrieved and is recommended, concrete grammar can comprise: the synonym that this keyword is current is replaced, then the synonym after replacing is retrieved, and the synonym that retrieval obtains sorted, select the forward some synonyms of rank to recommend set of words as the candidate.Wherein, the sort method of employing can comprise:

1) for comprising in the title and near the complete synonym of core word, directly retrieving inverted index with synonym;

2) for comprising in the title and near core contamination synonym, with this synonym and other core combinations of words, then retrieving inverted index.

S10: result for retrieval is filtered according to classification.Particularly, judge whether in candidate keywords corresponding to the title of above-mentioned product with issuing each belongs to the classification of the product of issuing.

S11: satisfy maximum and search number if satisfy the number of the candidate keywords of classification, then go to S12, otherwise go to S8.

S12: divide blue extra large word and popular word.Because the keyword of recommending has two class purposes, a class is popular word, and a class is blue extra large word, and the standard that two classes are divided is whether number of results is greater than predetermined threshold (for example, 100).Wherein, blue extra large word＜popular word.

S13-S14: sequencing of similarity, and recommend keyword after the ordering to seller user.Wherein, can come in such a way keyword is sorted:

1) arranges first blue extra large word, the popular word of rear arrangement.This is because the value of blue extra large word is greater than popular word.

2) between the extra large word of indigo plant, perhaps, score and searching times according to keyword between popular word sort, and be concrete

I) at first, according to the ordering of score size;

Ii) then, if the score same difference, is then pressed the searching times ordering apart from 0.01 scope.

The computing method of the score of keyword are below described: calculate the coupling correlativity of title and keyword (query), and normalization mates correlativity, classification correlativity, competition degree, obtain the score (score) of this keyword, wherein,

1) coupling correlation calculations (match_relevance):

Regard respectively title (title) and query word (query) as two vectorial X, Y, all unduplicated word is as the one dimension of vector in title and the query word, if X=[x1, x2 ... xn], Y=[y1, y2 ... yn], x1～xn wherein, y1～yn represents the score (if a certain word does not occur, then this dimension mark is 0) of each word in two vectors in query word or title.

Need simultaneously to filter the query word that is comprised by title fully, because this query word does not have help for improving the retrieval recall rate.

2) classification correlativity (cate_relevance):

Call the classification computational tool under the line, query word is calculated it belong to the probability of some classification;

Selected classification is classification i when supposing seller's release product on the line, and then the classification correlativity of query word is the probability that this query word belongs to classification i.

3) normalization relevance scores (relevance)

Relevance=(match_relevance* text matches relevance weight+cate_relevance* classification coupling relevance weight)/(text matches relevance weight+classification coupling relevance weight).

4) competition degree mark (competition)

The competition degree need to be considered searching times (search_cnt) and Search Results number (result_num), but the Search Results number is usually larger, therefore the number of results (page_num) that need to can show according to one page number of results, convert result's number of pages to, and to the no longer difference greater than 20 pages of result's number of pages, namely the number of results maximal value is 20; Number of pages to the result multiply by certain penalty value (page_penalty), calculates result_rank.Competition degree mark is directly proportional with searching times, is inversely proportional to the number of pages of Search Results.

Computing formula is as follows:

result_rank＝(result_num/page_num)×page_penalty+1.0

competition＝log10(search_cnt/result_rank)/4.0+0.3

5) normalization PTS (score)

Score=(relevance* relevance scores weight+competition* competition degree fractional weight)/(relevance scores weight+competition degree fractional weight).

Product Information Publishing System and the method described by above embodiment are passable, user's inquiry log and click logs have embodied user's query intention to a great extent, can set up offer to a mapping model between the query word by related-art technology such as machine learning and information processings, recommend to provide technical support for the offer publishing side provides keyword.

The application has higher commercial value, and in the ecommerce search field, Search Results is that zero few query word proportion is larger at present, and the website that has had a strong impact on inquiring user is experienced.Cause Search Results zero few reason to mainly contain: it is accurate not that the user inputs the query word that reflects its search intention; The seller does not fill in abundant information in publishing commodity information, the information point paid close attention to of user particularly, attribute for example, model etc.; The seller does not issue the needed commodity of user.Research before mainly lays particular emphasis on the previous case, and major technique comprises inquiry rewriting, query expansion etc.And the application lays particular emphasis on the solution latter event, main thought is: recommend user's attention rate height and belong to zero/less keyword of Search Results in the commodity publishing side, and guide the seller to fill in the keyword of recommendation, thereby finally reach the purpose that improves the overall recall rate of query word.

In this preferred embodiment, dynamically excavate the recommendation word according to user's inquiry log, click relation, and the ordering of recommendation word has effectively embodied the tendentiousness of user input query word; Fill in title and the classification of offer by analysis of key word and seller, calculate the correlativity of keyword and offer, regularly upgrade simultaneously and recommend to gather, strengthened recommendation word accuracy and ageing; In addition, generate the keyword variation, promote the coverage rate of electronic goods keyword in user's query word.

Obviously, those skilled in the art should be understood that, each module of above-mentioned the application or each step can realize with general calculation element, they can concentrate on the single calculation element, perhaps be distributed on the network that a plurality of calculation elements form, alternatively, they can be realized with the executable program code of calculation element, thereby, they can be stored in the memory storage and be carried out by calculation element, and in some cases, can carry out step shown or that describe with the order that is different from herein, perhaps they are made into respectively each integrated circuit modules, perhaps a plurality of modules in them or step are made into the single integrated circuit module and realize.Like this, the application is not restricted to any specific hardware and software combination.

The above is the application's preferred embodiment only, is not limited to the application, and for a person skilled in the art, the application can have various modifications and variations.All within the application's spirit and principle, any modification of doing, be equal to replacement, improvement etc., all should be included within the application's the protection domain.

Claims

1. the information issuing method in the website is characterized in that, comprising:

The server of information publishing side is by the subject information of the information to be released of client user input;

Described server is to the memory query query entries relevant with described subject information, and wherein, described memory stores has the query entries of historical record;

Described server sends to described client with the query entries that inquires as the keyword of described information to be released;

Described server by described client to releasing news that described keyword is selected to obtain.

2. method according to claim 1 is characterized in that, described subject information comprises: the title of described information to be released and classification.

3. method according to claim 2 is characterized in that, described server by following query steps to the memory query query entries relevant with described subject information:

Described server is divided into M independently keyword with described title, and chooses N keyword from a described M keyword, and wherein, M and N are natural number, and M 〉=N;

Described server inquires about whether there is the query entries that comprises a described N keyword from described storer;

If exist, then described server judges that whether the number of the query entries that belongs to described classification in the query entries that inquires is more than or equal to P, if more than or equal to, then will belong to front P query entries that inquires of described classification as the query entries relevant with described subject information, wherein, P is predefined natural number.

4. method according to claim 3, it is characterized in that, if described server is judged the number of the query entries that belongs to described classification in the query entries that inquires less than P, then described server also comprises to the step of the memory query query entries relevant with described subject information:

Repeat following steps, until the number of query entries that belongs to described classification in the query entries that inquires is more than or equal to P: described server makes N=N-1, and carries out described query steps in described storer.

5. method according to claim 2 is characterized in that, described server by following query steps to the memory query query entries relevant with described subject information:

Described server is selected the query entries that belongs to described classification from described storer;

Described server searches whether there is the query entries that comprises a described N keyword more than Q in the query entries that belongs to described classification that chooses, wherein, Q is predefined natural number;

If exist, then will belong to front Q query entries that inquires of described classification as the query entries relevant with described subject information.

6. method according to claim 5, it is characterized in that, if described server is judged the number of the query entries that inquires that belongs to described classification less than Q, then described server also comprises to the step of the memory query query entries relevant with described subject information:

Repeat following steps, until the number of the query entries that inquires that belongs to described classification is more than or equal to Q: described server makes N=N-1, and carries out described query steps in described storer.

7. each described method in 6 according to claim 1 is characterized in that described server comprises the query entries that inquires as the step that described keyword sends to described client:

Described server judges that whether on-line checking result's the number of each query entries in the described query entries that inquires is greater than predetermined threshold;

Described server is recorded as the first group polling clauses and subclauses with on-line checking result's number greater than the query entries of described predetermined threshold, and on-line checking result's number is recorded as the second group polling clauses and subclauses less than or equal to the query entries of described predetermined threshold;

Described server sends to described client with described the first group polling clauses and subclauses and described the second group polling clauses and subclauses as described keyword.

8. method according to claim 7 is characterized in that,

Described server comprises on-line checking result's number greater than the step that the query entries of described predetermined threshold is recorded as the first group polling clauses and subclauses:

Calculate described on-line checking result's number greater than each query entries in the query entries of described predetermined threshold and the degree of correlation of described title;

In described the first group polling clauses and subclauses, record the number of described Query Result greater than the query entries of described predetermined threshold according to degree of correlation order from big to small;

Described server comprises on-line checking result's number less than or equal to the step that the query entries of described predetermined threshold is recorded as the second group polling clauses and subclauses:

Calculate described on-line checking result's number less than or equal to each query entries in the query entries of described predetermined threshold and the degree of correlation of described title;

In described the second group polling clauses and subclauses, record the number of described Query Result less than or equal to the query entries of described predetermined threshold according to degree of correlation order from big to small.

9. each described method in 6 according to claim 1 is characterized in that, before the described server query entries relevant with described subject information to memory query, also comprises:

Described server upgrades the query entries of the historical record stored in the described storer.

10. the information issuing system in the website is characterized in that, comprising: the server and client side of information publishing side, wherein,

Described client is used for sending to described server the subject information of the information to be released of user's input, and wherein, described subject information comprises title and the classification of described information to be released;

The server of described information publishing side is used for receiving the described subject information that described client sends; To the memory query query entries relevant with described subject information, wherein, described memory stores has the query entries of historical record; The query entries that inquires is sent to described client as the keyword of described information to be released, and by described client to releasing news that described keyword is selected to obtain.

11. system according to claim 10 is characterized in that, described server comprises:

The first title processing unit is used for described title being divided into the individual independently keyword of M, and choosing N keyword from a described M keyword to the memory query query entries relevant with described subject information the time, and wherein, M and N are natural number, and M 〉=N;

The first query unit is used for whether having the query entries that comprises a described N keyword from described storer inquiry;

The first judging unit, be used for when having the described query entries that comprises a described N keyword, judge that whether the number of the query entries that belongs to described classification in the query entries that inquires is more than or equal to P, if greater than, then will belong to front P query entries that inquires of described classification as the query entries relevant with described subject information, wherein, P is predefined natural number.

12. system according to claim 11, it is characterized in that, the number of query entries that described server also is used for belonging to described classification in the query entries that described the first judgment unit judges goes out to inquire is during less than P, repeat following steps, until the number of query entries that belongs to described classification in the query entries that inquires is more than or equal to P: described server makes N=N-1; Notify described the first title processing unit from a described M keyword, to choose N keyword; Notify described the first query unit from described storer, to inquire about whether there is the query entries that comprises a described N keyword; And notify described the first judging unit judges the query entries that belongs to described classification in the query entries that inquires when having the described query entries that comprises a described N keyword number whether more than or equal to P, if more than or equal to, then will belong to front P query entries that inquires of described classification as the query entries relevant with described subject information.

13. system according to claim 10 is characterized in that, described server comprises:

Selected cell is used for selecting the query entries that belongs to described classification from described storer;

The second title processing unit is used for described title is divided into M independently keyword, and chooses N keyword from a described M keyword, and wherein, M and N are natural number, and M 〉=N;

The second query unit is used for searching whether there is the query entries that comprises a described N keyword more than Q in the query entries that belongs to described classification that chooses, and wherein, Q is predefined natural number; If exist, then will belong to front Q query entries that inquires of described classification as the query entries relevant with described subject information.

14. system according to claim 13, it is characterized in that, the number that described server also is used for the query entries that inquires that belongs to described classification that finds out in described the second query unit is during less than Q, repeat following steps, until the number of the query entries that inquires that belongs to described classification is more than or equal to Q: described server makes N=N-1; Notify described the second title processing unit from a described M keyword, to choose N keyword; And notify described the second query unit in the query entries that belongs to described classification that chooses, to search whether to exist the query entries that comprises a described N keyword more than Q, if exist, then will belong to front Q query entries that inquires of described classification as the query entries relevant with described subject information.

15. each described system in 14 according to claim 10 is characterized in that described server comprises:

The second judging unit, be used for when described server sends to described client with the query entries that inquires as the keyword of described information to be released, judge that whether on-line checking result's the number of each query entries in the described query entries that inquires is greater than predetermined threshold;

Record cell is used for on-line checking result's number is recorded as the first group polling clauses and subclauses greater than the query entries of described predetermined threshold, and on-line checking result's number is recorded as the second group polling clauses and subclauses less than or equal to the query entries of described predetermined threshold;

Transmitting element is used for described the first group polling clauses and subclauses and described the second group polling clauses and subclauses are sent to described client as described keyword.

16. system according to claim 15 is characterized in that, described record cell comprises:

The first record cell is used for recording the number of described Query Result greater than the query entries of described predetermined threshold by following steps: calculate described on-line checking result's number greater than each query entries of the query entries of described predetermined threshold and the degree of correlation of described title; In described the first group polling clauses and subclauses, record the number of described Query Result greater than the query entries of described predetermined threshold according to degree of correlation order from big to small;

The second record cell is used for recording the number of described Query Result less than or equal to the query entries of described predetermined threshold by following steps: calculate described on-line checking result's number less than or equal to each query entries of the query entries of described predetermined threshold and the degree of correlation of described title; In described the second group polling clauses and subclauses, record the number of described Query Result less than or equal to the query entries of described predetermined threshold according to degree of correlation order from big to small.

17. each described system in 14 according to claim 10 is characterized in that described server comprises:

Updating block was used for before the described server query entries relevant with described subject information to memory query, and the query entries of the historical record stored in the described storer is upgraded.