CN102354313B - Conceptive method and system for organizing and expressing information - Google Patents

Abstract

Provided are a method and a system providing responses to queries. A plurality of choices during the same searching dialogue are connected, a query is received by a user, a searching result is provided in responding to the query, and one or more connected choices are provided in responding to the choice sent by the user. In one embodiment of the invention, the searching result URL is connected with one or more queries, wherein the selection of the URL and the queries are included in a searching dialogue. In responding to the queries, searching results of any query including one or more URL and connected with all the URLs are provided.

Description

The method and system of conceptive tissue and statement information

The application is that application number is 200480035838.9, the applying date be Dec 7 in 2004 day, be called the divisional application of the Chinese invention patent application of " method and system of conceptive tissue and statement information ".

Priority request

The application relates to and requires the right of priority of the provisional application number 60/528,139 of submitting on Dec 8th, 2003, and its content is incorporated into this as quoting.

Related application

The name that the application relates on May 24th, 2004 and submits to is called the U.S. Patent application of " METHODS AND SYSTEMS FOR CONCEPTUALLY ORGANIZING AND PRESENTING INFORMATION ", and its content is incorporated into this as quoting.

Technical field

What embodiments of the invention were general relates to the field of notional organizational information, and the use analysis that relates in particular to conceptual dependency information is with effective organizational information.

Background technology

Along with increasing rapidly of information, the ability of organizational information is also increasing.For example, in the relevant network in internet (WWW) or other the Internet sources, can find bulk information source.Internet is the extended network of computer network, and information exchange is crossed the method for well known to a person skilled in the art (use of for example TCP and IP agreement etc.) and exchange on the internet.Internet allows user to transmit and receive data between the computing machine that is connected to this network.These data can comprise website, homepage, database, text collection, audio frequency, video or by being connected to the information of any other type that the computer server of internet can use on the internet.These information can be called as file or document, and can comprise other data that for example, comprise in data, webpage annex or the memory device (database) on webpage, webpage.

Understand the meaning of large volume document set like this and in this environment search information, in the time there is no special supplementary means, be very difficult.A kind of method of auxiliary positioning information is to use keyword.That is to say, document can comprise the keyword of the selected part that represents the information in document that is included in.These keywords are available to other computing machines on the internet and allow other computing machine location the document.

In order to locate the document on internet, the user of remote computer uses and is called as the search utility of search engine and searched key word.Search engine is to allow long-distance user to key in the program of one or more search words.Then search engine is compared search inquiry and is at least had the part of the keyword mating with search inquiry in search file with the keyword in document.Then search engine for example, to user display section document, title.User can be in the local document retrieving rolling view and select required document.

Early stage keyword search engine demonstrates major defect.For example, in order to increase the occurrence rate of particular document, document supplier can use the as much as possible search word relevant to document.In fact, the each word in some document or search engine use document is as keyword.Therefore, search engine can retrieve in a large number and need to combine the document that theme is irrelevant or only edge is relevant of finding by search word to user.And, thereby a lot of users of these search engines and be unfamiliar with form keyword search query skill and produce too wide in range search often retrieve thousands of documents.User must check about the summary info of each document with location information needed so.

This defect is solved by the differentiation of search engine, comprises the Information Organization of the search activities based on one or more users.The parameter (for example text) of these methods based on user preference degree prevailing view rather than Oriented Documents and to sort result.Wherein a kind of such method is according to the evolution score of the keyword based on used and to document ordering.That is to say, document receives the relevance score relevant to the keyword of search inquiry.When inquiring about and select document in the lists of documents of inquiry generation along with user's inputted search, the relevance score of document is adjusted.These marks are used to the results list of organizes documents to carry out subsequent searches." clicks " number of times (being the selecteed number of times of document) that these methods typical (at least in part) receive based on document and definite degree of correlation.These methods are commonly referred to " popularity sort method " or " click people gas method ", provide the document attracting most and meet the previous user of most is arranged in to top search result list.And, click people's gas method and produce the contextual result of reflection search.For example, previous searching method can return to the document that comprises all query words, but can automatically not get rid of the word that is not query portion.Therefore, most possibly return to the result about " New Mexico " for the text matches search of " Mexico ".Click people's gas method and can reduce this error result, because the user of search " Mexico " conventionally can not click about the page of " New Mexico " and can tend to click them and think and " Mexico " the maximally related page, thereby increase the degree of correlation of required document.

Because the information relevant to query word is along with the time may change, determine the degree of correlation result that may lead to errors based on number of clicks.For example, for the ad hoc inquiry of " democracy pioneer ", the document that relates to early stage pioneer Howard Dean may be selected many times in Dec, 2003, but the user who inputs this inquiry in March, 2004 may be desirably in now the result as leader's John Kerry.And, come top result and conventionally can be subject to out-of-proportion more use, cause like this Search Results of more and more distorting, wherein come top result replaced never.

The search engine of including a large amount of factors based on time and the factor based on using, the information providing in response to inquiry being organized by use has solved the part in these defects.For example, this method may be used previous user to adjust the degree of correlation of inquiry response document in response to the activity of ad hoc inquiry.Such User Activity can comprise number of clicks and the previously selection of user to customizing messages or the time of use.These methods in particular document can also being arranged in to previous the results list in the time that previously user selects it where, document actual and expect how the contrast of frequency of utilization and selected document are taken into account by use etc.

But, in current method, still there are a large amount of defects.For example, current method does not solve the problem of very rare inquiry, does not wherein collect enough user activity data.In the case, the possibility of result seldom or not exists.And click result depends on quality and the integrality of data source.Current method can not solve the greatest differences of data source quality.And current method is subject to deception impact, may affect the integrality of Search Results.Attempt to solve some problem by upgrade search-engine results on the basis of User Activity according to a kind of exemplary method of prior art.This method is called in name in the U.S. Patent No. 6,421,675 of " Search Engine " and is described, and its content is incorporated into this as quoting, thereby the comprehensive description to prior art is provided and has clearly distinguished the feature of each embodiment of the present invention.

Summary of the invention

The invention provides a kind of method, the method comprises: receive and record queries, described inquire-receive is from multiple users; Determine the Search Results corresponding to described inquiry; The one or more Search Results that receive and record described multiple users are selected; Described selection is associated with recorded inquiry; Receive the signal that comprises described inquiry from user; And in response to described inquiry, the one or more Search Results that comprise the selection associated with described inquiry are offered to described user.

The present invention also provides a kind of system, this system comprises: server count word processing system (DPS), wherein this DPS can receive inquiry from multiple users, in response to described inquiry, one or more Search Results are offered to described user, wherein each user can be provided by least one in provided Search Results, described DPS can also receive and record the signal of the one or more Search Results selections that comprise described user, recorded Search Results is selected to be associated with recorded inquiry, receive described inquiry, and provide one or more Search Results that comprise the Search Results associated with described inquiry in response to described inquiry, and one or more client DPS that can communicate by letter with described server DPS, wherein said communication comprise by inquiry and Search Results select at least one offer described server DPS and receive one or more Search Results.

The present invention also provides a kind of system, this system comprises: server count word processing system (DPS), wherein this DPS can receive the first inquiry from multiple users, determine the multiple Search Results corresponding to described the first inquiry, wherein each user can be provided by least one in provided Search Results, described DPS can also offer described user by described multiple Search Results, recorded Search Results is selected to be associated with recorded specific selection, receive inquiry, and provide multiple Search Results in response to described inquiry, wherein said multiple Search Results has comprised described specific selection and has selected with associated Search Results, and one or more client DPS that can communicate by letter with described server DPS, wherein said communication comprises that at least one during inquiry and specific selection and Search Results are selected offers described server DPS and receive one or more Search Results.

Accompanying drawing explanation

Can better understand the present invention by reference to description below, and utilize accompanying drawing to represent embodiments of the invention.In the accompanying drawings:

Fig. 1 shows that the association of the response based on from multiple users according to an embodiment of the invention revises the process flow diagram of the tissue of conceptual dependency information aggregate and the program of statement;

Fig. 2 shows according to an embodiment of the invention the process flow diagram that affects user's inquiry and the program contacting of storage content by the association of the User Activity that obtains during the similar information search sessions based on other users and/or user profile;

Fig. 3 shows according to an embodiment of the invention for organizing and explain conceptual dependency information and for contacting the structural drawing of system of particular memory content and each user inquiry;

Fig. 4 has shown the search daily record of three isolated users according to an embodiment of the invention;

Fig. 5 is for showing the process flow diagram of the program that more relevant search results are provided according to an embodiment of the invention;

Fig. 6 is for showing the process flow diagram of the program that relevant inquiring suggestion is provided according to an embodiment of the invention;

Fig. 6 A is for showing the process flow diagram of the program that query spelling correction suggestion is provided according to an embodiment of the invention;

Fig. 7 is the process flow diagram that shows the program of the inquiry of offering suggestions according to an embodiment of the invention;

Fig. 8 is for showing the process flow diagram of the program that similar results is provided according to an embodiment of the invention;

Fig. 9 is for showing the process flow diagram of according to an embodiment of the invention information being correlated with the program of the Search Results that provides more relevant;

Figure 10 is for showing according to an embodiment of the invention according to providing the process flow diagram of the program of more relevant Search Results based on customer location to selecting the position that is associated;

Figure 11 is for showing the process flow diagram that based on customer location, user is provided according to an embodiment of the invention the program of more relevant Search Results; And

Figure 12 shows the structural drawing of digital processing system according to an embodiment of the invention.

Embodiment

general introduction

Embodiments of the invention provide the method and system for conceptive tissue and statement information, and wherein the response of user to Information Organization and statement is associated to determine optimum tissue and the statement of information.In one embodiment of the invention, in the context of optimizing at search-engine results, during search sessions multiple users' User Activity and/or user profile and inquiry carry out associated, with affect the tissue of inquiry and document and explain between evolution contact.According to the User Activity in the whole search sessions process of the system storage of these embodiment, thereby can make a large amount of dissimilar User Activities can be associated with user profile.Use associated user to input and allow these systems relevant Search Results to be provided and can not to produce the restriction that in prior art, the system based on keyword is brought.

To provide in the following description a large amount of specific detail.But, should be appreciated that embodiments of the invention may be embodied as and do not have these specific detail.In other embodiments, known circuit, structure and technology are not described in detail to avoid fuzzy understanding of this description.

" embodiment " who mentions in whole instructions or " embodiment " represent that special characteristic, structure or the feature described in conjunction with this embodiment comprise at least one embodiment of the present invention.Therefore phrase " in one embodiment " or " in an embodiment " that, in whole instructions, many places occur might not all refer to same embodiment.And, can in any appropriate manner special characteristic, structure or feature be combined in one or more embodiments.

And the aspect of inventing is included in all features that are less than single disclosed embodiment.Therefore, the accompanying claims of instructions are clearly included in this embodiment, and every claims self are as an independent embodiment of the present invention.

program

Fig. 1 has shown the association of response based on from multiple users according to an embodiment of the invention and has revised the tissue of conceptual dependency information aggregate and the program of statement.

Program 100 shown in Fig. 1 starts from operating 105, and wherein conceptual dependency information aggregate is organized by informant and is explained to multiple users.In one embodiment, conceptual dependency information aggregate is the set page including about the information of the interested universal of particular group of users.This set page can comprise the correlator concept of a large amount of any kinds, comprise and the linking of file, catalogue, database, spreadsheet, news item, audio frequency, video, image, application program, advertisement, product description and reference information, with linking of any catalogue of list, form, tree or above-mentioned project, and gather linking of the pages with other, all these contents can be collected from the source of any amount.In one embodiment, for a change tissue and the statement of information aggregate of multiple set page creation forming separately, comprises quantity, type, arrangement and the significance of correlator concept.Like this, informant attempts the optimum way of intended tissue and statement information.

In operation 110, informant receives the response about Information Organization and statement from user.User's response can be for selecting the form of (or not selecting) certain partial information.For example, certain user can select the information of the specific sub-concept of the given set page, and does not select other.

In operation 115, the response receiving from multiple users is associated.That is to say, determine in the identical statistics that individual consumer carries out the effectively degree of contact.In alternative embodiment of the present invention, the association of response can be taked any form.The example association of various user's responses is hereinafter by more detailed description.

In operation 120, the response based on after association and revise tissue and the statement of conceptual dependency information aggregate.For example, can again form the set page to meet more fully individual consumer, user's group or certain class user or all users' needs.Except revising the statement of information of original statement, described again formation can comprise and adding or deletion information.For example, can add one or more sub-concepts to the set page or from wherein deleting.

Described user and respond to optimize information aggregate tissue and the statement of (for example gathering the page) although general above, embodiments of the invention can for example, for affecting the contact of user search engine queries and storage content (one or more documents).That is to say, embodiments of the invention can be for determining more relevant Search Results (being general more relevant or more relevant to specific user) in response to ad hoc inquiry.

Fig. 2 has shown that according to an embodiment of the invention the association of user search engine (USE) activity by obtaining during the similar information search sessions based on other users and/or user profile affects the program contacting of user's inquiry and storage content.

Program 200 shown in Fig. 2 starts from operating 205, wherein to USE action message and/or user profile during the session of some isolated user record searchings.Search sessions comprises the sequence of any search engine action (activity that can record by search engine) of given user.USE activity can comprise the link of sending inquiry, click guiding inside on searched page or external data, click the link of guiding inside in follow-up internal pages or external data and return to searched page or any internal pages after clicking inside or external linkage.USE activity can be for continuous or continue to occur in the cycle in reality.That is to say, can specify the time cycle that represents that search sessions stops.For example, if the interruption in the User Activity recording has exceeded the fixed time, can suppose that in practice search sessions finishes.Subsequent user activity can be thought new search sessions.

In operation 210, USE action message and/or the user profile of the some isolated users that record are associated.Described message reflection each user's whole search sessions, can carry out in every way association according to alternative embodiment of the present invention.The example association of various USE activities and/or user profile will be described in detail hereinafter more.

In operation 215, be used to affect the contact between particular memory content and respective user inquiry by the USE action message of the some isolated users that record and/or the association results of user profile that operate 210 acquisitions.This contact can provide the Search Results of given inquiry general or that be more correlated with for one or more specific users.

system

Embodiments of the invention can be implemented in network environment.Fig. 3 has shown the system for organizing and explain conceptual dependency information and inquiring about for contacting particular memory content and each user according to an embodiment of the invention.System 300 has as shown in Figure 3 shown the network of digital processing system (DPS) 300, comprises the one or more server DPS that are shown as server DPS 320, and is shown as multiple client DPS of client DPS 305-308.The DPS of system 300 is connected to each other and is configured to exchange multiple various types of storage contents that comprise document, and the content of for example, storing on Webpage, Webpage, comprises text, picture and Voice & Video content.For example, the content of storing can be audio/video file, for example, have the program of mobile image and audio frequency.Information can communicate by the communication network of any type between DPS, multiple distinct devices can communicate by described communication network, for example (but being not limited to) internet, the wide area network (WAN) not showing, LAN (Local Area Network) (LAN), Intranet etc.For example, as shown in Figure 3, DPS is connected to each other by internet 310, and internet 310 is the wherein one that comprise multiple networks with data communications method as above, and is known for those skilled in the art.The communication linkage of connection server DPS and client DPS might not be directly link, but can be indirect link, includes but not limited to broadcast wireless signal, network service etc.Although shown the DPS of example in Fig. 3, be appreciated that and can be connected to each other a large amount of such DPS.

Can will be described with reference to Figure 12 hereinafter for the embodiment of the digital processing system of server DPS 320 or client DPS 305-308 according to an embodiment of the invention.

According to one embodiment of the invention, the set of conceptual dependency information, for example, gather the page, explains to some users by client DPS 305-308.Described conceptual dependency information can also be the Search Results in response to the user's inquiry sending from one or more client DPS 305-308.Described information can adopt various ways, for example, can be the webpage URL address list providing by Webpage developer.Once it is upper to be expressed in client DPS, user makes a provisioning response to described statement.For example, user can carry out USE activity as above.Relevant user responds and other user profile are recorded and are sent to user's response/information affiliate application 321 to user's information is explained.The user response of this application program 321 to some users and information are carried out associated, and tissue and the statement of results modification conceptual dependency information based on associated.

data analysis

Embodiments of the invention obtain USE action message and/or user profile and these information are carried out to association, select and the auxiliary definition degree of correlation with the majority by using multiple users.Described association analysis comprises the program of evaluating common action or multiple user profile, to identify the upper effectively contact of statistics.The term " contact (association) " that these embodiment are used and " effectively contact (statistically significant association) in statistics " are defined as follows." contact " is that user is clearly or indirectly, consciously or any pairing of unconscious definite inquiry, term, concept, document or other network datas and combination thereof during search sessions.Contact can for example, represent by the link (hyperlink) of sending inquiry and/or selective guide inquiry, term, concept, document or other network datas.In statistics, effectively contact is the contact that can not belong to random occurrence on probability.When independently user adds up effectively contact on by two or more surface, record described association.

Embodiments of the invention by the more a large amount of information of record and more specifically information (comprising USE action message and/or user profile) provide than art methods more relevant search-engine results far away, and realized the more deep analysis of information.

One embodiment of the present of invention provide a kind of system of the data file that creates and operate all USE action messages of maintenance as shown in table 1 and user profile.(for the actual data structure that to describe the form of each embodiment of the present invention be only exemplary and not necessarily represent the embodiment of the present invention.)

Timestamp

User ID

Inquiry

Select (URL)

1/1/03 00:00:00	U1	Q1	P5
				1/1/03 00:01:00	U1	Q2	P1
1/1/03 00:02:00	U1	Q2	P2
				1/1/03 00:02:05	U1	Q2	P3
1/2/03 00:00:00	U2	Q2	P4
				1/2/03 00:01:00	U2	Q2	P1
1/2/03 00:02:00	U2	Q2	P4
				1/2/03 00:04:00	U2	Q1	P2
1/2/03 00:04:05	U2	Q1	P3
				1/3/03 00:00:00	U3	Q3	P3
1/3/03 00:04:00	U3	Q2	P1
				1/3/03 02:00:00	U3	Q3	P5
1/4/03 00:00:00	U4	Q1	*
				1/4/03 00:06:00	U4	Q2	P4

Table 1 (*=do not contact the inquiry of selection)

Table 1 has shown the data file that comprises mass data element, and these data elements have recorded the inquiry of each time point a large number of users and after each inquiry of each user, selected the URL (selection) of (click).This data file can comprise a large amount of other data elements that represent USE action message and/or user profile.These data elements, for example, can represent order, the IP address of the result that user clicks during the demonstration sequence, session of selection result, geographic position of IP address etc.

This is different from various art methods, and in the prior art, in cyclical process and after loading new data, the inquiry of simplification-result selects (Q2RP) associated by input database form, for example, show 1A.According to one embodiment of the invention, these not typings of information, but maintain with daily record form (log form), data element represents all USE action message and user profile.

Inquiry	Select (URL)	Score
			Q1	P2	1
Q1	P3	1
			Q1	P5	1
Q2	P1	3

Q2	P3	1
			Q2	P4	2
Q3	P3	1
			Q3	P5	1

Table 1A

Table 1A is the simplification example of prior art data structure, and the score known to can comprising in prior art is adjusted territory.For example, score can be simply for selecting the summation of number of times, can be also the result of more complicated adjustment algorithm.Score and adjustment also can be stored.

Prior art data structure in table 1A is enough to produce the Search Results of sequence, but it has represented the loss of a large amount of raw informations.As show as shown in 1A a large amount of available USE action message and user profile during art methods does not have record or analyzes search sessions.This comprises storage restriction, lacks the actual use to these information and do not recognize the adaptable prospect of these information because a variety of causes causes.

According to one embodiment of the invention, can omit and record these scores and adjust territory by recording and store the extension of information.That is to say, to these embodiment, do not store score adjustment information, because required any score can be based on institute's canned data and parametrization is calculated.And the parameter calculating the score and algorithm can change as required and not affect stored data.

Each embodiment according to the present invention, depends on the ability of identifying and recording the associated data (comprising USE action message and user profile) between isolated user to the in-depth analysis of information.That is to say, by recording and analyze more a large amount of information, almost comprise all information during user conversation, can obtain the Search Results with the higher degree of correlation far away.

For example, consider that user sends a series of inquiries and interleaves the user conversation of selection wherein.Common, be incoherent sending the URL selecting before inquiry A with inquiry A, because user often changes theme.Similarly, it is incoherent that the great majority that occur after follow-up inquiry B are selected with inquiry A.

And, consider all a large amount of isolated users that comprise the search sessions of inquiring about A that recorded.Can expect these users before inquiry A and select afterwards various irrelevant themes, therefore, irrelevant selection meeting wide dispersion is upper in a large amount of URL, and each URL can obtain very low score, has reflected that they lack and contact with inquiry A.Typically, only there is user-selected related urls very small amount of, that really remained on those themes relevant to inquiring about A by those can accumulate the rearrangement of sufficiently high click popularity mark with the impact Search Results associated with inquiring about A.

For example, suppose 1000 user search queries A.Then, 900 people in them select the inquiry B1 to B900 haveing nothing to do.100 remaining people continue to search for the various variations of original theme and select relevant inquiry A1-A9.Each of inquiring about by B in the result producing can receive once or twice and the associated selection of inquiry A, but the result producing by relevant inquiring A1-A9 on average can be accumulated the selection of ten times.

For one embodiment of the invention, can force such requirement, the URL after inquiry must be selected at least twice with relevant to original query.Such requirement can be eliminated and a large amount of B query selections of inquiring about A mistake and contacting.

According to one embodiment of the invention, provide the association analysis that depends on a large amount of statistical samples, to identify multiple relevant contacts.For these embodiment, reduce the restriction to analyzed contact and random data in prior art and divided to increase the degree of correlation.That is to say, record and the more a large amount of user session information of analysis also allow to analyze the contact of more correlation types.

the USE activity association of example

According to one embodiment of the invention, provide a kind of system of the small set and the combination thereof that use one or more basic associations.Common, each embodiment according to the present invention can determine the associated of any amount and increase or other targets for realizing the Search Results degree of correlation.Describe the association of some example below in detail.

inquiry-select (QUERY-TO-PICK)

Inquiry-select (Q2P) association that inquiry is associated with selection.In the time that multiple isolated users carry out identical contact, this identical contact is associated candidate.When search engine response returns results and when user selects this result, this is the particular case (Q2RP) of this association in inquiry.In practice, search engine algorithms substitutes the second isolated user.According to one embodiment of the invention, Q2P association is associated inquiry with selections all in user conversation.This is different from the scheme of prior art, in the prior art, stops contacting between given inquiry and selection once send subsequent query.

By Q2P, the given inquiry of sending during all selections of recording during user conversation and this user conversation is associated.In one embodiment, be that each contact distributes score based on various factors, these factors comprise the quantity of time, the inquiry interleaving and/or selection between inquiry and selection and the inquiry order with respect to selection.

And, can adjust based on known factor the score of each contact, the time limit of the time limit of the sequence of selecting in the results list when these factors comprise contact, the perdurability of selection (interval before next known users action), contact or the first known example of order (with respect to contact older or that upgrade) and contact.

Each user conversation can have the unlimited duration.In actual applications, can force the restriction of rational time restriction or insert action, exceed outside this restriction and selecting specified relationship not between inquiry.Interchangeable or additional, the interruption of enough duration can represent the interruption of session.Searching for according to an embodiment of the invention Log Summary is presented in table 2 below.In various alternative embodiments, can in this search daily record, catch any other project, but for clarity in this omission.

Table 2 (*=do not contact the inquiry of selection)

Fig. 4 has shown the search daily record of three isolated users according to an embodiment of the invention.Search Log Summary 400 shown in Fig. 4 comprises the search daily record 410,420 and 430 of the search information of describing respectively three isolated user U1, U2 and U3, described in above reference table 2.Q2RP part in empty wire frame representation Q2P search information in each search daily record.For example, search daily record 410 comprises Q2RP part 411, wherein inquires about Q1 and has produced selection P5.Search daily record 410 also comprises Q2RP part 412, wherein inquires about Q2 and has produced selection P1, P2 and P3.

Search daily record 420 comprises that the duration is the interruption 421 of 48 hours.In one embodiment of the invention, interruption can represent two independent sessions so for a long time, can not specify the relation between any selection and inquiry between the two.Contrary, search daily record 430 comprises that the duration is the interruption 431 of 2 hours.In one embodiment of the invention, this interruption can not represent two independent user conversations.That is to say, all search activities of search daily record 430 can be thought the information of single user session and respective associated.

Table 2A has shown the form of the click information comprising in table 2 according to an embodiment of the invention.For relatively, show 2B and shown according to the form of the click information comprising in the table 2 of the typical art methods of employing Q2RP association.

	Q1	Q2	Q3
				P1	2	3	1
P2	1	1	-

P3	2	3	1
				P4	1	1	-
P5	1	2	1

Table 2A (Q2P result)

	Q1	Q2	Q3
				P1	-	3	-
P2	1	-	-
				P3	1	1	1
P4		2	-
				P5	1	-	1

Table 2B (the Q2RP result of prior art)

In damaed cordition below, our supposition once selects to represent that score increases 0, and because a large amount of factors may change score or make score loss, we suppose selection=score+1 1 time.Suppose a time threshold, the click in row 103 is all lost in two forms, because user only spends the very short time on this URL.Assumption database is carried out batch processing renewal every day, in the prior art form of the click in row 203 in table 2B, conventionally can be lost as clicking 201 repetition.According to one embodiment of the invention, the click in row 203 and 402 is lost in form as clicking 201 repetition.

According to one embodiment of the invention, for inquiry Q1, the URLP1 never clicking immediately Q1 after has obtained high score in form because multiple user send inquire about before Q1 or afterwards (although be not immediately after) selected it.According to one embodiment of the invention, the whole score matrix of form is more substantial, because recorded more contact.Some mark is lower, the mark of for example Q2P4, and this is because the session data of preserving represents that all clicks are from unique user, thereby allows the more repetition of identification.

In the practical application of Q2P, we can keep specific contact is that Q2RP is also the difference of non-Q2RP.Single, not associated non-Q2RP clicks (for example Q3P1 in form), and the enough degree of confidence of result to user that provide can be provided, but click for single, not associated Q2RP, provide the fact of the result of original search to strengthen contact by search engine.

select-inquiry (PICK-TO-QUERY)

Select-inquiry (P2Q) association is that all inquiries of recording during user conversation are associated, and during these inquiries and this user conversation, the given selection of sending is associated.Search Log Summary in table 2 has shown the output that P2Q is relevant.That is to say, the identical data that Q2P is produced can be regrouped index to P2Q.

inquiry-inquiry (QUERY-TO-QUERY)

Inquiry-inquiry (Q2Q) association is that the every other inquiry of sending during all inquiries of sending during user conversation and this session is associated.In one embodiment, can be that score is specified in each contact based on various factors, these factors comprise whether time limit of quantity, contact of the time between inquiry, the inquiry interleaving and/or selection or order (with respect to contact older or that upgrade), Query Result have produced and select and the paired order of the inquiry of contact.

Determine whether Query Result has produced and select and the paired order of the inquiry of contact can provide many especially information, because they can represent whether one query is " association " of another one query.For any practical application, know which correct which mistake in the inquiry of two contacts is of great use.

Searching for according to an embodiment of the invention Log Summary is presented in following table 3.Only need the query portion of searching for daily record to create Q2Q form.

Table 3

Table 3A has shown the form (order of sending inquiry is ignored in supposition) of the click information comprising in table 3 according to an embodiment of the invention.

	Q1	Q2	Q3
				Q1	-	2	-
Q2	-	-	1
				Q3	-	-	-

Table 3A (Q2Q result)

The lower Delta Region of table 3A can be for preserving paired inquiry order information, avoids two registration (double-booking) situations as colleague 301-303.

As mentioned above, can adopt the point system of utilizing various factors to change score or making score loss.For example, can punish repeating (such as the contact in the contact in row 101 and 102 and row 401 and 402).Or, not associated Q2Q contact, for example Q2Q3, the enough degree of confidence of result to user that provide can be provided.

select-select (PICK-TO-PICK)

Selecting-select (P2P) association is that the every other selection of sending during all selections of sending during user conversation and this session is associated, and like this, the associated and above-mentioned Q2Q association class of P2P seemingly.Same, according to each embodiment, can specify score to each contact based on various factors, these factors comprise the paired order of quantity, the time limit of contact or the selection of order (with respect to contact older or that upgrade) and contact of the time between selection, the inquiry interleaving and/or selection.

Searching for according to an embodiment of the invention Log Summary is presented in following table 4.Only need the selection part of searching for daily record to create P2P form.

Table 4

Table 4A has shown the form (order of sending selection is ignored in supposition) of the click information comprising in table 4 according to an embodiment of the invention.

P1

P2

P3

P4

P5

P1	-	1	3	1	2
						P2	-	-	1	1	-
P3	-	-	-	1	2
						P4					-
P5

Table 4A (P2P result)

Same, repeat (such as relating to the contact of row 201 and relating to the contact of row 203) and may incur loss, cost very short time, the user on URL also can incur loss.

The lower Delta Region of table 4A can be for preserving paired order of preference information, avoids the two registration situations as colleague 201-203.

the combination of the USE activity association of example

The various alternative embodiments according to the present invention, can for example, link together by two or more association (described above substantially associated) Search Results that provides more relevant.For example, thus the basic association that can connect two or more strengthens its result to imitate basic association, particularly needs the additional result of broad match in rare data or generation.

Association after connection is used selection or inquiry rather than user, to form the link between other selections and inquiry.Common, the association of connection is more, and result departs from initial selected or inquiry is far away.Therefore, under many circumstances, connecting minimum association is optimum approach to produce results needed.

Table 5 has shown movable basic two kinds of associated associated may the combining of above-mentioned USE.

Associated	Combination	Basic analog	Produce thing
				QPQ	Q2P+P2Q	Q2Q	Inquiry is to the selection relevant to original query
QQQ	Q2Q+Q2Q	Q2Q	Inquiry is to the inquiry relevant to original query
				PQQ	P2Q+Q2Q	P2Q	Inquiry is to the inquiry relevant to original selection
PPQ	P2P+P2Q	P2Q	Inquiry is to the selection relevant to original selection
				PPP	P2P+P2P	P2P	Select the selection relevant to original selection
PQP	P2Q+Q2P	P2P	Select the inquiry relevant to original selection

QQP	Q2Q+Q2P	Q2P	Select the inquiry relevant to original query
				QPP	Q2P+P2P	Q2P	Select the selection relevant to original query

Table 5

Connect association and may introduce wrong relation, therefore, in one embodiment of the invention, association is carried out to association.For example, if Q2Q association needs two isolated users, QQQ associated character string should need to link two inquiries of original and final inquiry.In QQQ, the inquiry (Q3) interkniting is associated an inquiry (Q1) with another inquiry (Q2).If identical contact is carried out in two or more inquiry independently, interkniting, this is a kind of associated.

As above with reference to described in basic association, the contact between original selection or inquiry and output selection or inquiry at least forms (or adding body one by one by a search engine) by two each and every one bodies.By the association connecting, can not have individual consumer's (or search engine) that original selection or inquiry are selected or inquire about with any output to be associated.Indirect association is two isolated users of minimum needs also.

In effect, the association of connection tend to future that prediction is collected in more a large amount of data when certain time point basic association what may be.In essence, the possible association that their identification is not yet observed.

Associated combination needs the association between multiple middle selections or inquiry.Link the different selection of original and relevant selection or inquiry and/or the quantity of inquiry, than centre selection and/or inquiry and original and relevant selection and/or query link approach more important.For an association, must there be at least two different link paths, and no matter there are how many users to set up these links.That is to say, by an intermediate node Q3 contact Q1 and Q2, also can't form the association between Q1 and Q2 even if multiple user has set up this contact.

A large amount of factors have affected associated intensity, comprise the quantity of link path, the intensity of each direct correlation component link and the uniqueness of each intermediate node.For example, link for example, may be produced than passing through unique intermediate queries (such as " 1965 Ford wild horse roadster ") and be linked two selections more weak association far away by public and general inquiry two associated selections of (" automobile ").

Connecting the association " inquiry-select-inquiry " of associated advantage after by connection below example (QPQ) can better understand.

Table 6A and table 6B have shown respectively the multiplied result of the intersection QP score of table 2A and 2B, and for inquiring about these results added to determine the contact score after combination.(this might not be optimal algorithm, but for example object).

P1

P2

P3

P4

P5

Summation

Q1-Q2

2×3

1×1

2×3

1×1

1×2

16

Q1 to Q3

2×1

-

2×1

1×1

5

Q2 to Q3

3×1

-

3×1

2×1

8

Table 6A

	P3	P5	Summation
				Q1 to Q2	1×1	-	1
Q1 to Q3	1×1	1×1	2
				Q2 to Q3	1×1	-	1

Table 6B

Table 6C and table 6D have shown respectively the equivalent Q2Q association table of table 6A and table 6B.

	Q1	Q2	Q3
				Q1	-	16	5
Q2	-		8
				Q3	-

Table 6C

	Q1	Q2	Q3
				Q1	-	1	2
Q2	-	-	1
				Q3	-	-	-

Table 6D

As mentioned above, according to one embodiment of the invention, QPQ association results has produced more associated far away than the narrower QRPQ association of more specific alternative embodiment of the present invention, and beguine is more associated far away according to the associated generation of Q2Q of another embodiment of the present invention.And the inquiry that uses the associated permission of QPQ can not produce Search Results to incorrect formation is offered suggestions.This is impossible in art methods.

user-user

As inquiry and/or select can be by user-association, user can be by inquiring about and/or selecting and associated.As being called as user to user (U2U) with the general program of indirect association like QPQ association class.The U2U association (two users have inputted identical inquiry and selection) of selecting based on result is presented in table 7 below.This association should be user-result selection-user (URPU), although there are other U2U associations of the alternative embodiment according to the present invention.Same, according to various embodiment, can specify score to each contact based on various factors.For example, suppose time threshold, the click in row 103 is lost, because user only spends little time on URL.

Table 7

Table 7A has shown the form of the click information comprising in table 7 according to an embodiment of the invention.

	U1	U2	U3
				Q1P2	-	1	-
Q1P3	-	1	-
				Q1P5	1	-	-
Q2P1	1	1	1
				Q2P2
Q2P3	1	1
				Q2P4		1
Q3P3			1
				Q3P5			1

Table 7A

According to one embodiment of the invention, a given user (for example U1) for example,, to the unison quantity that can be defined as inquiry/selection that this user and another user share of another user's (U2) class, (is that class is unison divided by inquiry/selection sum of given user _u1-U2=(shared QP _u1-U2)/(QP _u1sum)).In alternative embodiment of the present invention, can use the unison algorithm of more complicated class.For example, according to one embodiment of the invention, class is unison, and algorithm can be taken the difference of search rate into account and inquiry, selection and inquiry-selection are imposed to different weights.

Table 7B has shown the unison information of class that the click information of his-and-hers watches 7A is calculated.

Table 7B

Use the unison information of this class, in the time that the inquiry that given user is sent in the future bears results, adjust the selection score of respective user according to the unison information of described class.For example, the selection score producing with 0.67 adjustment user U2 and 0.33 is adjusted the selection score that user U3 produces by the future Query result of user U1.There is no the selection of the unison user's generation of class by designated certain default value.

example application

The method and system of the tissue of above-mentioned each alternative embodiment of reference the present invention and statement data can be for various practical applications, and this it will be apparent to those skilled in the art that.These application are more comprehensively discussed below.The use of specific USE action message and/or user profile can be more suitable for application-specific than other information.For example, for application-specific, specific USE activity is carried out to association will be more practical, more effectively or more accurate.Example application is below by for being particularly suitable for the USE action message of actual enforcement of application-specific and/or user profile and describing.

the Q2P application of example

Of the present inventionly Q2P and/or Q2P equivalent combinations USE action message are carried out to associated embodiment allow user to obtain in every way more relevant Search Results.For example, user can search for and some part of revised result is associated with original search by precision.That is to say, the document of selection might not be associated with original search on word, and be only notional relation.Conceptual relation can provide original search is better responded.Associated Search Results based on Q2P USE action message has been avoided the restriction of text based retrieval in art methods.User's search procedure again can be preserved and utilize to embodiments of the invention.This ability can have a large amount of obviously systems of advantage for realizing than art methods.

Not only utilize previous user's degree of correlation judgement according to the association of the Q2P USE action message of a large amount of isolated users of various embodiments of the invention, and utilized its research effort.Subsequent user does not need the previous user's of repetition mistake, but can from the lesson of trial-mistake of previous user, be benefited.

Fig. 5 has shown the program that more relevant search results are provided according to an embodiment of the invention.Program 500 shown in Fig. 5 starts from operating 505, wherein receives inquiry from user.Described inquiry can have one or more particular characteristics, once these characteristics are identified, can be used as according to the present invention each alternative embodiment provides the basis of more relevant Search Results.

In operation 510, the Q2P USE action message of a large number of users carries out association for received inquiry.Each inquiry can have various particular characteristics, and these characteristics can be determined by the association of Q2P USE action message.These particular characteristics for example can comprise: inquiry can have different meanings to different user, inquiry can be spelt by mistake, inquiry can have wording of equal value, inquiry can have comparatively relevant part and not too relevant part, inquiry can be associated with particular result output or result output mix, and inquiry can have wider or narrower Search Results.Each characteristic in these characteristics, regardless of being independent or combining, may be useful for more relevant Search Results is provided under different condition.

In operation 515, the Search Results of the Q2P USE action message based on after association is provided for user in response to inquiry.The Search Results providing can based on inquiry one or more characteristic.For example, in the time that definite described inquiry has different meaning, can provide in response to inquiry the result with the meaning based on more popular.

Some example results of obtaining by program 500 and the result obtaining by typical prior art compare, and show below with respect to the special characteristic of above-mentioned inquiry.

Table 8 has shown the exemplary search results having more than a kind of some inquiries of meaning.It is as shown in the table, and compared with art methods, embodiments of the invention allow Search Results to be related to the more popular meaning of ad hoc inquiry.

Table 8

Table 9 has shown the exemplary search results (for example " encycopidea ") of mistake spelling.It is as shown in the table, and embodiments of the invention allow Search Results to be related to the inquiry that may correctly spell.In this way, embodiments of the invention can be determined the correct spelling of mistake spelling inquiry.Art methods can not produce Search Results conventionally to this mistake spelling, or very poor Search Results.

Table 9

In one embodiment of the invention, this spelling correction is that a kind of " soft " proofreaies and correct.That is to say, according to one embodiment of the invention, the result providing in response to inquiry is the accurately result of the most of user institute preference of inquiry of input.If most users thinks that this inquiry is for mistake spelling, large result is by the inquiry comprising after correction.If most users thinks that this inquiry is intentional, large result will comprise the inquiry of not changing.If two kinds of explanations are all legal, result is the two combination.For this embodiment, because all corrections are all that conceptual dependency is not only literal similarity, therefore unlikely provide wrong spelling correction.

Contrary, the spelling correction of prior art is generally " firmly " and proofreaies and correct.That is to say, these method identification mistake spelling inquiries, attempt proofreading and correct then based on proofreading and correct Search Results.When legal inquiry is that mistake is spelt by wrong diagnosis, or inquire about when spelling is not still still required inquiry by the correction of algorithm by mistake really, these methods can provide incoherent result." initiatively " correction can require user clicks on links so that the inquiry of suggestion is searched for again, may be so also error recovery again.

Table 10 has shown the exemplary search results of the inquiry for example, with two kinds or more explanations or wording of equal value (" Burma and Myanmar ").It is as shown in the table, and embodiments of the invention provide Search Results to having each highly concomitant wording of equal value.For provided example, in nine top Search Results URL for each wording of equal value, there are five to be identical.This has higher similarity far away compared with art methods.

Table 10

User usually comprises in inquiry the keyword that relevant information can not be provided, but search engine has been brought to unnecessary text matches requirement, has caused the Search Results of the lower degree of correlation in art methods.Embodiments of the invention have overcome this defect.

Table 11 has shown the exemplary search results of the inquiry that comprises redundance.It is as shown in the table, and embodiments of the invention allow to ignore unnecessary, the inessential and uncorrelated part in inquiry, and more relevant more succinct inquiry is provided like this.

Table 11

According to one embodiment of the invention, fair play for example, is selected the Search Results of various result products (picture, audio/video, text, image, news item etc.).That is to say, user does not need to specify them finding the result of which kind of type.The result providing can reflect the preference of preference that this user had previously shown or previous isolated user.For example, if thereby inquiry produces the selection that picture search specific image is top score very continually, and this image may be provided as Search Results according to its score.In one embodiment, each top score result of Different Results product need to not be staggered according to score order, returns group but can be expressed as according to product.For this embodiment, the result product of user search need to not understood and identify inquiry with language tool in the time attempting to confirm customer objective.

Table 12 has shown the sample query Search Results for example, being produced by Search Results product combination (comprising picture result).It is as shown in the table, and embodiments of the invention allow Search Results to comprise product combination.

Table 12

Embodiments of the invention can be by regulating Q2P to select the weight of score to change the scope of Search Results to Q2RP.Table 13 has shown the narrower and wider Search Results to inquiry " Stanford ".

Table 13

The first row of table 13 comprises the narrower result that relates to the selection to inquiry.The secondary series of table 13 has shown broad result (for example only having shown the selection of not making in response to the result of original query).The result of secondary series represents to inquire about the relative broad range of theme, and the result of first row helps user to probe into the degree of depth of inquiry theme.In fact, combination narrower and wider result can provide maximally related Search Results.In one embodiment of the invention, create a large amount of combinations with different range, allow user to select " widening result " or " focusing result " link, or change control to adjust combination.

the Q2Q application of example

One embodiment of the present of invention are carried out association to Q2Q and/or Q2Q equivalent combinations USE action message, allow user to obtain the inquiry of the suggestion relevant to its search.Fig. 6 has shown the program that relevant inquiring suggestion is provided according to an embodiment of the invention.Program 600 shown in Fig. 6 starts from operating 605, wherein receives inquiry from user.

Associated at operation 610, Q2Q (and/or Q2Q equivalent combinations) USE action message.Associated Q2Q USE action message directly tends to the result of the remarkable precision that produces inquiry.Associated Q2Q equivalent combinations USE action message tends to produce more various result, although conventionally have the repetition of height.In one embodiment of the invention, to the associated Q2Q equivalent combinations of the original query of Relative Fuzzy USE action message, because this association produces more query suggestion far away conventionally.

The query suggestion providing in response to original query " electronic eavesdropping devices " has according to an embodiment of the invention been provided table 14.The first row of table 14 comprises the directly associated query suggestion based on Q2QUSE action message, and the secondary series of table 14 comprises the associated query suggestion based on Q2Q equivalent combinations (being QPQ) USE action message.

Table 14

In operation 615, provide one or more query suggestion of Q2Q (and/or Q2Q equivalent combinations) the USE action message based on after association to user.In one embodiment of the invention, query suggestion can be expressed in results page.Replaceable or additional, can on the page of query suggestion, provide link.For example, in the time that original query produces the query suggestion of a large amount of height correlations (high score), this embodiment is very practical.In one embodiment of the invention, query suggestion can be classified as accurate (comprising all original search words) and relevant search before statement.

According to one embodiment of the invention, the association of Q2Q (and/or Q2Q equivalent combinations) USE action message is used to produce query spelling bearing calibration initiatively.In query spelling bearing calibration initiatively, user selects the query correction of suggestion to obtain the Search Results based on query correction.

Fig. 6 A has shown the program that query spelling correction suggestion is provided according to an embodiment of the invention.Program 600A shown in Fig. 6 A starts from operating 605A, wherein receives inquiry from user.This inquiry may be the misspellings of the required inquiry of user.

In operation 610A, the operation 610 of Q2Q (and/or Q2Q equivalent combinations) USE action message referral procedure 600 described above and associated.

At operation 611A, Q2Q (and/or Q2Q equivalent combinations) the USE action message based on after association is determined one or more query suggestion.

At operation 612A, one or more query suggestion is confirmed as the spelling correction (inquiry receiving at operation 605A) of original query.According to alternative embodiment of the present invention, the definite of query suggestion is to be affected with various different modes according to received inquiry.For example, for the inquiry previously having observed, in the time there is Q2Q information, the similar inquiry of conceptual dependency and text can be used Q2Q related information and editing distance computational algorithm and be identified.In the case, can use various standards to realize determining as the query suggestion of the spelling correction of original (being received) inquiry.For example, in the time that determined query suggestion is associated with received inquiry, similar and can be confirmed as the spelling correction of received inquiry than the more general determined query suggestion of received inquiry with received query text.In alternative embodiment, can consider definite confidence level of the query suggestion of the spelling correction of more standards using raising as received inquiry.For example, before the inquiry than received, when query suggestion is tended to send after received inquiry more frequently, or when query suggestion is tended to produce more user and select than received inquiry, these standards can improve to be determined as the confidence level aspect the query suggestion of the spelling correction of received inquiry.

When the inquiry receiving at operation 605A is the inquiry previously not observing, there is not Q2Q information.In the case, according to one embodiment of the invention, if suspect that a word in inquiry exists mistake, according to previously whether observing the word of described suspection assessed in one of two ways.

In the time previously observing the word of described suspection, identify other inquiries that occur this word.Keyword based on identical with the inquiry receiving and these inquiries are weighted.Finally, the correction of the suspicious word to previous consideration checks and the correction of offering suggestions for the keyword weight of the frequency of the correction based on previous suggestion and the inquiry of this suspicious word appearance.

In the time previously not observing the word of described suspection, occur that other inquiries of the every other keyword in received inquiry can be identified.In one embodiment of the invention, in the time that inquiry does not meet this standard, can identify the inquiry with received inquiry with the public word of the most unusual (the most distinctive) (low frequency).For each embodiment, the inquiry of identifying can be based on the text similarity of received inquiry and be weighted and identification and received inquiry have the most general inquiry of very high text similarity.

In one embodiment of the invention, in the time that condition does not allow to use the method for above-mentioned spell correction suggestions program, the Q2Q USE action message after association uses together with traditional n-gram-type models coupling.In this embodiment, extract word contact frequency so that known keyword or phrase are determined and jointly followed word from enquiry frequency data.These data and editing distance combine and are used to the definite inquiry as possible spell correction suggestions of unknown word in many word inquiries.

At operation 615A, one or more query suggestion that is defined as the spelling correction of received inquiry is provided for the possible spelling correction of user as received inquiry.In one embodiment of the invention, in the time not having query suggestion to be confirmed as the possible spelling correction of received inquiry, with reference to operation 615 query suggestion that provides as above of the program 600 of figure 6.

the P2Q application of example

According to one embodiment of the invention, can use P2Q or P2Q equivalent combinations and provide about the suggestion inquiry of the particular result page.

Fig. 7 has shown the program of the inquiry of offering suggestions according to an embodiment of the invention.Program 700 shown in Fig. 7 starts from operating 705, wherein receives inquiry and the corresponding Search Results that large result URL is provided.

In operation 710, to associated P2Q (or P2Q equivalent combinations) the USE action message of each result URL.

In operation 715, the inquiry that the P2Q USE action message based on after association is offered suggestions to each result URL.That is to say, provide with Search Results in the suggestion Query List that is closely related of any page.These suggestion in the situation that of fuzzy query may for user provide by URL in one rather than other the focusing meaning realizing.

Table 15 has shown the inquiry exemplary search results page of " rangers " and the correspondence of example suggestion inquiry according to an embodiment of the invention.

Table 15

the P2P application of example

According to one embodiment of the invention, similar to the particular result one or more results that provide in response to received inquiry are used P2P or P2P equivalent combinations and provide.For example, receive inquiry and the assessment result corresponding to this inquiry.Based on described assessment, provide equally similar results.That is to say, for example, can specify some related pages and/or link and can lead and there is the new results page of additional similar results along with result.Common, the similar results of image can produce other images mostly, and the similar results of webpage can produce other webpages mostly, the like.

Fig. 8 shows the program that similar results is provided according to an embodiment of the invention.Program 800 shown in Fig. 8 starts from operating 805, wherein receives inquiry and determines corresponding Search Results.

In operation 810, to associated P2P (or P2P equivalent combinations) the USE action message of corresponding Search Results.

At operation 815, the P2P USE action message based on after association and one or more similar results (being similar to received result) is provided.That is to say, the results list being closely related with received result is provided.Similar results can form search selection, image, news item etc.

The example similar results providing in response to inquiry " phoenix " has according to an embodiment of the invention been provided table 16.

Table 16

the user profile application of example

personalized search

The concept of personalized search is based on following prerequisite, when know about user some information time more relevant Search Results can be provided.Shared in the past at least recommendation of certain user's interest and the user of suggestion and be considered to there is larger value than the user's who does not share its interest and taste recommendation.

Individuation search method of the prior art is identified user's demographics conventionally, but customizes result according to the preference of other members in this demographics group.This method has serious defect, and in a population statistical body, preference changes widely different.Each user belongs to a lot of and the common demographics group that is difficult to mediation conventionally, and user does not provide reliable demographic information conventionally.

According to one embodiment of the invention, each user is the group of thinking people's composition, has unison with respect to other users' class.

Fig. 9 shown according to an embodiment of the invention information is carried out associated so that the program of more relevant Search Results to be provided.Program 900 shown in Fig. 9 starts from operating 905, wherein the unison matrix of user's compute classes at least minimum search activities of each performance.In one embodiment of the invention, calculate the unison matrix of described class and can carry out as follows.First, extract inquiry and the selection of all given user U1.Then, at least N the user U2 repeating with the inquiry of U1 and/or in selecting is identified, and identifies the remaining inquiry of U2 and selection.Then to each inquiry and selection specified weight and the unison mark of compute classes.Described weight can be inversely proportional to this inquiry and/or the common degree being chosen in all users, and by selection/inquiry total after weighting and select/inquiry of total weighting are compared and the unison mark of compute classes.

In operation 910, the Search Results that the new inquiry of U1 is produced is analyzed, and increases the mark of each result and if any previous selection belongs to the user with the unison mark of non-zero class.

In operation 915, based on operation 910 mark adjustment and to result rearrangement and by the expression of results after reordering to user.

An alternative embodiment according to the present invention, with the similar program of program 900 can be for increasing the degree of correlation of search suggestion.

localization search

A kind of important kind of user profile is position.Position has advantage than other demographic informations, because it does not need user to provide itself and no matter who is all to keep real using computing machine.

In one embodiment of the invention, user profile comprises user's geographical location information, and comprises its inquiry and selection by expansion.Described geographical location information can comprise longitude and latitude and city, state name and name of the country.According to an embodiment, these user profile are used to provide Search Results based on user's geographic position.For example, the U.S. user of input inquiry " osu " may represent " Ohio State University (Ohio State University) ", " Oklahoma State University (Oklahoma State University) " or " Oregon State University (Oregon State University) ".In one embodiment, the Search Results providing also with the user-association for example, with similar user profile (similar position).

Figure 10 has shown according to an embodiment of the invention according to the program that more relevant Search Results is provided based on customer location to the position of selecting to be associated.Program 1000 shown in Figure 10 starts from operating 1005, has wherein obtained " " center " and " coverage " that are assigned with latitude-longitude more than the URL of the selection of specified quantity.Center is the position that arrives certain function minimization of the distance summation of the effective user's selection of statistics of each record.For example, in one embodiment, center can represent that URL (having the URL of local tendency) concrete (brick-and-mortar) position in real world is to realize height precision.In the time not there is not local tendency, center has little or at all nonsensical.For example, the URL that does not have a real world particular location can not have nationwide local tendency.

Coverage is to expect that URL, to the highly attractive geographical radius of user, expects not attractive outside this.Radius is less, steeper or more " localization " of attractive force of attractive force decline.The radius of coverage is inversely proportional to certain tolerance of local tendency.For example, highly for example www.canariesbaseball.com of the URL of localization forms center at Sioux Falls, the very compact coverage of SD (being the local of Sioux Falls Canaries).

In operation 1010, coverage adjusts according to statistics.For example, some position generates and more clicks than other position.Adjust original click data to reflect this species diversity.

In operation 1015, ad hoc inquiry is compared to respond with precalculated position data set in user's geographic position, and geographic position adjustment Search Results score based on user.Near those and have the highly mark of the URL of localization and be increased, and the mark of URL is reduced (mark without the URL of very large localization tendency remains unchanged) at a distance.

In operation 1020, user is provided and has the Search Results in the geographic position based on user at least in part.Allow like this user to receive the subjective correlated results in initial some results.For example, compared with prior art, the user of the Wyoming State of input " state tax form (state tax forms) " is more easily provided the state tax form of the Wyoming State.

subscriber's local

As mentioned above, localization can be based on selection, or can be according to alternative embodiment and based on user.In one embodiment of the invention, the localization based on user allows the result of popular inquiry to be cached by compass of competency, and then position-based offers user and without any need for real-time calculating.

Figure 11 has shown according to an embodiment of the invention the program that based on customer location, user is provided more relevant Search Results.Program 1100 shown in Figure 11 starts from operating 1105, wherein determines the compass of competency for the specified portions of the click of given inquiry, starts from the compass of competency of highest level.For example, inquiry " car insurance " produces the mixing of U.S. and Britain's website, and the user who is positioned at Britain more clicks Britain's website, and the user who is positioned at the U.S. more clicks U.S.'s website.

In operation 1110, be increased for the click mark of compass of competency of the assign thresholds part of clicking, and create the independent the results list that is labeled as this compass of competency.

In operation 1115, the click mark of the user outside described compass of competency is reduced, and has produced the general list for the user outside the compass of competency of the assign thresholds part of clicking.In alternative embodiment, create the real-time list that mixes with the adjacency in region in a given jurisdiction based on user.

In operation 1116, if not for the compass of competency of the specific threshold part of clicking, provide result based on user compass of competency in operation 1120.

In operation 1116, if there is the compass of competency for the specific threshold part of clicking, in the circulation again by proceeding to other compass of competency of even lower level of operation 1125 Programs, and arrive thus operation 1105.

In alternative embodiment, by associated with URL position and do not consider inquiry.Have advantages of like this in effective statistical significance, because can assemble more data.For example, consider all users' to the specific state tax website of any query selection source place, then give extra weight for the user in topmost state to particular station and no matter the inquiry that relates to.Potential defect is, if the user of Wyoming State input " Florida state tax form ", if Wyoming State tax website among result, Ze Qi get branch obtains effective and without basis increase.

general issue

Embodiments of the invention provide the method and system of conceptive combination and statement information, and wherein optimum tissue and the statement of information determined in user's response with the associated of statement with information combination.Although describe for some example embodiment above, alternative embodiment of the present invention has a lot of additional application.

The present invention includes various operations.A lot of methods are described with its citation form, but can any method interpolation or deletion action not deviated to base region of the present invention.Operation of the present invention can or can realize by hardware implement in the executable instruction of machine as above.Interchangeable, these steps can be carried out by the combination of hardware and software.The present invention can be provided as computer program, can comprise the machine readable media of having stored instruction, and these instructions can be for programming to carry out according to program as above of the present invention to computing machine (or other electronic equipments).

Figure 12 is that demonstration according to an embodiment of the invention can be for the embodiment of the digital processing system of the server DPS 320 with reference to described in figure 3 or client DPS 305-308.In alternative embodiment of the present invention, disposal system 1201 can be computing machine or Set Top Box, comprises the processor 1203 being connected with bus 1207.In one embodiment, internal memory 1205, storage unit 1211, display controller 1209, communication interface 1213 and i/o controller 1215 are also connected to bus 1207.

Disposal system 1201 is connected to external system by communication interface 1213.Communication interface 1213 can comprise analog modem, integrated services digital network network (ISDN) modulator-demodular unit, cable modem, digital subscriber line (DSL) modulator-demodular unit, T-1 line interface, T-3 line interface, light carrier interface (for example OC-3), token-ring network interface, satellite transmission interface, wave point or for connecting devices to other interfaces of other equipment.Communication interface 1213 can also comprise transceiver or radiophone signal etc.

In one embodiment of the invention, between communication interface 1213 and cloud shape symbol 1230, receive/send signal of communication 1225.In one embodiment of the invention, signal of communication 1225 can be for being connected to another computer system, hub, router etc. by disposal system 1201.In one embodiment of the invention, signal of communication 1225 is machine-readable medium, can pass through circuit, cable, optical fiber or atmosphere etc. transmission.

In one embodiment of the invention, processor 1203 can be traditional microprocessor, such as but not limited to Intel Pentium series processors, Motorola systematic microprocessor etc.Internal memory 1205 can be machine readable media, for example dynamic RAM (DRAM) and can comprise static random-access memory (SRAM).Display controller 1209 is according to traditional approach control display 1219, in one embodiment of the invention, display 1219 can be cathode-ray tube (CRT) (CRT) display, liquid crystal display (LCD), Active Matrix Display, TV monitor etc.Input-output apparatus 1217 is connected to i/o controller 1215, can be keyboard, disc driver, printer, scanner and other input and output devices, comprises mouse, roller ball, touch pad etc.

Storage unit 1211 can comprise machine readable media, such as but not limited to hard disk, floppy disk, CD, smart card or other forms of data storage cell.In one embodiment of the invention, storage unit 1211 can comprise erasable medium, read-only medium, read/writable medium etc.Some data can be passed through direct memory access program and write memory 1205 in the software execute process of computer system 1201.Should be appreciated that software can reside in storage unit 1211, internal memory 1205, or can send or receive by modulator-demodular unit or communication interface 1213.For illustration, term " machine readable media " thereby be considered to be comprises can store data, information or to the instruction sequence any medium that causes processor 1203 to carry out method of the present invention to carry out by processor 1203 of encoding.Term " machine readable media " should include but not limited to solid-state memory, CD and disk, carrier signal etc.

Although described the present invention with reference to some embodiment, it will be understood by those skilled in the art that the present invention is not limited to described embodiment, rather than can implement by the modifications and variations within appended claims essence and scope.Therefore this instructions should be considered to illustrative rather than restrictive.

Claims (9)

1. a method for conceptive tissue and statement information, the method comprises:

Received and record searching engine queries by server count word processing system (DPS), this search engine inquiry receives from multiple isolated users via client DPS;

In response to described search engine inquiry, provide one or more Search Results via described client DPS to described isolated user by described server DPS, wherein each isolated user can be provided by least one Search Results in provided Search Results;

The multiple Search Results that received and recorded described isolated user by described server DPS are selected;

Determined in the Internet activity rank of being selected to carry out during the search sessions of Search Results by isolated user by described server DPS;

In the time that the Internet activity rank of carrying out during described search sessions exceedes activity grade threshold value, by described server DPS, weight is divided and tasked selected Search Results;

By described server DPS, the Search Results of institute's weighting is selected to carry out associated with recorded search engine inquiry;

Receive the described search engine inquiry from user by described server DPS via described client DPS;

Determined the degree of correlation of Search Results and the described search engine inquiry of described weighting by described server DPS, wherein determine that the degree of correlation of the Search Results of described weighting comprises:

Determine that by described server DPS the isolated user quantity of the Search Results of selecting institute weighting and described isolated user check at least one in duration of Search Results of institute's weighting;

The degree of correlation of the residue Search Results in described multiple Search Results that at least one in the duration of being checked by the isolated user quantity of Search Results and the Search Results of institute's weighting of the weighting of described server DPS alternative institute and described isolated user are selected;

Search Results by described server DPS based on institute's weighting and the degree of correlation of described search engine inquiry are determined provides the Search Results of which institute's weighting to isolated user; And

In response to described search engine inquiry, by described server DPS via described client DPS

The Search Results of the institute weighting associated with described search engine inquiry is provided to described isolated user.

2. method according to claim 1, wherein selects multiple Search Results by described isolated user, and the method also comprises:

Determined the degree of correlation of selected Search Results and described search engine inquiry by described server DPS; And

The degree of correlation by described server DPS based on selected Search Results and described search engine inquiry and determine provide which selected Search Results to described isolated user.

3. method according to claim 2, wherein determine that the degree of correlation of selected Search Results and described search engine inquiry comprises:

Determine and select the isolated user quantity of described Search Results and described isolated user to check at least one in the duration of selected Search Results by described server DPS; And

By described server DPS, isolated user selected to quantity and compare with the degree of correlation of residue Search Results in multiple Search Results of being selected by described isolated user at least one of duration of described Search Results.

4. method according to claim 1, the method also comprises:

Determined the characteristic of described search engine inquiry by described server DPS; And

By described server DPS, the Search Results of institute's weighting is selected to be associated with the characteristic of determined described search engine inquiry.

5. method according to claim 4, the method also comprises:

To offer described isolated user with the search-engine results of one or more institute weighting of the association of characteristics of determined described search inquiry by described server DPS.

6. method according to claim 4, wherein said search engine inquiry comprises that the characteristic of one or more words and described search engine inquiry comprises at least one in the wording of equal value of the synonym of word in the spelling of the word in search engine inquiry or common mistake spelling, search engine inquiry and the word of search engine inquiry.

7. method according to claim 4, the word in wherein said search engine inquiry has different meanings, and the method also comprises:

By described server DPS, the Search Results of institute's weighting of the more popular meaning based on institute's predicate is offered to described isolated user.

8. method according to claim 1, the Search Results of wherein said institute weighting is URL.

9. a device for conceptive tissue and statement information, this device comprises:

First device, for being received and record searching engine queries by server count word processing system (DPS), this search engine inquiry receives from multiple isolated users via client DPS;

The second device, be used in response to described search engine inquiry, provide one or more Search Results via described client DPS to described isolated user by described server DPS, wherein each isolated user can be provided by least one Search Results in provided Search Results;

The 3rd device, selects for the multiple Search Results that received and recorded described isolated user by described server DPS;

The 4th device, for being determined by described server DPS in the Internet activity rank of being selected to carry out during the search sessions of Search Results by isolated user;

The 5th device, in the time that the Internet activity rank of carrying out during described search sessions exceedes activity grade threshold value, is divided weight to task selected Search Results by described server DPS;

The 6th device, for being selected to carry out associated with recorded search engine inquiry to the Search Results of institute's weighting by described server DPS;

The 7th device, for receiving the described search engine inquiry from user by described server DPS via described client DPS;

The 8th device, for determined the degree of correlation of Search Results and the described search engine inquiry of described weighting by described server DPS, wherein determine that the degree of correlation of the Search Results of described weighting comprises:

Determine that by described server DPS the isolated user quantity of the Search Results of selecting institute weighting and described isolated user check at least one in duration of Search Results of institute's weighting;

The degree of correlation of the residue Search Results in described multiple Search Results that at least one in the duration of being checked by the isolated user quantity of Search Results and the Search Results of institute's weighting of the weighting of described server DPS alternative institute and described isolated user are selected;

Search Results by described server DPS based on institute's weighting and the degree of correlation of described search engine inquiry are determined provides the Search Results of which institute's weighting to isolated user; And

In response to described search engine inquiry, provided the Search Results of the institute weighting associated with described search engine inquiry to described isolated user via described client DPS by described server DPS.

Images