CN102930041A - Retrieval result real-time updating method based on user behavior information and system thereof - Google Patents
Retrieval result real-time updating method based on user behavior information and system thereof Download PDFInfo
- Publication number
- CN102930041A CN102930041A CN2012104534649A CN201210453464A CN102930041A CN 102930041 A CN102930041 A CN 102930041A CN 2012104534649 A CN2012104534649 A CN 2012104534649A CN 201210453464 A CN201210453464 A CN 201210453464A CN 102930041 A CN102930041 A CN 102930041A
- Authority
- CN
- China
- Prior art keywords
- subclauses
- clauses
- retrieval
- result
- user
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a retrieval result real-time updating method based on user behavior information and a system of the retrieval result real-time updating method based on the user behavior information. The retrieval result real-time updating method based on the user behavior information and the system of the retrieval result real-time updating method based on the user behavior information comprise achieving retrieval results which comprise at least one entry, achieving the user behavior information, choosing a processing mode based on the user behavior information, processing the entries in the retrieval results according to the similarity degree of the entries and obtaining and showing the updated retrieval results. Based on the user behavior information, the retrieval result real-time updating method and the system of the retrieval result real-time updating method can improve the retrieval results in real time. The entries which are ranked front are always entries which are needed by users and not visited, and therefore the efficiency that the users check the retrieval results and the user retrieval experience are improved.
Description
Technical field
The present invention relates to the information search technique field, relate in particular to a kind of result for retrieval real time updating method and system thereof based on user behavior information.
Background technology
Along with the fast development of computing machine and infotech, information automatic by computing machine or that assisted user generates is more and more, how to retrieve customizing messages and become thus very important in magnanimity information.In order to solve problems, various computer information retrieval technology are arisen at the historic moment, and comprise computer documents searching system, network search engines, online database on-line retrieval system etc.The computer user utilizes these systems to search information needed by keying in keyword, and this type systematic has solved the difficult problem of user search information to a great extent, but the ability that shortage is accurately retrieved and good user experience.In recent years machine learning techniques development is rapid, and it can utilize the people's behavior of artificial intelligence correlation technique natural imitation and form of thinking, user behavior is inferred make the calculating function reach the interchange purpose.Yet this type of technology not yet is widely used in improving user's experience of information retrieval system.There is not yet in the prior art based on the user and experience and the user behavior information such as feedback result for retrieval is recombinated and the technology of dynamic real-time update.
Summary of the invention
The present invention proposes a kind of result for retrieval real time updating method based on user behavior information, comprising:
Step 1: obtain the result for retrieval that comprises at least one clauses and subclauses;
Step 2: obtain user behavior information;
Step 3: based on described user behavior Information Selection tupe, and according to the clauses and subclauses similarity, the clauses and subclauses in the described result for retrieval are processed;
Step 4: obtain and display update after result for retrieval.
The similarity of each clauses and subclauses " contrast " are between each clauses and subclauses, carry out the comparison of similarity between the content of each clauses and subclauses.The algorithm of similarity comparison comprises the algorithm of the Jaccard index that calculates two set etc.Similarity is for the similarity degree between the content of each clauses and subclauses of expression or each clauses and subclauses.If the similarity value is higher, then represents in the content of these two clauses and subclauses or clauses and subclauses and exist identical or akin information more.
Wherein, the generation step that comprises the result for retrieval of at least one clauses and subclauses described in the described step 1 comprises:
Steps A 1: obtain keyword, described keyword is carried out pre-service;
Steps A 2: judge whether described pretreated keyword meets the requirements; If satisfactory, then retrieve the generation result for retrieval according to described keyword; If undesirable, then re-execute described steps A 1, A2, until generate result for retrieval.
Wherein, in the described step 1, according to weights described clauses and subclauses are sorted from high to low.
Wherein, in the described step 1, clauses and subclauses weights are identical according to described clauses and subclauses citation times sort from high to low.
Among the present invention, described user behavior information refers to that the user can produce different interest or concern, for example for the every clauses and subclauses that comprise in the described result for retrieval, may access or click certain/a little clauses and subclauses, may selective access certain/a little clauses and subclauses, may skip certain/a little clauses and subclauses, different in size to the concern time of each clauses and subclauses, etc., because of different demands or reaction, the different behaviors that the user makes, thus form corresponding user behavior information.
Described user behavior information comprises the physiology sign information when the historical accesses entry of user, user ignore clauses and subclauses, user and access length reading time of the content-length of the time interval of different clauses and subclauses, the historical accesses entry of user, the historical accesses entry of user, user's accesses entry.Also comprise user's possible other reaction informations when accesses entry.
The historical accesses entry of user refers to the user in browsing the process of result for retrieval, clicks to enter a certain clauses and subclauses and further access, and then these clauses and subclauses be user's history accesses entry.
The user ignores clauses and subclauses, refers to that the user browses in the process of result for retrieval at jumping characteristic, still exists some clauses and subclauses not accessed before a certain clauses and subclauses of access, and then these clauses and subclauses are ignored clauses and subclauses for the user.
The user accesses the time interval of different clauses and subclauses, refers to that the user clicks the time interval that different clauses and subclauses are further accessed.
The content-length of the historical accesses entry of user refers to the further quantity of information of the content of these clauses and subclauses of access of user, comprises the number of words in these clauses and subclauses, the information such as time of video playback.
Length reading time of the historical accesses entry of user refers to that the user clicks clauses and subclauses of rear further access until access the complete time that the result for retrieval page spends of returning.
Physiology sign information during user's accesses entry refers to various physiology or limbs characteristic parameter and the change information thereof of user when reading clauses and subclauses that the user behavior information acquisition device captures.For example, facial expression, eye movement, limbs feature, changes in heart rate, respiratory variations or other physiology sign information applicatory of when reading clauses and subclauses, producing of user.
The present invention further comprises step 5: namely, repeat described step 2 to step 4, until stop when stopping to obtain described user behavior information.
Among the present invention, described step 3, the tupe of the every clauses and subclauses that comprise based on the described result for retrieval of described user behavior Information Selection comprises: hide historical accesses entry, similar historical accesses entry ordering or similarly ignore the clauses and subclauses ordering.Further, the present invention can use any one of above-mentioned three kinds of patterns to process, or uses any multinomial combination wherein to process, and for example, the clauses and subclauses of having accessed is hidden, in the result for retrieval after making it not be presented at renewal.For example, the similar clauses and subclauses of having ignored by descending sort, no longer are presented in the result for retrieval after the renewal.For example, similar historical accesses entry is arranged in the result for retrieval be presented at after the renewal by ascending order.
Wherein, described hiding historical accesses entry may further comprise the steps:
Step R1: the clauses and subclauses of choosing the user to access based on described user behavior information;
Step B2: the clauses and subclauses that described user has been accessed shift out from described result for retrieval;
Step B3: described clauses and subclauses of having accessed are deposited in the historical accesses entry set.
Among the present invention, historical accesses entry set refers to be comprised of the clauses and subclauses of having accessed.Historical accesses entry set is stored in the described entry process device.
Wherein, described similar historical accesses entry ordering may further comprise the steps:
Step C1: the clauses and subclauses of choosing the user to access based on described user behavior information;
Step C2: described historical accesses entry is deposited in the described historical accesses entry set;
Step C3: the clauses and subclauses in the described historical accesses entry set are carried out the similarity contrast, obtain the similar content between the described clauses and subclauses;
Step C4: according to described similar content each clauses and subclauses in the described result for retrieval are carried out similarity contrast, generate the similarity value of described each clauses and subclauses and described similar content;
Step C5: from high to low each clauses and subclauses in the described result for retrieval are sorted according to described similarity value.
Similar content between the described clauses and subclauses refers to the highest information of similarity between the historical accesses entry.For example, a certain vocabulary ABC all appears in the clip Text of historical accesses entry, calculate this vocabulary of rear identification ABC as the highest content of similarity in the historical accesses entry according to the similarity compare device, then this vocabulary ABC is as the similar content between each historical accesses entry.
According to described similar content each clauses and subclauses in the described result for retrieval are carried out the similarity contrast, each clauses and subclauses in the calculating result for retrieval and the similarity degree of this similar content.The similarity value is higher, and namely similarity degree is higher, shows that the clauses and subclauses in the described result for retrieval are more similar to the historical accesses entry of user.Further, the entry process device sorts to the clauses and subclauses in the described result for retrieval according to this similarity value, makes the ordering of the interested clauses and subclauses of user forward.
Wherein, describedly similarly ignore clauses and subclauses orderings and may further comprise the steps:
Step D1: the clauses and subclauses of choosing the user in access, to ignore based on described user behavior information;
Step D2: described user's the clauses and subclauses of ignoring are deposited in history and ignore in the entry set;
Step D3: each clauses and subclauses that each clauses and subclauses in the described result for retrieval and described history are ignored in the entry set are carried out the similarity contrast, obtain the similarity weights of each clauses and subclauses in the described result for retrieval;
Step D4: from low to high each clauses and subclauses in the described result for retrieval are sorted according to described similarity weights.
Among the present invention, history is ignored entry set and is referred to be comprised of the clauses and subclauses of having ignored.History is ignored entry set and is stored in the described entry process device.
Wherein, further comprise: calculate the similarity value of clauses and subclauses and described similar content in the described historical accesses entry set, each clauses and subclauses during described historical accesses entry is gathered sort from high to low according to described similarity value.
The invention allows for a kind of result for retrieval real-time update system based on user behavior information, comprising:
The user behavior information acquisition device, it obtains user behavior information;
Similarity compare device, it comprises the functional module of calculating similarity;
The entry process device, it is connected with described user behavior information acquisition device and similarity compare device, for the described user behavior Information Selection pattern of obtaining according to described user behavior information acquisition device, and according to the similarity comparing result of described similarity compare device for described clauses and subclauses, process the clauses and subclauses in the described result for retrieval;
Display device, it is connected with described entry process device, and reception also shows the clauses and subclauses that sent by described entry process device.
In the result for retrieval real-time update of the present invention system, further comprise:
Database, it stores magnanimity information;
Indexing unit, it is connected with described database and described entry process device, is used for generating described result for retrieval according to the described magnanimity information of keyword retrieval.
Further, described indexing unit generates the weights of the matching degree of each clauses and subclauses and described keyword in the described result for retrieval.
Wherein, the user behavior information acquisition device comprises mouse, keyboard, image acquisition equipment, built-in timing device, infrared induction equipment, GPS, the sense of touch sensing apparatus of computer system.
The present invention according to the user with searching system reciprocal process in the collateral information that produces dynamically update and the mechanism of the Search Results of recombinating, the user who improves Machine Retrieval System experiences.The present invention is based on user behavior information and improve in real time result for retrieval, the forward clauses and subclauses that sort are always the user to be needed and not accessed clauses and subclauses, make the user more promptly retrieval and inquisition to its information needed, improve the user and checked the efficient of result for retrieval, thereby be embodied as the purpose that the user provides fast accurate retrieval service.The present invention with the process of user interactions in the iterate improvement result of retrieving each time progressively, what the assurance user at first saw at every turn all is the as a result clauses and subclauses of not reading, the as a result clauses and subclauses that are arranged in the result for retrieval prostatitis all are that the net result of wanting to the user is the most similar, and when the user returns when looking into historical accesses entry, the most forward clauses and subclauses of ordering are exactly that the user thinks back the clauses and subclauses looked into most, have improved user search experience.
The present invention does not need the direct access inquiry customer problem, but infers user's preference situation by the interbehavior information that indirectly reads user and system.For example, when the user used system retrieval information of the present invention, system can return the correlated results clauses and subclauses and show each as a result general introduction content of clauses and subclauses, and the user can check the as a result complete content of clauses and subclauses according to these property summarized content choice.For example, (establishing result for retrieval is RS when the user has consulted the clauses and subclauses of some result for retrieval according to content summary information, ResultSet) after, native system extracts them automatically from these clauses and subclauses analog information (is designated as CI, be similar content), and utilize this analog information CI again retrieval (but also the connection data storehouse is retrieved again) in result for retrieval, dynamically update result for retrieval RS content and RS discal patch purpose rank.
For another example, the clauses and subclauses that the detection user of system has accessed and the clauses and subclauses statistical information of not accessing infer the uninterested aspect of user the rank of this type of classification in the dynamic reducing result for retrieval.For example: suppose that the user has accessed the clauses and subclauses R1 among the RS, R2, R4 and R5, then system infers that this user loses interest in to clauses and subclauses R3 and result similarly, further with clauses and subclauses dynamic reducing rank similar to R3 among the RS, to reduce such clauses and subclauses to user's interference.
The present invention hides the historical accesses entry of user automatically, and for example, when the user has accessed certain bar when again turning back to the result for retrieval page after the clauses and subclauses as a result, system of the present invention hides the clauses and subclauses that the user had accessed automatically.What see when the user turns back to the result for retrieval page at every turn like this all is the as a result clauses and subclauses of not accessing, thereby has reduced the interference of repeated and redundant information to the user.Namely, the clauses and subclauses that the present invention will access are transferred in the clauses and subclauses of having accessed from the clauses and subclauses of result for retrieval, keep the user when checking result for retrieval, to avoid repeated accesses to the clauses and subclauses of having accessed, the user can be checked all the time without the clauses and subclauses of accessing, thereby reduced the interference of repeated and redundant information to the user.
The historical accesses entry auto-sequencing of the present invention to hiding for example, after historical accesses entry is hidden automatically, the invention provides interface so that the user checks historical access.And, when launching the content of historical accesses entry each time, according to user's interactive information historical accesses entry is sorted, so that the user can find the clauses and subclauses of oneself paying close attention to rapidly from historical visit information.For example, carry out the similarity contrast in the clauses and subclauses that the present invention has accessed to the user, choose the content that similar content that wherein similarity is the highest is most interested in as the user.By according to this similar content the clauses and subclauses in the result for retrieval being carried out similarity ordering from high to low, make the ordering of the clauses and subclauses that comprise the content that the user is most interested in the result for retrieval forward, be convenient to the user and view sooner interested clauses and subclauses, improved the efficient that the user checks clauses and subclauses, improved user's retrieval and experienced.
The present invention is classified as history to user's uncared-for clauses and subclauses when the access of selectivity or jumping characteristic and ignores entry set, each clauses and subclauses in the result for retrieval and historical each clauses and subclauses that are left in the basket of ignoring in the entry set are carried out the similarity contrast, from low to high each clauses and subclauses in the result for retrieval are sorted according to the similarity weights that obtain, after the ordering of the uninterested content of user is leaned on, improved the efficient that the user checks clauses and subclauses, improved user's retrieval and experienced.
Description of drawings
Fig. 1 represents to the present invention is based on the process flow diagram of the result for retrieval real time updating method of user behavior information.
Fig. 2 represents to the present invention is based on the detail flowchart of the result for retrieval real time updating method of user behavior information.
Fig. 3 represents to the present invention is based on the structural drawing of the result for retrieval real-time update system of user behavior information.
Fig. 4 represents that clauses and subclauses are transferred to the synoptic diagram that historical accesses entry is gathered in the embodiment of the invention.
Fig. 5 represents the detail flowchart of the similar historical accesses entry sequencing model of the present invention.
Fig. 6 represents in the embodiment of the invention synoptic diagram based on the entry process of user behavior information.
Fig. 7 represents that the present invention hides the process flow diagram of historical accesses entry pattern.
Fig. 8 represents the process flow diagram of the similar historical accesses entry sequencing model of the present invention.
Fig. 9 represents the similar process flow diagram of ignoring the clauses and subclauses sequencing model of the present invention.
Embodiment
In conjunction with following specific embodiments and the drawings, the present invention is described in further detail.Implement process of the present invention, condition, experimental technique etc., except the following content of mentioning specially, be universal knowledege and the common practise of this area, the present invention is not particularly limited content.
Such as Fig. 1 to Fig. 9,1-database, 2-indexing unit, 3-user behavior information acquisition device, 4-entry process device, 5-similarity compare device, 6-display device.
Such as Fig. 1 and shown in Figure 2, the present invention is based on the result for retrieval real time updating method of user behavior information, comprising:
Step 1: obtain the result for retrieval that comprises at least one clauses and subclauses, this result for retrieval is by display device 6 demonstrations and show the user.
The user inputs keyword.Obtain the keyword of user's input and this keyword is carried out pre-service by indexing unit 2.The pre-service of keyword refers to the pruning, fractionation, synthetic etc. to keyword, analyzes and extract the retrieval that core in the keyword is used for database 1.
After the keyword pre-service is complete, further whether this keyword is met retrieval and require to judge.Retrieval requires to generally include the length requirement of keyword, the sensitive information examination requirement that keyword relates to etc.
Retrieval requires when keyword does not meet, and then prompting user re-enters keyword retrieval requires or the user withdraws from retrieval until keyword meets.
If keyword meets the retrieval requirement, then indexing unit 2 utilizes this keyword to retrieve until generate corresponding result for retrieval in the magnanimity information of database 1 storage, this result for retrieval is transferred to entry process device 4, entry process device 4 obtains this result for retrieval and preserves, and comprises one or more clauses and subclauses in this result for retrieval.Entry process device 4 transfers to display device 6 with clauses and subclauses to be shown, shows this result for retrieval and the every clauses and subclauses that comprise thereof by display device 6.
Among the present invention, preferably, the every clauses and subclauses in 4 pairs of result for retrieval that obtain of entry process device sort according to its weights (weights refer to the matching degree of keyword and each clauses and subclauses), for example, from high to low each clauses and subclauses are sorted by weights.
Further preferably, when a plurality of clauses and subclauses that comprise in the result for retrieval have identical weights, the situation that a plurality of clauses and subclauses identical with the keyword matching degree namely occur, the clauses and subclauses that then number of times that is cited according to each clauses and subclauses of entry process device 4 is identical with these weights are minor sort again, for example, from high to low each clauses and subclauses is further sorted according to citation times.
Step 2: obtain user behavior information.
The user checks the result for retrieval that the above-mentioned steps one of display device 6 displayings obtains.The user operates by common equipments such as mouse, keyboard, touch-screens according to the difference of actual conditions, and each relevant entry of clicking, choose in this result for retrieval is further accessed.User behavior information acquisition device 3 is by monitoring user operation, judge the user access certain/a little clauses and subclauses or ignore certain/a little clauses and subclauses, thereby obtain user behavior information, this user behavior message reflection user to each clauses and subclauses in various degree concern or ignore.
User behavior information comprises: the physiology sign information when the historical accesses entry of user, user ignore clauses and subclauses, user and access length reading time of the content-length of the time interval of different clauses and subclauses, the historical accesses entry of user, the historical accesses entry of user, user's accesses entry.Also comprise user's possible other reaction informations when accesses entry.For example, user behavior information acquisition device 3 monitors the user and has clicked the further access of a certain clauses and subclauses do, and then user behavior information acquisition device 3 judges that these clauses and subclauses are the historical accesses entry of user.
For example, the user is when the clauses and subclauses of access result for retrieval, and possible jumping characteristic accessing is some clauses and subclauses wherein, and has ignored other clauses and subclauses.When the user clicked a certain clauses and subclauses, if there are not accessed clauses and subclauses before these clauses and subclauses, user behavior information acquisition device 3 judged that these not accessed clauses and subclauses ignore clauses and subclauses as the user.
For example, user behavior information acquisition device 3 can be inferred the actual interested content of user by detecting user's institute's time spent time of accessing different clauses and subclauses and energy.Further, infer the precision of assessment user preferences in order to improve system of the present invention, access the time span of this entry contents of reading divided by the metric that content-length was obtained of these clauses and subclauses, as the assessment of the user being accessed certain bar result items institute's time spent and energy with this user.For example, when user behavior information acquisition device 3 detects the user at certain as a result during unusual high of the energy that spends of clauses and subclauses, update system infers that this user is to these clauses and subclauses and interested with the similar information of these clauses and subclauses, and in the step of back, promote the rank of this type of information, so that this user obtains relevant analog information as early as possible; Otherwise, when user behavior information acquisition device 3 detects the user at certain as a result during unusual low of the energy that spends of clauses and subclauses, update system infers that this user loses interest in to these clauses and subclauses and with the similar information of these clauses and subclauses, and in the step of back, reduce the rank of this type of information, obtain as early as possible the probability of this type of information to reduce this user.
For example, user behavior information acquisition device 3, judge roughly that by the historical reading rate statistical information that records a certain individual consumer this user reads the scheduled time of certain clauses and subclauses, in reading process, be subject to other interference and cause elongated phenomenon reading time to distinguish the user.
Preferably, for more accurate deduction user preferences, user behavior information acquisition device 3 can also further be judged by user's facial expression information of obtaining when reading clauses and subclauses.Preferably, the present invention arranges image acquisition equipment, catch the user read certain/real-time facial expression during a little clauses and subclauses.Further, the facial expression analysis module is set, the real-time expression of analysis user is set up the expression parameter of each clauses and subclauses, by other behavioral parameters in conjunction with the user, catches user behavior information, to determine the user preferences degree.For example, in user's agreement situation, can directly pass through the image-capturing apparatus of the mode invoke user computing machine of browser plug-in.
Further, can also obtain eye movement variation, limbs changing features, changes in heart rate, respiratory variations or other physiology sign information applicatory that the user occurs when reading clauses and subclauses.
Step 3: based on user behavior Information Selection tupe, and according to the clauses and subclauses similarity, the clauses and subclauses in the result for retrieval are processed.
Among the present invention, the tupe of the every clauses and subclauses that result for retrieval comprised based on user behavior information comprises: hide historical accesses entry, similar historical accesses entry ordering, the similar clauses and subclauses ordering isotype of ignoring.The present invention can adopt any one in above-mentioned three kinds of patterns, also available any two kinds or any two or more mode combinations.
What Fig. 7 showed is the process flow diagram that the present invention hides historical accesses entry pattern.After carrying out hiding historical accesses entry pattern, when after the user accesses a certain clauses and subclauses, again turning back to the result for retrieval page, the clauses and subclauses that the user had accessed are hidden automatically, on the page of result for retrieval tabulation, then no longer show the clauses and subclauses that this user had just accessed, that is whole clauses and subclauses that, the user sees after the page that returns the result for retrieval tabulation must be that this user did not access.Wherein, user's clauses and subclauses of having accessed are transferred to from the content of result for retrieval in the historical accesses entry set.Historical accesses entry set is stored in the entry process device 4.
Further, after historical accesses entry is automatically hidden, hide the record of historical access if the user need to check institute, the invention provides and show history reading clauses and subclauses function again, the user can consult all historical visit informations.Further, from historical visit information, find out fast for the convenience of the user the concern clauses and subclauses, the present invention can sort to historical accesses entry according to user behavior information when launching historical accesses entry, for example, the information such as summary by each the historical accesses entry in 5 pairs of historical accesses entry set of similarity compare device are carried out the similarity contrast, extract in each clauses and subclauses summary the content of the most normal appearance as similar content, for example, maximum contents to occur as similar content.Carry out from high to low ordering according to each clauses and subclauses and the similarity of similar content again, make the ordering of the clauses and subclauses that comprise the content that the user is most interested in the historical accesses entry set forward.Further, the present invention also can sort to each clauses and subclauses the accessed time of clauses and subclauses from the near to the remote.
What Fig. 8 showed is the process flow diagram of the similar historical accesses entry sequencing model of the present invention.Similar historical accesses entry sequencing model refers to according to historical accesses entry set the clauses and subclauses in the result for retrieval be sorted.
The clauses and subclauses that entry process device 4 automatic preservation users have accessed in current sessions are included in the historical accesses entry set (being designated as VisitedItemSet).For example, as shown in Figure 4, when user behavior information acquisition device 3 detected the user and accessed a certain clauses and subclauses A in the result for retrieval (being designated as ResultSet) or subset of items A, these clauses and subclauses A or subset of items A can be transferred among the historical accesses entry set VisitedItemSet automatically.VisitedItemSet is user's formed entry set in the process of access result for retrieval, automatically is kept in the entry process device 4.
Calculated the similarity (being designated as Similarity) of each historical accesses entry among the VisitedItemSet by similarity compare device 5.Each clauses and subclauses similarity relatively can calculate the strongest similar content of similarity according to contents such as the summary of each clauses and subclauses or interior perhaps keywords.Similarity compare device 5 carries out similarity contrast to each clauses and subclauses in the result for retrieval again according to this similar content, draws the similarity value of each clauses and subclauses and this similar content in the result for retrieval, according to the ordering of each clauses and subclauses among this similarity value renewal ResultSet.
Upgrading among the ResultSet each clauses and subclauses ordering can User need different and arranges, and for example, can sort from high to low according to the similarity value, or sorts from low to high according to the similarity value, also can condition be set according to other and sort.
Preferably, the similarity value according to the clauses and subclauses similarity sorts from high to low in the result for retrieval (ResultSet after the renewal) after upgrading among the present invention.
Preferably, the historical accesses entry of each in the historical accesses entry set (accesses entry set) makes the user can find rapidly the clauses and subclauses of oneself paying close attention to most from historical information according to sorting from high to low to the similarity value of similar content.Each clauses and subclauses after upgrading among the ResultSet of rearrangement do not show the user at once, but wait until when the user gets back to the result for retrieval page and showed to the user by display device 6.
Such as Fig. 5, preferred embodiment of the present invention is concrete adopts following mode to realize:
Wherein, used an outside algorithmic procedure in the said process: upgrade the updateSimilarity algorithm of similarity of clauses and subclauses and similarity set, this algorithm can be general similarity algorithm, such as the algorithm of the Jaccard index that calculates two set:
What Fig. 9 showed is the similar process flow diagram of ignoring the clauses and subclauses sequencing model of the present invention.The similar clauses and subclauses sequencing model of ignoring is further to sort according to the clauses and subclauses of clauses and subclauses to result for retrieval that are left in the basket.
Entry process device 4 can judge in the following way that the user loses interest in to a certain clauses and subclauses: when the every clauses and subclauses in user's jumping characteristic/selective access result for retrieval tabulation, if there are not accessed clauses and subclauses before the accessed clauses and subclauses, then think the not accessed clauses and subclauses that are skipped and with it similarly clauses and subclauses belong to uncared-for clauses and subclauses.For example, as shown in Figure 6, comprise five clauses and subclauses of R1 to R5 in the result for retrieval, the user has accessed the clauses and subclauses R1 in the result for retrieval, R2, and R4 and R5, then clauses and subclauses R3 is the clauses and subclauses that are left in the basket.Entry process device 4 infers that this user loses interest in to the clauses and subclauses of clauses and subclauses R3 and result for retrieval similarly, further, the clauses and subclauses similar to R3 among the result for retrieval clauses and subclauses RS are carried out the dynamic reducing rank, to reduce such clauses and subclauses (clauses and subclauses that R3 is similar) to user's interference.In the present embodiment, entry process device 4 is automatically preserved users' uncared-for clauses and subclauses in current sessions and is included into history and ignores in the entry set and (be designated as IgnoredItemSet).History is ignored entry set and is stored in the entry process device 4.When the user had ignored a certain clauses and subclauses among the result for retrieval ResultSet or subset of items, these clauses and subclauses or subset can be added among the IgnoredItemSet automatically.Similarity compare device 5 calculates among the result for retrieval ResultSet similarity weights SimilarityWeight of each clauses and subclauses in each clauses and subclauses and IgnoredItemSet and according to the ordering of each clauses and subclauses among these similarity weights SimilarityWeight renewal ResultSet.For example, the clauses and subclauses of the result for retrieval after the renewal sort from low to high according to SimilarityWeight.
The similarity weights carry out the similarity contrast by similarity compare device 5 with clauses and subclauses in the result for retrieval and the historical clauses and subclauses of ignoring of ignoring entry set, and draw these clauses and subclauses and all have ignored the similarity value of clauses and subclauses.The similarity value of these clauses and subclauses is weighted and computing, draws the similarity weights of these clauses and subclauses.After all clauses and subclauses, calculate the similarity weights of all clauses and subclauses in similarity compare device's 5 traversal result for retrieval.The concrete formula of similarity weights is: SimilarityWeight=Similarity_1*Weight (Similarity_1)+Similarity_2*Weight (Similarity_2)+... + Similarity_n*Weight (Similarity_n), wherein, Weight is the Similarity-Weighted saturation of system definition, can determine the weighting factor of different similarities by using this function, and the higher weighting factor of similarity is higher, for example, the Weight function definition in the present embodiment is:
Weight(Similarity)；
RETURN?Similarity
This function is so that similarity has identical with it weighting factor, and the higher weighting factor of similarity is higher.This moment, the weights formula namely was derived as the quadratic sum of similarity, i.e. SimilarityWeight=Similarity_1^2+Similarity_2^2+ ... + Similarity_n^2.Illustrate as follows, if certain clauses and subclauses (being designated as I1) are respectively 1 with IgnoredItemSet (supposing to comprise three clauses and subclauses) discal patch purpose similarity, 0.1 and 0.2, the weighting factor of these three similarities is respectively 1 so, 0.1 with 0.2, then the similarity weights of I and IgnoredItemSet equal 1^2+0.1^2+0.2^2=1.05; Establish again certain clauses and subclauses (being designated as I2) and be respectively 0.5 with IgnoredItemSet (supposing to comprise three clauses and subclauses) discal patch purpose similarity, 0.6 and 0.1, the weighting factor of these three similarities is respectively 0.5 so, 0.6 with 0.1, then the similarity weights of I2 and IgnoredItemSet equal 0.5^2+0.6^2+0.2^2=0.65.Because 1.05 greater than 0.65, so I1 is more similar to clauses and subclauses among the IgnoredItemSet than I2, after I1 just is arranged in I2 in next step renewal process.
The result for retrieval that display device 6 will not be upgraded at once shows the user, but waits until that the user shows when getting back to the result for retrieval page.In the preferred embodiment of the present invention, similarity compare device's 5 implementation procedures of the present invention are similar to the implementation procedure of above-mentioned similar historical accesses entry sequencing model, difference is to detect after the user loses interest in for a certain clauses and subclauses when entry process device 4, then more new historical is ignored entry set, and reduces in the result for retrieval and ignore the rank of the similar clauses and subclauses of clauses and subclauses to this.
Step 4: when entry process device 4 judges that the user upgrades result for retrieval, the result for retrieval behind the display device display update.
When user behavior information acquisition device 3 monitors the user following behavior occurs, entry process device 4 will upgrade result for retrieval:
The user inputs the search key behavior: after the user inputs search key, and the keyword retrieval database 1 that indexing unit 2 provides according to the user, and provide result for retrieval for the user.When the user re-entered keyword at existing result for retrieval interface, the result for retrieval after indexing unit 2 uses new keyword retrieval database 1 and renewal is provided was to entry process device 4.
The behavior of user's accesses entry: entry process device 4 recording users have been accessed and have been consulted clauses and subclauses, and infer according to this user to the hobby of some/a certain class clauses and subclauses, thereby with resequencing with the same or analogous content of such clauses and subclauses in the result for retrieval, to promote rank.
The user is back to the behavior of whole result for retrieval from single clauses and subclauses: when user's end the access of some concrete clauses and subclauses is read, when returning the result for retrieval page, the result for retrieval after entry process device 4 will upgrade shows the user by display device 6.
The user ignores the behavior of some or a certain class clauses and subclauses: entry process device 4 recording users have been accessed the statistical information of consulting clauses and subclauses, and infer according to this user to the detest of a certain class clauses and subclauses or ignore, thereby with resequencing with the same or analogous content of such clauses and subclauses in the result for retrieval, its rank is reduced.
Preferably, repeated execution of steps two is to step 4 after execution of step four, user behavior information acquisition device 3 constantly obtains user behavior information, browse in the process of result for retrieval the user result for retrieval is realized implementing to optimize, improve user's retrieval and experience, until the user withdraws from and browses or user behavior information acquisition device 3 stops when stopping to obtain user behavior information.
What Fig. 3 showed is the structural representation that the present invention is based on the result for retrieval real-time update system of user behavior information.System of the present invention comprises user behavior information acquisition device 3, entry process device 4, similarity compare device 5 and display device 6.Further comprise database 1, and indexing unit 2.
User behavior information acquisition device 3 is used for catching the user in access and the behavior of checking result for retrieval.By the monitor user ' behavior to obtain the user for the relevant information of the interest-degree of clauses and subclauses height, a foundation when processing clauses and subclauses as entry process device 4.User behavior information acquisition device 3 comprises the input media that monitoring is common, for example, and mouse, keyboard, touch screen etc.User behavior information acquisition device 3 further comprises image acquiring device, for example, countenance information when being used for obtaining the user and checking the result for retrieval clauses and subclauses, judging whether the user is interested in the clauses and subclauses of checking, and the user checks the information such as time that clauses and subclauses spend in conjunction with user's expression information data.For example, obtain the changes in heart rate of user when consulting clauses and subclauses, to judge the user to the concern of this accesses entry or to ignore situation.
5 pairs of each clauses and subclauses of similarity compare device are carried out the similarity contrast and are obtained similarity value or similarity weights, a parameter when processing clauses and subclauses as entry process device 4.Have similarity contrast algorithm among the similarity compare device 5, it is connected with entry process device 4, all kinds of clauses and subclauses of the preservation in the entry process device 4 is carried out similarity calculate, and the result is returned in the entry process device 4.
The present invention is based on the result for retrieval real-time update system of user behavior information, further comprise database 1 and indexing unit 2.Store magnanimity information in the database 1.Indexing unit 2 is connected with database 1 and entry process device 3, and for example, indexing unit 2 can be the data retrieval devices such as search engine, literature search engine.The user is to indexing unit 2 input keywords, and indexing unit is implemented pre-service to keyword, and generates the information such as weights, citation times of result for retrieval, each clauses and subclauses according to the magnanimity information in the keyword retrieval database 1 that meets system requirements.
Protection content of the present invention is not limited to above embodiment.Under the spirit and scope that do not deviate from inventive concept, variation and advantage that those skilled in the art can expect all are included in the present invention, and take appending claims as protection domain.
Claims (15)
1. the result for retrieval real time updating method based on user behavior information is characterized in that, may further comprise the steps:
Step 1: obtain the result for retrieval that comprises at least one clauses and subclauses;
Step 2: obtain user behavior information;
Step 3: based on described user behavior Information Selection tupe, and according to the clauses and subclauses similarity, the clauses and subclauses in the described result for retrieval are processed;
Step 4: obtain and display update after result for retrieval.
2. result for retrieval real time updating method as claimed in claim 1 is characterized in that, the result for retrieval that comprises at least one clauses and subclauses described in the described step 1 is to obtain by following steps:
Steps A 1: obtain keyword, described keyword is carried out pre-service;
Steps A 2: judge whether described pretreated keyword meets the requirements; If satisfactory, then retrieve the generation result for retrieval according to described keyword; If undesirable, then re-execute described steps A 1, A2, until generate result for retrieval.
3. result for retrieval real time updating method as claimed in claim 1 is characterized in that, in the described step 1, according to weights described clauses and subclauses is sorted from high to low.
4. result for retrieval real time updating method as claimed in claim 3 is characterized in that, in the described step 1, described clauses and subclauses weights are identical according to described clauses and subclauses citation times sort from high to low.
5. result for retrieval real time updating method as claimed in claim 1, it is characterized in that described user behavior information comprises the physiology sign information when the historical accesses entry of user, user ignore clauses and subclauses, user and access length reading time of the content-length of the time interval of different clauses and subclauses, the historical accesses entry of user, the historical accesses entry of user, user's accesses entry.
6. result for retrieval real time updating method as claimed in claim 1 is characterized in that, further comprises step 5: repeat described step 2 to step 4, until stop when stopping to obtain described user behavior information.
7. result for retrieval real time updating method as claimed in claim 1 is characterized in that, the tupe of described step 3 comprises:
Hide historical accesses entry pattern, similar historical accesses entry sequencing model or similar ignore in the clauses and subclauses sequencing model any one or more.
8. result for retrieval real time updating method as claimed in claim 7 is characterized in that, described hiding historical accesses entry comprises:
Step B1: the clauses and subclauses of choosing the user to access based on described user behavior information;
Step B2: the historical accesses entry of described user is shifted out from described result for retrieval;
Step B3: described clauses and subclauses of having accessed are deposited in the historical accesses entry set.
9. result for retrieval real time updating method as claimed in claim 7 is characterized in that, described similar historical accesses entry ordering comprises:
Step C1: the clauses and subclauses of choosing the user to access based on described user behavior information;
Step C2: described historical accesses entry is deposited in the described historical accesses entry set;
Step C3: the clauses and subclauses in the described historical accesses entry set are carried out the similarity contrast, obtain the similar content between the described clauses and subclauses;
Step C4: according to described similar content each clauses and subclauses in the described result for retrieval are carried out similarity contrast, generate the similarity value of described each clauses and subclauses and described similar content;
Step C5: from high to low each clauses and subclauses in the described result for retrieval are sorted according to described similarity value.
10. result for retrieval real time updating method as claimed in claim 7 is characterized in that, describedly similarly ignores clauses and subclauses orderings and comprises:
Step D1: the clauses and subclauses of choosing the user in access, to ignore based on described user behavior information;
Step D2: described user's the clauses and subclauses of ignoring are deposited in history and ignore in the entry set;
Step D3: each clauses and subclauses that each clauses and subclauses in the described result for retrieval and described history are ignored in the entry set are carried out the similarity contrast, obtain the similarity weights of each clauses and subclauses in the described result for retrieval;
Step D4: from low to high each clauses and subclauses in the described result for retrieval are sorted according to described similarity weights.
11. result for retrieval real time updating method as claimed in claim 9, it is characterized in that, further comprise: calculate the similarity value of clauses and subclauses and described similar content in the described historical accesses entry set, each clauses and subclauses during described historical accesses entry is gathered sort from high to low according to described similarity value.
12. the result for retrieval real-time update system based on user behavior information is characterized in that, comprising:
User behavior information acquisition device (3), it obtains user behavior information;
Similarity compare device (5), it comprises the functional module of calculating similarity;
Entry process device (4), it is connected with described user behavior information acquisition device (3) and similarity compare device (5), for the described user behavior Information Selection pattern of obtaining according to described user behavior information acquisition device (3), and according to the similarity comparing result of described similarity compare device (5) for described clauses and subclauses, process the clauses and subclauses in the described result for retrieval;
Display device (6), it is connected with described entry process device (4), and reception also shows the clauses and subclauses that sent by described entry process device (4).
13. result for retrieval real-time update as claimed in claim 12 system is characterized in that, further comprises:
Database (1), it stores magnanimity information;
Indexing unit (2), it is connected with described database (1) and described entry process device (4), is used for generating described result for retrieval according to the described magnanimity information of keyword retrieval.
14. result for retrieval real-time update as claimed in claim 13 system is characterized in that, described indexing unit (2) generates the weights of the matching degree of each clauses and subclauses and described keyword in the described result for retrieval.
15. result for retrieval real-time update as claimed in claim 12 system, it is characterized in that described user behavior information acquisition device (3) comprises mouse, keyboard, image acquisition equipment, built-in timing device, infrared induction equipment, GPS, the sense of touch sensing apparatus of computer system.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012104534649A CN102930041A (en) | 2012-11-12 | 2012-11-12 | Retrieval result real-time updating method based on user behavior information and system thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012104534649A CN102930041A (en) | 2012-11-12 | 2012-11-12 | Retrieval result real-time updating method based on user behavior information and system thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102930041A true CN102930041A (en) | 2013-02-13 |
Family
ID=47644838
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2012104534649A Pending CN102930041A (en) | 2012-11-12 | 2012-11-12 | Retrieval result real-time updating method based on user behavior information and system thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102930041A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103440271A (en) * | 2013-08-02 | 2013-12-11 | 亚太宝龙科技（湖南）有限公司 | Method and device for displaying historical directories in operating system |
| CN104035933A (en) * | 2013-03-06 | 2014-09-10 | 腾讯科技（深圳）有限公司 | Method and system for subscribing to reading source |
| CN105468652A (en) * | 2014-09-12 | 2016-04-06 | 北大方正集团有限公司 | Retrieval sorting method and system |
| CN107451141A (en) * | 2016-05-30 | 2017-12-08 | 阿里巴巴集团控股有限公司 | Processing exchange method, the apparatus and system of a kind of data recommendation |
| CN107590176A (en) * | 2017-07-31 | 2018-01-16 | 北京奇艺世纪科技有限公司 | A kind of preparation method of evaluation index, device and electronic equipment |
| CN108062391A (en) * | 2017-12-15 | 2018-05-22 | 上海速邦信息科技有限公司 | Knowledge pushes management system in a kind of ITSM platforms |
| CN108270815A (en) * | 2016-12-30 | 2018-07-10 | 上海互联网软件集团有限公司 | Hot information supplying system |
| CN109800319A (en) * | 2018-12-26 | 2019-05-24 | 中国科学院自动化研究所南京人工智能芯片创新研究院 | Image processing method, device, computer equipment and storage medium |
| CN110275955A (en) * | 2019-06-21 | 2019-09-24 | 中国科学院计算机网络信息中心 | Recognition methods, device, storage medium and the processor of text type |
| CN111414534A (en) * | 2019-01-07 | 2020-07-14 | 阿里巴巴集团控股有限公司 | Information processing method, device, equipment and storage medium |
| CN112966172A (en) * | 2019-12-12 | 2021-06-15 | 北京沃东天骏信息技术有限公司 | Search method and search device |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1996316A (en) * | 2007-01-09 | 2007-07-11 | 天津大学 | Search engine searching method based on web page correlation |
| CN101246499A (en) * | 2008-03-27 | 2008-08-20 | 腾讯科技（深圳）有限公司 | Network information search method and system |
| CN102682090A (en) * | 2012-04-26 | 2012-09-19 | 焦点科技股份有限公司 | System and method for matching and processing sensitive words on basis of polymerized word tree |
-
2012
- 2012-11-12 CN CN2012104534649A patent/CN102930041A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1996316A (en) * | 2007-01-09 | 2007-07-11 | 天津大学 | Search engine searching method based on web page correlation |
| CN101246499A (en) * | 2008-03-27 | 2008-08-20 | 腾讯科技（深圳）有限公司 | Network information search method and system |
| CN102682090A (en) * | 2012-04-26 | 2012-09-19 | 焦点科技股份有限公司 | System and method for matching and processing sensitive words on basis of polymerized word tree |
Non-Patent Citations (1)
| Title |
|---|
| 黄磊: ""基于实例学习的搜索引擎结果优化系统设计与实现"", 《中国硕士学位论文全文数据库（电子期刊）》, 15 May 2010 (2010-05-15), pages 138 - 312 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104035933A (en) * | 2013-03-06 | 2014-09-10 | 腾讯科技（深圳）有限公司 | Method and system for subscribing to reading source |
| WO2014135092A1 (en) * | 2013-03-06 | 2014-09-12 | Tencent Technology (Shenzhen) Company Limited | Method and system for subscribing reading feed |
| CN104035933B (en) * | 2013-03-06 | 2019-01-29 | 腾讯科技（深圳）有限公司 | A kind of reading source method for subscribing and system |
| CN103440271A (en) * | 2013-08-02 | 2013-12-11 | 亚太宝龙科技（湖南）有限公司 | Method and device for displaying historical directories in operating system |
| CN103440271B (en) * | 2013-08-02 | 2017-12-12 | 江苏智光创业投资有限公司 | The method and its device of history catalogue are shown in operating system |
| CN105468652A (en) * | 2014-09-12 | 2016-04-06 | 北大方正集团有限公司 | Retrieval sorting method and system |
| CN107451141A (en) * | 2016-05-30 | 2017-12-08 | 阿里巴巴集团控股有限公司 | Processing exchange method, the apparatus and system of a kind of data recommendation |
| CN107451141B (en) * | 2016-05-30 | 2021-01-29 | 阿里巴巴集团控股有限公司 | Data recommendation processing interaction method, device and system |
| CN108270815A (en) * | 2016-12-30 | 2018-07-10 | 上海互联网软件集团有限公司 | Hot information supplying system |
| CN107590176B (en) * | 2017-07-31 | 2021-01-15 | 北京奇艺世纪科技有限公司 | Evaluation index obtaining method and device and electronic equipment |
| CN107590176A (en) * | 2017-07-31 | 2018-01-16 | 北京奇艺世纪科技有限公司 | A kind of preparation method of evaluation index, device and electronic equipment |
| CN108062391A (en) * | 2017-12-15 | 2018-05-22 | 上海速邦信息科技有限公司 | Knowledge pushes management system in a kind of ITSM platforms |
| CN109800319A (en) * | 2018-12-26 | 2019-05-24 | 中国科学院自动化研究所南京人工智能芯片创新研究院 | Image processing method, device, computer equipment and storage medium |
| CN111414534A (en) * | 2019-01-07 | 2020-07-14 | 阿里巴巴集团控股有限公司 | Information processing method, device, equipment and storage medium |
| CN111414534B (en) * | 2019-01-07 | 2023-06-30 | 阿里巴巴集团控股有限公司 | Information processing method, apparatus, device and storage medium |
| CN110275955A (en) * | 2019-06-21 | 2019-09-24 | 中国科学院计算机网络信息中心 | Recognition methods, device, storage medium and the processor of text type |
| CN112966172A (en) * | 2019-12-12 | 2021-06-15 | 北京沃东天骏信息技术有限公司 | Search method and search device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102930041A (en) | Retrieval result real-time updating method based on user behavior information and system thereof | |
| US10180967B2 (en) | Performing application searches | |
| CN102279851B (en) | Intelligent navigation method, device and system | |
| US7739221B2 (en) | Visual and multi-dimensional search | |
| CN109271574A (en) | A kind of hot word recommended method and device | |
| TWI636416B (en) | Method and system for multi-phase ranking for content personalization | |
| US9183281B2 (en) | Context-based document unit recommendation for sensemaking tasks | |
| US7917514B2 (en) | Visual and multi-dimensional search | |
| US8626768B2 (en) | Automated discovery aggregation and organization of subject area discussions | |
| Jiang et al. | Mining search and browse logs for web search: A survey | |
| US20090100015A1 (en) | Web-based workspace for enhancing internet search experience | |
| CN111859160A (en) | Method and system for recommending session sequence based on graph neural network | |
| CN105022775A (en) | Apparatus and method for structuring web page access history | |
| Prajapati | A survey paper on hyperlink-induced topic search (HITS) algorithms for web mining | |
| CN110175895A (en) | A kind of item recommendation method and device | |
| CN108959580A (en) | A kind of optimization method and system of label data | |
| Wang et al. | Recommending high-utility search engine queries via a query-recommending model | |
| Takano et al. | An adaptive e-learning recommender based on user's web-browsing behavior | |
| Krohn et al. | Concept lattices for knowledge management | |
| Nawazish et al. | Integrating “random forest” with indexing and query processing for personalized search | |
| Cheng et al. | Context-based page unit recommendation for web-based sensemaking tasks | |
| Ahamed et al. | Deduce user search progression with feedback session | |
| Fung et al. | Discover information and knowledge from websites using an integrated summarization and visualization framework | |
| Rana et al. | Analysis of web mining technology and their impact on semantic web | |
| CN105740255B (en) | Network search method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20130213 |