Background technology
In field of computer technology, especially it is currently based on the various application that the Internet provides, in applicationWebpage on provided the user customized menu.When user selects a functional keys on webpageAfter, can eject for user and carry out, for user, the menu that next operation selects, the menu ejected also may be usedTo be translucent mask, mask has and carries out, for user, the options that next operation selects, forUser provides function to guide.
But those optionies being used for guiding are all to set according to business experience in advance in existing maskPutting, in the application system that function is various, the options that user can select is the most, it is impossible to willAll optionies are included in mask.Even if it is the most reasonable that the options being included in mask is also not necessarilyOptions, some customized options perhaps user is the most conventional, and conventional may not be in mask.
The Chinese invention patent of Publication No. CN 103092471A discloses a kind of dynamic function menuImplementation method and terminal, the method is added up the user of each function menu and is used frequency, obtains functionMenu uses frequency statistics information, uses frequency statistics information according to function menu, with systemic presuppositionSpread pattern or distribution form, dynamic adjustment function menu list or the sequence of layout.Can according toThe use habit at family or form demand, carry out suitable arrangement or distributing adjustment to menu, to improve dishSingle Consumer's Experience, user friendly operation.
But only user to each function menu having had uses frequency to carry out in this inventionStatistics, and according to using frequency statistics information to adjust function menu list, i.e. to the function having hadMenu carries out the adjustment arranged.And for the upper net operation of user, user's next one operation targetOptional item get more and more, can be another website, or other plates on same webpageBlock.Guiding of next operation cannot be provided the user in the prior art according to the current operation of user,Need user oneself to find next step operation page click through operation, this for a user,Waste the most unnecessary time and efforts, for the application system that function is numerous and diverse, Yong HucaoIt is more inconvenient that work gets up.
Summary of the invention
It is an object of the invention to provide a kind of personalized implementation method guided and device, it is possible to according toCurrent operation behavior, historical operation information and the customer attribute information at family, for next step behaviour of userMake to provide personalization to guide, facilitate the operation of user, improve Consumer's Experience.
To achieve these goals, technical solution of the present invention is as follows:
A kind of personalized implementation method guided, including:
According to user access information, identify user's access sequence collection；
Each access sequence concentrated by user's access sequence, with the item in frequent access paths setMate, it is thus achieved that corresponding rule set；
The consequent of strictly all rules in rule set is recommended user as function boot-list；
Wherein, described frequent access paths set is to carry out frequently according to user's history access sequence setAccess module excavation obtains.
Further, described carry out frequent access paths excavation according to user's history access sequence set,Including:
Carry out pretreatment according to Web server log file and obtain user's history access sequence collection；
User's history access sequence collection is carried out frequent access paths excavation and obtains frequent access paths collectionClose.
Wherein, described carry out pretreatment according to Web server log file, including data purification,User's identification, session identification, path supplement step.Described user's history access sequence collection is carried out frequencyNumerous access module excavates, and uses prefixspan data mining algorithm.
Each access sequence concentrated by user's access sequence of the present invention, with frequent access pathsItem in set mates, and uses the longest match selection method.
A kind of implementation of the present invention, the longest described match selection method, arrange and access sequence containing userIn row, web page joint number sets preset frequent access paths and frequently accesses mould as mate more than firstFormula.
The another kind of implementation of the present invention, the longest described match selection method, arrange and access sequence with userIn row, web page joint number mates the frequent access paths that most top n frequent access paths is coupling.
The present invention propose the most simultaneously a kind of personalized guide realize device, including:
Excavate module, excavate for carrying out frequent access paths according to user's history access sequence setTo frequent access paths set；
Identification module, for according to user access information, identifies user's access sequence collection；
Matching module, for each access sequence concentrated by user's access sequence, with frequently accessItem in set of modes mates, it is thus achieved that corresponding rule set；
Guide module, for being recommended as function boot-list by the consequent of strictly all rules in rule setUser.
Further, described excavation module, when carrying out frequent access paths and excavating, performs to operate as follows:
Carry out pretreatment according to Web server log file and obtain user's history access sequence collection；
User's history access sequence collection is carried out frequent access paths excavation and obtains frequent access paths collectionClose.
Wherein said excavation module is when carrying out pretreatment, to Web server log file numberSupplement process according to purification, user's identification, session identification, path；Carrying out frequent access paths excavationTime, use prefixspan data mining algorithm.
Further, described matching module uses the longest match selection method to mate.
The implementation method of a kind of personalized guiding that the present invention proposes and device, by access userWeb log data carries out pretreatment, excavates and is frequently visited by the user set of modes, then matches ruleThen collect, to subscription client browser output function boot-list, it is simple to user selects next step willThe operation carried out.There is provided personalization to guide for next step operation of user, facilitate the operation of user, carryRise Consumer's Experience.
Detailed description of the invention
With embodiment, technical solution of the present invention is described in further details below in conjunction with the accompanying drawings, real belowExecute example and do not constitute limitation of the invention.
The general thought of the present invention is by web log data pretreatment, then utilizes prefixspan to calculateMethod excavates frequent access sequence set of modes from the web access log of user, then according to acquisitionUser's current accessed sequence, utilize the longest match selection method to carry out in frequent access paths setJoin, thus obtain corresponding rule set, finally by Web server end program, regular collection is carried outProcess backward client browser output function boot-list.The present embodiment is with " emperor's diet room " Ali's cloudIllustrating as a example by public calculating platform, " emperor's diet room " Ali's cloud public calculating platform is Taobao businessFamily, independent software vendors, enterprise, scientific research institution provide precision marketing solution, businessman's investigationSolution, membership promotion solution, enterprise's cloud data solution etc., user is exercisableOptions is the most, needs to provide the user and a kind of guides based on the function being frequently visited by the user pattern.
A kind of personalized implementation method guided of the present embodiment, including off-line process and online treatmentProcess, illustrates individually below.
As it is shown in figure 1, off-line process be by Web server log file through data purification,The steps such as user's identification, session identification, path supplement process, and obtain user's history access sequenceSet, then utilizes the prefixspan data mining algorithm after optimizing to user's history access sequence collectionConjunction carries out frequent access paths excavation and obtains frequent access paths set, i.e. to user's history access sequenceSet carries out frequent access paths excavation and obtains frequent access paths set.
User's access on " emperor's diet room " Ali's cloud public calculating platform all records in Web serviceIn device journal file, by Web server log file through data purification, user's identification, sessionAfter the steps such as identification, path supplement process, it is possible to obtain user's history access sequence set.VisitAsk that sequence is the user once access to website, wherein comprise all pages joint once accessedPoint, the page node accessed is arranged as access sequence sequentially in time.
Specifically, data purification is to extract necessary field from Web server log file, as withFamily ID, time, page node ID etc.；It is the user that will be not logged on that user identifies, passes through cookieID is supplemented by table corresponding with ID；Session identification be by regulation time in units of such as60 minutes, by the access path cutting of user be multiple session, i.e. cutting be multiple access sequence, oftenThe corresponding access sequence of individual session；Finally, for due to server sync data, having canThe page node that energy user accessed has the situation of loss, by site Diagrams, the path that will loseOn Bu Chonging, in order to follow-up analysis.After above-mentioned pretreatment, it becomes possible to obtain user's history and visitAsk arrangement set.
Then utilize the prefixspan algorithm after optimizing that user's history access sequence set is carried out frequentlyAccess module excavates and obtains frequent access paths set, and concrete prefixspan algorithm realizes such as following table:
Wherein, the sequence pattern of input is equivalent to an access sequence, and the Frequent episodes finally exportedPattern is exactly frequent access paths, the composition frequent access paths set of all of frequent access paths.CloseIn prefixspan algorithm, for algorithm conventional in data mining, in data mining, conventional calculationMethod is also a lot, and such as Apriori algorithm etc. repeat no more here.
The core concept of online treatment part is, the user access information that client is collected is passed through WebServer-side processes, obtains the access path of user, and conversate identification to the access path of user,Obtain user's access sequence collection；Then each access sequence concentrated access sequence, utilizes the longestAccess sequence collection is mated by match selection method with the item in frequent access paths set, it is thus achieved that correspondingRule set；Finally the consequent of strictly all rules in rule set is recommended use as function boot-listFamily.
Specifically, as in figure 2 it is shown, comprise the steps:
S1, according to user access information, identify user's access sequence collection；
User logs in " emperor's diet room " Ali's cloud public calculating platform, " emperor's diet room " on the computer of oneselfAli's cloud public calculating platform collects user access information by client browser, and mails to Ali's cloudThe rear end Web server of public calculating platform, user access information is conversated by Web serverIdentify, in units of the time of regulation (such as 60 minutes), be many by the access path cutting of userIndividual session, i.e. cutting are multiple access sequence, obtain user's access sequence collection.With off-line processExcept for the difference that, obtain here is the current access sequence collection of user.
S2, each access sequence concentrated by user's access sequence, and in frequent access paths setItem mate, it is thus achieved that corresponding rule set；
Use the longest match selection method, by each access sequence and the item in frequent access paths setMating, each item in frequent access paths set is exactly a frequent access paths.UseLong match selection method, the item in the frequent access paths set i.e. matched contains in user's access sequenceMost of web page joint, such as can arrange the item in frequent access paths set contain user accessIn sequence, web page joint number is more than 70%, then the item in this frequent access paths set is occurrence；OrPerson is arranged mates most top n frequent access paths and is with web page joint number in user's access sequenceThe frequent access paths joined.
For example, it is assumed that user access sequence A1Include 10 web page joints, frequent access paths collectionConjunction has B1, B2, B3Three frequent access paths are respectively and user access sequence A1Mate mostFirst three frequent access paths, wherein B1Comprise all user access sequence A110 webpages jointPoint, B2Comprise user access sequence A110 web page joints in 9, B3Comprise user to visitAsk sequence A110 web page joints in 8, if the N set is as 3, then B1, B2,B3Being the frequent access paths matched, their set is exactly rule set.
S3, the consequent of strictly all rules in rule set is recommended user as function boot-list.
Regular is consequent, refers to the web page joint that in rule, user's access list does not comprises, and these are visitedAsk that node is all the follow-up web page joint that may go to access of user.Such as B1The web page joint comprisedAssume there are 15, wherein have 10 identical with the web page joint in user's access sequence, then 5Web page joint is exactly the web page joint that user may go to access.
Thus the consequent of strictly all rules is recommended user as function boot-list, browsing userIt is shown as mask on device, selects for user.Thus user can directly select in mask and want to accessWeb page joint, it is achieved directly access.
Such as in the data engine region of " emperor's diet room " Ali's cloud public calculating platform, user is mouseAfter being placed in data mining module, according to the historical operation of this user, excavate the frequent of this userPattern is also stored in data base, now user's current accessed sequence (the most current courses of action)As parameter, find down from the frequent access paths set of this user according to " the longest match pattern "Single stepping collection, such as " going to authorization center ", and passes to the user operation page, and now the page is dynamicChange navigation tag in link, options shows in navigation tag " will to go to authorization center " etc.In, selecting for user, navigation tag shows on user browser with the form of mask.
What the present embodiment a kind of personalization based on said method guided realizes device, as it is shown on figure 3,Including:
Excavate module, excavate for carrying out frequent access paths according to user's history access sequence setTo frequent access paths set；
Identification module, for according to user access information, identifies user's access sequence collection；
Matching module, for each access sequence concentrated by user's access sequence, with frequently accessItem in set of modes mates, it is thus achieved that corresponding rule set；
Guide module, for being recommended as function boot-list by the consequent of strictly all rules in rule setUser.
The present embodiment excavation module, when carrying out frequent access paths and excavating, performs to operate as follows:
Carry out pretreatment according to Web server log file and obtain user's history access sequence collection；
User's history access sequence collection is carried out frequent access paths excavation and obtains frequent access paths collectionClose.
Wherein excavation module is when carrying out pretreatment, needs Web server log file numberSupplement process according to purification, user's identification, session identification, path；Carrying out frequent access paths excavationTime, use prefixspan data mining algorithm.
In the present embodiment, matching module uses the longest match selection method to mate.
Above example is only limited in order to technical scheme to be described, is not carrying on the backIn the case of present invention spirit and essence thereof, those of ordinary skill in the art work as can be according to the present inventionMake various corresponding change and deformation, but these change accordingly and deformation all should belong to institute of the present inventionAttached scope of the claims.