WO2017005119A1

WO2017005119A1 - Method and device for implementing individualized guidance

Info

Publication number: WO2017005119A1
Application number: PCT/CN2016/087465
Authority: WO
Inventors: 王有良
Original assignee: 阿里巴巴集团控股有限公司; 王有良
Priority date: 2015-07-09
Filing date: 2016-06-28
Publication date: 2017-01-12
Also published as: CN106326320A

Abstract

A method and device for implementing individualized guidance. The method comprises: according to a historical access sequence set of a user, conducting frequent access pattern mining to obtain a frequent access pattern set, and then, according to current access information about the user, recognizing a user access sequence set (S1); matching each access sequence in the user access sequence set with items in the frequent access sequence set, so as to obtain a corresponding rule set (S2); and recommending to the user the latter items of all the rules in the rule set as a function guidance list(S3). Accordingly, the device comprises a mining module, a recognition module, a matching module and a guidance module. The method and device can provide individualized guidance for the next operation of a user, thereby facilitating the operation of the user, and improving the user experience.

Description

Method and device for realizing personalized guidance

Technical field

The invention belongs to the technical field of computers, and in particular relates to a method and a device for implementing personalized guidance.

Background technique

In the field of computer technology, especially the various applications currently provided based on the Internet, a customized menu is provided for users on the application webpage. When the user selects a function button on the webpage, the user pops up a menu for the user to make the next operation selection, and the pop-up menu may also be a translucent mask layer, and the mask layer has a user for the next operation selection. Select items to provide functional guidance to the user.

However, the options used in the existing layer are used according to the business experience. In a variety of applications, the user can choose a large number of options, and can not include all the options in the layer. in. Even the choices included in the mask are not necessarily the most reasonable choices, and some custom choices may be less common for the user, and the usual ones are not necessarily in the mask.

The Chinese invention patent with the publication number of CN 103092471A discloses a method and a terminal for realizing a dynamic function menu. The method counts the user usage frequency of each function menu, obtains the function menu usage frequency statistics information, and uses the frequency statistics information according to the function menu to The system presets the layout or layout form, dynamically adjusting the function menu list or layout order. The menu can be properly arranged or adjusted according to the user's usage habits or form requirements, so as to improve the user experience of the menu and facilitate the operation of the user.

However, in the invention, only the user usage frequency of each function menu that has been already performed is counted, and the function menu list is adjusted according to the usage frequency statistical information, that is, the adjustment of the already arranged function menu. For the user's online operation, the user has more and more options for the next operation target, which can be another website or other sections on the same webpage. In the prior art, it is impossible to provide the user with the guidance of the next operation according to the current operation of the user, and the user needs to find the page click operation of the next operation, which wastes a lot of unnecessary time and energy for the user. For a complicated application system, the user operates It is even more inconvenient to do.

Summary of the invention

An object of the present invention is to provide a method and a device for implementing personalized guidance, which can provide personalized guidance for the user in the next operation according to the current operation behavior, historical operation information, and user attribute information of the user, which facilitates the operation of the user. Improved user experience.

In order to achieve the above object, the technical solution of the present invention is as follows:

A method of implementing personalized guidance, including:

Identifying a user access sequence set based on user access information;

Matching each access sequence in the user access sequence set with the items in the frequent access mode set to obtain a corresponding rule set;

Recommend the latter item of all rules in the rule set as a function guide list to the user;

The frequent access mode set is obtained by performing frequent access pattern mining according to a user history access sequence set.

Further, the performing frequent access pattern mining according to the user history access sequence set includes:

Preprocessing according to the web server log file to obtain a user history access sequence set;

Frequent access pattern mining for user history access sequence sets results in a set of frequent access patterns.

The pre-processing according to the web server log file includes data purification, user identification, session identification, and path replenishing steps. The frequent access pattern mining is performed on the user history access sequence set, and the prefixspan data mining algorithm is adopted.

According to the invention, each access sequence in the user access sequence set is matched with the items in the frequent access mode set, and the longest matching selection method is adopted.

In an implementation manner of the present invention, the longest matching selection method sets a frequent access mode that includes a frequent access mode in which the number of webpage nodes in the user access sequence is greater than the first preset.

In another implementation manner of the present invention, the longest matching selection method sets a frequent access mode in which the first N frequent access modes that match the number of webpage nodes in the user access sequence are matched.

The invention also proposes a device for implementing personalized guidance, which comprises:

a mining module, configured to perform frequent access pattern mining according to a user history access sequence set to obtain a frequent access mode set;

An identification module, configured to identify a user access sequence set according to user access information;

a matching module, configured to match each access sequence in the user access sequence set with an item in the frequent access mode set to obtain a corresponding rule set;

A boot module that is used to recommend the latter item of all rules in the rule set as a function boot list to the user.

Further, when performing the frequent access mode mining, the mining module performs the following operations:

The mining module performs data purification, user identification, session identification, and path supplement processing on the web server log file during preprocessing; and uses the prefixspan data mining algorithm when performing frequent access pattern mining.

Further, the matching module uses the longest matching selection method for matching.

The method and device for implementing the personalized guidance provided by the present invention pre-processes the web log data accessed by the user, excavates the frequent access mode set of the user, and then matches the rule set, and outputs the function guide to the user client browser. A list that allows the user to select what will be done next. Provide personalized guidance for the user's next operation, which is convenient for the user's operation and improves the user experience.

DRAWINGS

1 is a flow chart of an offline processing process in a method for implementing personalized booting according to the present invention;

2 is a flow chart of a method for implementing personalized guidance according to the present invention;

FIG. 3 is a schematic structural diagram of an apparatus for implementing personalized guidance according to the present invention.

detailed description

The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. The examples do not constitute a limitation of the invention.

The general idea of the present invention is to preprocess the web log data, and then use the prefixspan algorithm to mine the frequent access sequence pattern set from the user's web access log, and then use the longest matching selection method to frequently access according to the obtained user current access sequence. The pattern set is matched to obtain the corresponding rule set. Finally, the web server program processes the rule set and outputs a function boot list to the client browser. This example takes the “Yu-Yi Room” Alibaba Cloud Public Computing Platform as an example to illustrate. The “Yu-Yi Room” Alibaba Cloud Public Computing Platform provides precision marketing solutions, merchant research and solutions for Taobao merchants, independent software developers, enterprises and scientific research institutions. Programs, affiliate marketing solutions, enterprise cloud data solutions, etc., there are many user-operable options, and it is necessary to provide users with a function guide based on the user's frequent access mode.

In this embodiment, a method for implementing personalized guidance includes an offline processing process and an online processing process, which are separately described below.

As shown in FIG. 1 , the offline processing process processes the web server log file through data purification, user identification, session identification, path supplementation, etc., to obtain a user history access sequence set, and then uses the optimized prefixspan data mining algorithm to the user. The historical access sequence set performs frequent access pattern mining to obtain a frequent access mode set, that is, frequent access pattern mining for the user historical access sequence set to obtain a frequent access mode set.

The user's access to the "Yu Yu Fang" Alibaba Cloud public computing platform is recorded in the Web server log file, and the Web server log file can be processed by data purification, user identification, session identification, path supplementation, etc. Access the sequence collection. An access sequence is a user's access to a website, which contains all the page nodes that have been visited at one time. The accessed page nodes are arranged in chronological order as access sequences.

Specifically, the data cleaning is to extract necessary fields from the web server log file, such as user ID, time, page node ID, etc.; user identification is to add the user ID through the cookie and the user ID correspondence table; The identification is divided into multiple sessions by a predetermined time unit, for example, 60 minutes, that is, the user's access path is divided into multiple access sequences, each session corresponds to one access sequence; finally, for the reason that the server synchronizes data It is possible that the page node that the user has visited has a lost situation, and the lost path is added through the website structure map for subsequent analysis. After the above pre-processing, you can get a user history visit. Ask the sequence collection.

Then, using the optimized prefixspan algorithm to perform frequent access pattern mining on the user history access sequence set, the frequent access pattern set is obtained. The specific prefixspan algorithm is implemented as follows:

Wherein, the input sequence pattern is equivalent to an access sequence, and the final output is frequently sequenced. The mode is the frequent access mode, and all the frequent access modes constitute a frequent access mode set. As for the prefixspan algorithm, which is commonly used in data mining, there are many commonly used algorithms in data mining, such as Apriori algorithm, which will not be described here.

The core idea of the online processing part is to process the user access information collected by the client through the web server, obtain the user's access path, perform session identification on the user's access path, and obtain the user access sequence set; then each of the access sequence sets An access sequence is used to match the access sequence set with the items in the frequent access mode set by using the longest matching selection method to obtain a corresponding rule set; finally, the latter item of all rules in the rule set is recommended as a function boot list to the user.

Specifically, as shown in FIG. 2, the following steps are included:

S1, identifying a user access sequence set according to user access information;

The user logs in the “Yu Yu Fang” Alibaba Cloud public computing platform on his computer. The “Yu Yu Fang” Alibaba Cloud public computing platform collects user access information through the client browser and sends it to the back-end web server of the Alibaba Cloud public computing platform. The web server performs session identification on the user access information, and divides the user's access path into multiple sessions in a predetermined time unit (for example, 60 minutes), that is, splits into multiple access sequences to obtain a user access sequence set. Different from the offline processing process, what is obtained here is the user's current access sequence set.

S2: Match each access sequence in the user access sequence set with an item in the frequent access mode set to obtain a corresponding rule set;

The longest match selection method is used to match each access sequence with an item in the frequent access mode set, and each item in the frequent access mode set is a frequent access mode. The longest matching selection method is adopted, that is, the items in the matched frequent access mode set contain most of the webpage nodes in the user access sequence, for example, the items in the frequent access mode set may be set to contain more than 70% of the number of webpage nodes in the user access sequence. The item in the frequent access mode set is a match; or the frequent access mode in which the first N frequent access modes match the maximum number of webpage nodes in the user access sequence is matched.

For example, suppose the user access sequence A ₁ contains 10 web page nodes, and the frequent access mode set has B ₁ , B ₂ , and B ₃ three frequent access modes respectively being the first three frequent accesses that best match the user access sequence A _{1 .} Mode, where B ₁ contains 10 web page nodes of all user access sequence A ₁ , B ₂ contains ₉ of 10 web page nodes of user access sequence A ₁ , B ₃ contains 10 web page nodes of user access sequence A ₁ Of the eight, if the set N is 3, then B ₁ , B ₂ , and B _{3 are} matched frequent access patterns, and their set is the rule set.

S3. Recommend the latter item of all the rules in the rule set as a function guide list to the user.

The latter part of the rule refers to the webpage nodes that are not included in the user access list in the rule. These access nodes are webpage nodes that the user may visit later. Node B _1, for example, page 15 contains hypothesis, of which 10 have the same user accessing the page sequence node, then the node is the page 5 page node to the user may access.

Therefore, the latter item of all rules is recommended to the user as a function guide list, and displayed as a mask on the user's browser for the user to select. Therefore, the user can directly select the webpage node in the mask layer that he wants to access, and realize direct access.

For example, in the data engine area of the "Yu Yu Fang" Alibaba Cloud Public Computing Platform, after the user places the mouse on the data development module, according to the historical operation of the user, the frequent pattern of the user is mined and stored in the database. The user's current access sequence (ie, the current operation path) is used as a parameter, and the next operation set is found from the user's frequent access mode set according to the “longest matching mode”, such as “go to the authorization center” and transmitted to the user operation page. At this time, the page dynamically changes the link in the navigation tab, and the selection items such as “Go to the authorization center” are displayed in the navigation tab for the user to select, and the navigation label is displayed on the user browser in the form of a layer.

The device for implementing the personalized guidance based on the foregoing method, as shown in FIG. 3, includes:

When the mining module of this embodiment performs frequent access mode mining, the following operations are performed:

The mining module needs to perform data purification, user identification, session identification, and path supplement processing on the web server log file during the pre-processing; when performing frequent access pattern mining, the prefixspan data mining algorithm is adopted.

In this embodiment, the matching module uses the longest matching selection method to perform matching.

The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to be limiting, and those skilled in the art can make various corresponding changes according to the present invention and without departing from the spirit and scope of the present invention. Modifications, but such corresponding changes and modifications are intended to be included within the scope of the appended claims.

Claims

A method for implementing personalized guidance, comprising:

Identifying a user access sequence set based on user access information;

Matching each access sequence in the user access sequence set with the items in the frequent access mode set to obtain a corresponding rule set;

Recommend the latter item of all rules in the rule set as a function guide list to the user;

The frequent access mode set is obtained by performing frequent access pattern mining according to a user history access sequence set.
The method according to claim 1, wherein the performing frequent access pattern mining according to the user history access sequence set comprises:

Preprocessing according to the web server log file to obtain a user history access sequence set;

Frequent access pattern mining for user history access sequence sets results in a set of frequent access patterns.
The method according to claim 2, wherein the pre-processing according to the web server log file comprises data purification, user identification, session identification, and path replenishing steps.
The method according to claim 2, wherein the frequent access pattern mining is performed on the user history access sequence set, and the prefixspan data mining algorithm is adopted.
The method according to claim 1, wherein the matching of each access sequence in the user access sequence set with the items in the frequent access mode set uses the longest matching selection method.
The method according to claim 5, wherein the longest matching selection method sets a frequent access mode that includes a frequent access mode in which the number of web page nodes in the user access sequence is greater than the first set preset.
The method according to claim 5, wherein the longest matching selection method sets a frequent access mode in which the first N frequent access modes that match the number of webpage nodes in the user access sequence are matched.
An apparatus for implementing personalized guidance, comprising:

a mining module for performing frequent access pattern mining based on a user history access sequence set To frequent access mode collections;

An identification module, configured to identify a user access sequence set according to user access information;

a matching module, configured to match each access sequence in the user access sequence set with an item in the frequent access mode set to obtain a corresponding rule set;

A boot module that is used to recommend the latter item of all rules in the rule set as a function boot list to the user.
The device according to claim 8, wherein the mining module performs the following operations when performing frequent access mode mining:

Preprocessing according to the web server log file to obtain a user history access sequence set;

Frequent access pattern mining for user history access sequence sets results in a set of frequent access patterns.
The device according to claim 9, wherein the mining module performs the following operations when performing preprocessing:

Data purification, user identification, session identification, and path replenishment processing are performed on the web server log file.
The apparatus according to claim 9, wherein the mining module uses a prefixspan data mining algorithm when performing frequent access pattern mining.
The apparatus of claim 8 wherein said matching module employs a longest match selection method for matching.
The apparatus according to claim 12, wherein said longest matching selection method sets a frequent access mode including a frequent access mode in which the number of web page nodes in the user access sequence is greater than the first set preset.
The apparatus according to claim 12, wherein the longest matching selection method sets a frequent access mode in which the first N frequent access modes that match the number of webpage nodes in the user access sequence are matched.