CN108009215B - A kind of search results pages user behavior pattern assessment method, apparatus and system - Google Patents

Abstract

The invention discloses a method, device and system for evaluating user behavior patterns on a search result page. The method includes: receiving user information, and classifying user types according to the cognitive style in the user information; receiving at least two user information in the search engine result page. different types of test behavior data; visually process test behavior data and conduct preliminary qualitative analysis; mine different types of users and different types of user behavior patterns under page layout for analysis, and evaluate user behavior patterns on search results pages. Through various types of test behavior data, the analysis of common user behavior patterns in search engine results pages under different types of layouts and different types of users has made important contributions to optimizing page layouts and improving link delivery effects.

Description

A method, device and system for evaluating user behavior patterns on search result pages

technical field

The invention belongs to the technical field of search engine result page optimization, and relates to a search result page user behavior pattern evaluation method, device and system, in particular to a search result page user behavior pattern evaluation method, device and system based on multimodal features .

Background technique

In the information age, the Internet has penetrated into all aspects, not only manifested as changes in people's lifestyles, but also brought about tremendous changes in corporate marketing strategies. There are more and more ways for people to obtain information, and the ways for enterprises to disseminate information are also increasing, and the most important system is the search engine (Search Engine). A search engine is a user who sends a search task through the search box of the search engine when there is a certain need, collects information from the Internet through certain strategies and procedures, and after processing and organizing, sends relevant information to the search engine result page according to certain rules. (SERP:Search EngineResults Page) The fixed mode is displayed to the user's application system. Moreover, commercial promotion related to search content is often accompanied by search engine result pages, which is one of the most common marketing methods on the Internet. It can be seen that search engines have an important impact on people's lives. Reasonable analysis of user behavior patterns in search engine results pages can help understand users' daily behaviors and habits, and better page layout and typesetting for information processing and business promotion design. Therefore, exploring user behavior patterns not only has important research value and commercial value, but also has important guiding significance for system designers.

At this stage, there are many studies on user behavior in search engine results pages. Most researchers obtain user behavior information by mining users' network behavior logs or recording user cursor information, and then perform relevant analysis, processing and recommend.

However, the current research on user behavior in search engine results pages mainly has the following shortcomings:

(1) Most of the information sources used are subjective information, which is not accurate enough;

(2) Although some researchers are aware of the role of context, they seldom analyze the interaction relationship between different regions in the search results page, and seldom analyze the common behavior patterns of different types of users under different layouts.

To sum up, in the prior art, there is still no effective solution to the problem of how to scientifically and accurately evaluate user behavior patterns in search engine result pages and optimize page layout to improve user experience.

Contents of the invention

Aiming at the deficiencies in the prior art and solving the problems in the prior art on how to scientifically and accurately evaluate the user behavior pattern in the search engine result page and optimize the page layout to improve the user experience, the present invention provides a search result page user behavior Mode evaluation method, device and system, specifically a method, device and system for user behavior mode evaluation of search result pages based on multi-mode features, through various types of test behavior data, analysis of different types of layouts and different types of search engine result pages. It is a common user behavior pattern under different types of users, which makes an important contribution to optimizing the page layout and improving the effect of link delivery.

The first object of the present invention is to provide a method for evaluating user behavior patterns on a search result page.

In order to achieve the above object, the present invention adopts the following technical scheme:

A method for evaluating user behavior patterns on a search result page, the method comprising:

Receive user information, and classify user types according to the cognitive style in the user information;

receiving at least two different types of test behavior data of users within search engine results pages;

Visually process test behavior data and conduct preliminary qualitative analysis;

The behavior patterns of different types of users and different types of users under the page layout are respectively excavated for analysis, and the user behavior patterns of search result pages are evaluated.

As a further preferred solution, before receiving user information, users who are not suitable for eye movement collection are eliminated according to visual conditions, and a number of users are randomly selected from the remaining users to evaluate user behavior patterns on the search result page.

As a further preferred solution, the user information includes basic user information and user cognitive information;

The basic user information includes the user's name, gender, age and occupation;

The user cognition information is obtained by mosaic graph test method, including cognition style.

As a further preferred solution, the test behavior data is the information source data generated during the user's browsing process in the engine result page, including but not limited to cursor data and eye movement data;

The cursor data is a cursor event triggered by an input device obtained during the user's browsing of the search engine result page;

The eye-movement data is eye-tracking information acquired during a user's browsing of a search engine result page.

As a further preferred solution, the specific steps of visually processing test behavior data and performing preliminary qualitative analysis include:

Data visualization of cursor data and eye movement data;

Preliminary qualitative analysis of the impact of page layout on the eye movement data.

As a further preferred solution, the specific steps for analyzing different user types and different types of behavior patterns of users under different user types and page layouts include:

Using the same frequent pattern mining method as the number of evaluation behavior data types,

Mining the cursor behavior patterns under different user types and page layouts, based on the cursor data, analyzing the correlation patterns between different links in the search results page and the cursor behaviors between user types and page layouts, as well as the cursor behaviors in links under different user types and page layouts Combination mode of parameters;

And mining the eye-movement browsing patterns under different user types and page layouts, according to the eye-movement data, the time-series information in the process of users browsing the search engine result pages is obtained, and the sequence relationship between the links in the process of users browsing the search engine result pages is analyzed.

As a further preferred solution, the frequent itemset mining algorithm is used to mine the cursor behavior patterns under different user types and page layouts;

Using the directional frequent browsing pattern mining algorithm to mine eye movement browsing patterns under different user types and page layouts, the directional frequent browsing pattern mining algorithm is used to mine user directional and fixed-length eye movement browsing patterns under different page layouts to obtain user behavior processes Time series information in .

As a further preferred solution, the directional frequent browsing pattern mining method includes:

Obtain the browsing sequence data of the user on the search engine result page according to the test behavior data;

Add the adoption length and its corresponding support attribute in the browsing sequence data, and initialize it;

Process the support of each browsing sequence to make it zero and get a new sequence;

Determine whether the new sequence is a frequent sequence, and output the frequent sequence.

As a further preferred solution, in this method, the range of the link area is preset, and the time when the user enters the range of each link area is extracted from the test behavior data.

As a further preferred solution, the specific steps of obtaining the browsing sequence of the user on the search engine result page according to the test behavior data include:

Roughly classifying the test behavior data according to the page layout;

Arrange according to the order of time when users enter the area of each link, and each user views each web page corresponding to a piece of browsing sequence data.

As a further preferred solution, the specific steps of processing the support of each browsing sequence to make it zero include:

Preset support threshold;

Calculate the support of the first element of the browsing sequence data, set the support of the first element less than the support threshold to zero and eliminate the sequence;

Sort and classify the browsing sequence data according to the value of the first element, create a queue corresponding to it, enter the sequence into different queues by category, and delete the first element of each sequence;

Update the adopted length of the sequence and its corresponding support attribute until the elements in each sequence end with support equal to 0.

As a further preferred solution, the specific steps for judging whether a new sequence is a frequent sequence are:

Calculating the score of the new sequence, the score being the product of the adopted length of the sequence and the corresponding support degree;

Sort the scores to determine the maximum score;

The new sequence corresponding to the maximum score is a frequent browsing pattern sequence; otherwise, it is determined that the new sequence is a non-frequent browsing pattern sequence.

A second object of the present invention is to provide a computer-readable storage medium.

In order to achieve the above object, the present invention adopts the following technical scheme:

A computer-readable storage medium, in which a plurality of instructions are stored, and the instructions are adapted to be loaded by a processor of a terminal device and perform the following processing:

Receive user information, and classify user types according to the cognitive style in the user information;

receiving at least two different types of test behavior data of users within search engine results pages;

Visually process test behavior data and conduct preliminary qualitative analysis;

The behavior patterns of different types of users and different types of users under the page layout are respectively excavated for analysis, and the user behavior patterns of search result pages are evaluated.

The third object of the present invention is to provide a device for evaluating user behavior patterns on a search result page.

In order to achieve the above object, the present invention adopts the following technical scheme:

A device for evaluating user behavior patterns on a search result page, using Internet terminal equipment, including a processor and a computer-readable storage medium, the processor is used to implement instructions; the computer-readable storage medium is used to store multiple instructions, and the instructions are suitable for Loaded by the processor and performs the following processing:

Receive user information, and classify user types according to the cognitive style in the user information;

receiving at least two different types of test behavior data of users within search engine results pages;

Visually process test behavior data and conduct preliminary qualitative analysis;

The behavior patterns of different types of users and different types of users under the page layout are respectively excavated for analysis, and the user behavior patterns of search result pages are evaluated.

The fourth object of the present invention is to provide a system for evaluating user behavior patterns on a search result page.

In order to achieve the above object, the present invention adopts the following technical scheme:

A search result page user behavior pattern assessment system, the system is based on the above search result page user behavior pattern assessment method, comprising: a user information collection device, a behavior data collection device and a behavior pattern assessment device;

The user information collection device is used to collect user information and send it to the behavior pattern evaluation device;

The behavior data collection device is used to collect at least two different types of test behavior data of the user in the search engine result page, and send it to the behavior pattern evaluation device;

The behavior pattern evaluation device is used to receive user information, classify user types according to the cognitive style in the user information; receive at least two different types of test behavior data of the user in the search engine result page; visually process the test behavior data and Carry out preliminary qualitative analysis; excavate different user types and different types of user behavior patterns under the page layout for analysis, and evaluate the user behavior patterns of the search results page.

Beneficial effects of the present invention:

(1) A search result page user behavior pattern evaluation method, device and system according to the present invention, obtain at least two different types of test behavior data of the user in the search engine result page, for example, the user's eye movement data and cursor data, mining frequent behavior patterns through at least two information source characteristics generated during the user’s browsing process in the search engine result page, and analyzing the combination mode between page layout, user type, search engine result page and various links, As well as the timing relationship, the present invention is of great significance for improving the link layout mode in the search engine result page, personalized recommendation information, and advertisement delivery effect.

(2) A search result page user behavior pattern evaluation method, device and system according to the present invention collects behavior information of various information sources in the search engine result page, and proposes a novel, directional and fixed-length frequent browsing The pattern mining algorithm mines the frequent sequential patterns in the user's browsing process, and uses the frequent item set mining algorithm to mine the user's cursor behavior pattern, in order to accurately analyze the relationship between various layouts and links under user types, optimize the page layout, and improve user experience. Important reference.

Description of drawings

The accompanying drawings constituting a part of the present application are used to provide further understanding of the present application, and the schematic embodiments and descriptions of the present application are used to explain the present application, and do not constitute improper limitations to the present application.

Fig. 1 is the method flowchart among the present invention;

Fig. 2 is an example figure of the search engine result page in the behavioral data collection experiment of the present invention;

Fig. 3 is a domain-of-interest division diagram of the search result page area in the behavioral data collection experiment of the present invention:

Fig. 4 is the saccade path diagram of the eye movement behavior of the present invention;

Fig. 5 is the heat map of the eye movement behavior of the present invention;

Fig. 6 is a key performance indicator diagram of the eye movement behavior of the present invention;

Fig. 7 is an association rule diagram of clicking behaviors of each link under the action of cognitive factors in the present invention;

Fig. 8 is an association rule diagram of each link click behavior under the action of page layout factors in the present invention;

Fig. 9 is a schematic diagram of the system structure in the present invention.

Detailed ways:

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

It should be pointed out that the following detailed description is exemplary and intended to provide further explanation to the present application. Unless otherwise specified, all technical and scientific terms used in this embodiment have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

It should be noted that the terminology used here is only for describing specific implementations, and is not intended to limit the exemplary implementations according to the present application. As used herein, unless the context clearly dictates otherwise, the singular is intended to include the plural, and it should also be understood that when the terms "comprising" and/or "comprising" are used in this specification, they mean There are features, steps, operations, means, components and/or combinations thereof.

Aiming at the deficiencies in the prior art and solving the problems in the prior art on how to scientifically and accurately evaluate the user behavior pattern in the search engine result page and optimize the page layout to improve the user experience, the present invention provides a search result page user behavior The mode evaluation method, device and system, through various types of test behavior data, analyze common user behavior modes in different types of layouts and different types of users in search engine results pages, making important contributions to optimizing page layouts and improving link delivery effects.

In the case of no conflict, the embodiments in the present application and the features in the embodiments can be combined with each other. The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

Example 1:

The purpose of Embodiment 1 is to provide a method for evaluating user behavior patterns on a search result page.

In order to achieve the above object, the present invention adopts the following technical scheme:

As shown in Figure 1,

A method for evaluating user behavior patterns on a search result page, the method comprising:

Step (1): Receive user information, and classify user types according to the cognitive style in the user information;

Step (2): receiving at least two different types of test behavior data of the user in the search engine result page;

Step (3): visually process test behavior data and conduct preliminary qualitative analysis;

Step (4): Mining and analyzing different user types and user behavior patterns under different page layouts, and evaluating user behavior patterns on the search result page.

It should be noted that before receiving user information in step (1), users who are not suitable for eye movement collection are eliminated according to visual conditions, and a number of users are randomly selected from the remaining users to evaluate user behavior patterns on the search results page. The test users in the experiment are random people who have been screened for visual conditions, with uncorrected or corrected vision of 1.0 or more, and no visual diseases or obstacles that affect the data collection work of the eye tracking experiment.

In this embodiment, a total of 63 test users were recruited, the male to female ratio was 1:1.2, and the users were between 18 and 21 years old, with an average age of 19.7 years. In order to make the test results close to the actual situation, students of different majors are selected. In addition, all subjects were screened for visual conditions, and those with uncorrected or corrected vision less than 1.0, as well as those with color blindness and color weakness that affected the collection of experimental eye movement data were excluded.

In step (1), the user information includes user basic information and user cognitive information;

The basic user information includes the user's name, gender, age and occupation; the basic user information is collected by filling in a form;

The user cognition information is obtained by mosaic graph test method, including cognition style. Classify user types according to the cognitive style in user information.

In this embodiment, the category of the search task is randomly selected, covering different types as much as possible, such as: TV, amusement park, etc. The search engine selects the Chinese Baidu search engine. Considering the impact of network uncertainties, the search engine result pages corresponding to the search tasks are crawled, and then the search result pages are purified, that is, only ten result links and relevant recommendations on the right are kept. Then add advertising links on the top, bottom and right side of the result link list by manually adding. An example of the search result page of the present invention is shown in FIG. 2 . According to the link, it is divided into 12 interest areas (interest area 1-10 where the ten result links are located, interest area AD where the advertisement link is located, and interest area R where the relevant recommendation is located on the right). The division diagram of interest areas on the search result page of the present invention is shown in FIG. 3 . Among them, L1, L2 and L3 represent the commercial promotion layout situation where the advertisement is located above, below and on the right side of the result link list respectively.

In step (2), the test behavior data is the information source data generated during the browsing process of the user in the engine result page, including but not limited to cursor data and eye movement data; in this embodiment, the cursor data and eye movement data are collected. Behavioral information from these two sources of information is eye-tracking data.

The cursor data is a cursor event triggered by an input device obtained during the user's browsing of the search engine result page; Cursor data.

It should be noted that in this embodiment, the input device can trigger a series of cursor events for one element or multiple elements of the search result page, including but not limited to, for example, cursor left click, cursor right click, cursor Cursor events such as wheel scrolling or cursor movement to control cursor movement, or keyboard events such as enter confirmation events, menu pop-up events, or direction key scrolling events, or for example, touchpad sliding to control cursor movement, touchpad left button single touchpad events such as clicking or right-clicking on the touchpad, or, for example, for a touch-type terminal device, touchscreen events such as sliding to control page movement or touchscreen clicking.

The eye-movement data is eye-tracking information acquired during a user's browsing of a search engine result page. In this embodiment, the eye movement data generated during the process of browsing the search engine result page by the user is acquired by means of eye movement tracking.

In this embodiment, the eye tracking device is the SMI RED (Version 2.5) eye tracker developed by Pusheng Technology Co., Ltd. of Germany, and the selected sampling frequency is 120 Hz. It should be noted that when using this device, the subjects were required to fix their heads in the U-shaped bracket on the edge of the table, and they were not allowed to move their heads arbitrarily during the experiment. Before the formal experiment, the user has to perform eye movement calibration work twice, and the experiment can only be continued if the standards are met twice.

In this embodiment, IViewX, Experiment Center and BeGaze that come with the eye tracker are used for the recording of eye movement behavior and preliminary qualitative analysis, and Matlab R2014a is used for frequent pattern mining.

In this embodiment, the specific steps of collecting the test behavior data are:

Step (2-1) Pre-experiment preparation: It is necessary to introduce the basic steps of the experiment, the rules of using hardware equipment, specific precautions during the experiment and eye calibration to the user;

Step (2-2) collecting eye movement data: during the user's browsing of the search engine result page, the subject is required to obtain the eye movement data generated during the browsing process in accordance with the rules of use of the eye tracking device;

Step (2-3) collecting cursor behavior data: embed the corresponding Javascript code in the search engine result page that the user needs to browse, and record the cursor behavior data generated during the user's browsing process;

In this embodiment, the scenario design for collecting user behavior data is:

In order to make the experimental results more in line with the actual situation, the evaluation scene designed by the present invention is: given a certain amount of search tasks for the user, the subject understands the brief introduction of the search task through the screen display, presses the space bar after understanding the information to start browsing the search result page, browses The process is the same as real life, you can scroll, slide and click the cursor. Close the browser after browsing each search result page, and the next search task profile will pop up automatically. Use eye-tracking method to obtain the unconscious attention information of the user in the process of browsing the webpage, and obtain the user's cursor information through the JavaScript code embedded in the webpage. A design of 6 (search tasks)×3 (page layout)×2 (cognitive style) is adopted. In order to prevent user fatigue, the maximum number of search tasks defined here is 6, and users can stop at any time according to their own conditions.

In this embodiment, a total of 319 pieces of eye movement and cursor behavior information of users browsing the search result page are obtained, and the data summary is shown in Table 1.

Table 1

In step (3), the concrete steps of said visual processing test behavior data and carrying out preliminary qualitative analysis include:

Step (3-1): Data visualization of cursor data and eye movement data;

Step (3-2): Preliminary qualitative analysis of the impact of the page layout on the eye movement data.

In this embodiment, the eye movement data is visualized using BeGaze eye movement analysis software, such as the saccade path diagram shown in FIG. 4 , the heat zone diagram shown in FIG. 5 , and the key performance indicator diagram shown in FIG. 6 . Cursor data is displayed in an Excel table, and Table 2 shows a summary table of click behaviors. Preliminary qualitative analysis is performed through these visual data graphs. It can be found that users pay more attention to the top of the search engine results page, and the commercial promotion is more attractive to users, but the advertisement on the right is less attractive to users than the commercial promotion at the top and bottom.

Table 2

1 L1 Dependence 5 9 10 2 L1 Dependence 1 2 3 6 7 3 L1 Independence 2 5 10 4 L1 Independence 2 4 6 8 R 5 L1 Dependence 1 3 4 9 AD 6 L1 Independence 2 5 7 9 7 L1 Independence 1 3 4 8 L1 Dependence 2 9 AD 9 L1 Dependence 1 3 4 7 10 L2 Dependence 3 4 6 7 11 L2 Dependence 7 8 12 L2 Dependence 2 6 R 13 L2 Independence 2 3 AD 14 L2 Independence 3 4 6 8 9 AD 15 L2 Dependence 4 AD 16 L2 Independence 2 3 AD 17 L2 Dependence 2 4 18 L2 Independence 2 5 9 19 L3 Dependence 4 5 6 7 8 9 10 20 L3 Independence 1 2 3 4 7 8 10 twenty one L3 Independence 1 4 5 twenty two L3 Dependence 2 3 2 twenty three L3 Dependence 1 5 9 twenty four L3 Dependence 1 5 25 L3 Independence 1 2 5 7 R 26 L3 Independence 4 9

In step (4), the specific steps for analyzing the different types of behavior patterns of users under different user types and page layouts are analyzed, including:

Using the same frequent pattern mining method as the number of evaluation behavior data types,

Mining the cursor behavior patterns under different user types and page layouts, based on the cursor data, analyzing the correlation patterns between different links in the search results page and the cursor behaviors between user types and page layouts, as well as the cursor behaviors in links under different user types and page layouts Combination mode of parameters;

And mining the eye-movement browsing patterns under different user types and page layouts, according to the eye-movement data, the time-series information in the process of users browsing the search engine result pages is obtained, and the sequence relationship between the links in the process of users browsing the search engine result pages is analyzed.

In this embodiment, the frequent itemset mining algorithm is used to mine cursor behavior patterns under different user types and page layouts;

Using the directional frequent browsing pattern mining algorithm to mine eye movement browsing patterns under different user types and page layouts, the directional frequent browsing pattern mining algorithm is used to mine user directional and fixed-length eye movement browsing patterns under different page layouts to obtain user behavior processes Time series information in .

(4-1) Cursor behavior pattern analysis: use frequent pattern mining algorithm to analyze different links in the search results page (including ten result links, advertising links and related recommendations) and two factors (user type, namely user cognitive style, page layout) The correlation mode of the cursor behavior among the two factors, understand the link and combination mode closely related to the two factors, and establish the relationship between the cognitive mode, page layout and each link;

The two factors of user cognition and page layout are discussed separately. There are many cursor behavior parameters. Here we only take representative click characteristics as an example:

User type impact analysis (i.e. user cognition impact analysis): According to the click behavior data, classify the data according to the user cognition mode, mine the frequent item sets of click features in each category, and analyze the user cognition style and search engine result pages The association rules between the links in the website, and the combination mode of the click behavior parameters in the links under different cognitions.

Page layout impact analysis: According to the click behavior data, classify the data according to the page layout, that is, the advertisement position, mine the frequent item sets of the click characteristics in each category, and analyze the page layout and the association rules between the links in the search engine result page, Understand the combination mode of click behavior parameters in each link under different positions;

In this embodiment, multiple classic methods of frequent pattern mining are used to analyze the click behavior and cognitive style of different interest areas in the search engine result page, frequent item sets and association rules between advertisement positions. Since the results of different algorithms are similar, only the running results of the most classic frequent pattern mining Apriori algorithm are given. Figure 7 and Figure 8 respectively show the calculated association rules (minSup=0.03, minConf=0.4) between the click behavior of each interest area on the search engine result page and different cognitive styles and advertisement positions. For the convenience of presentation, the present invention visualizes the connection between items in the association rule as a directed graph. First, it is necessary to generate an adjacency matrix of regular antecedents and subsequent items. This matrix is a sparse matrix, and the values of the elements in the matrix are the correlation degrees between antecedents and subsequent items. Then, let the antecedents in the association rules be the predecessors of the directed graph, and the latter items in the association rules be the consequents of the corresponding antecedents in the directed graph. Taking the bold connection line in Figure 8 as an example, it means that {7}{2}→{Independence}, {7}→{Independence}, {7}{2}→{Dependence} are all association rules that satisfy the threshold.

Under the influence of user cognition factors, the dimensions of the items in the frequent itemset of field-independent cognitive style click behavior are significantly larger than those of field-dependent cognitive style. This shows that field-independent users have a strong subjective initiative when browsing the web, are not limited to a certain range, and have a strong overall awareness; field-dependent users have relatively fixed browsing areas and patterns in the process.

Analyzing the effects of page layout factors, it can be seen that the dimensions of items in the frequent itemset are significantly larger when the advertisement is located in L1 and L3 than when the advertisement is located in L2. The layout format at L1 affects the most items and has a greater influence on the advertisement. The layout format at L2 affects the least items, but has a greater influence on the ad. This echoes the conclusion of the preliminary qualitative analysis in step (3).

(4-2) Analysis of eye-movement browsing patterns: This embodiment proposes a directional frequent browsing pattern mining algorithm DFBP (Directional Frequent Browsing Patterns), which is used to mine user-directed and fixed-length browsing patterns under different page layouts, and understand user behavior in the process. Time-series information, combined with the actual situation to analyze the sequence relationship between the links in the user's browsing process;

In said step (4-2), the specific steps of the DFBP algorithm of eye movement browsing pattern analysis are:

Step (4-2-1): Roughly classify the collected data according to the location of the advertisement. Each category is arranged according to the order of the time when the collected users first entered each area of interest. Each user views each Each web page corresponds to a browsing mode sequence Q _i ;

For example: ad placement, including: the top of the web page, the middle of the web page, or the bottom of the web page;

Each page viewed by a user corresponds to a sequence of browsing patterns:

For example, open Baidu, enter the search term, there are 10 search links corresponding to the search term, and the numbers are 1-10 respectively. For the advertisement area, it is defined as AD, and the area related to the search term is defined as R;

Assume that the browsing order of the user's eyes is AD, 1, 2, 3, 12, 4, 5, 6, 8, 7, 9, 10; then the user's browsing pattern sequence is: 11→1→2→3→ 12→4→5→6→8→7→9→10;

Assume that the browsing order of the user's eyes is 1, 2, 3, AD, 4, 5, 6, 7, 8, 9, 10, R; then the user's browsing mode sequence is: 1→2→3→AD→ 4→5→6→7→8→9→10→R;

Assume that the browsing order of the user's eyes is AD, 2, 3, 4, 5; then the user's browsing pattern sequence is: AD→2→3→4→5;

Step (4-2-2): Add three attributes to all the data obtained in step (4-2-1) and initialize: adopted length L=(l ₁ ,l ₂ ,…,l _p ), support sequence S=(s ₁ ,s ₂ ,…,s _p ), l ₁ =0, s ₁ =0, support threshold s;

Element support, for example:

The support degree of an element is recorded as the frequency of the element. When there are 300 pieces of data in a data set, there are 30 pieces of data whose first element is 11. At this time, the support degree of the first element is 30; on the premise that the first element is consistent, the second element There are 10 items for 1, and the support degree of the second element is 10 at this time; on the premise that the first two elements are consistent, there are 7 items for the third element of 2, and the support degree of the third element is 7 at this time , and so on.

Accepted length, for example:

Adopted length, understand the specific length information of a sequence with frequent conditions. The sequence calculates its support from the first element, if it is greater than the support threshold, then l ₁ =1, L=(1), continue, otherwise l ₁ =1, L=(0), end; when l ₁ =1 , calculate the support degree of the second element, if it is greater than the support degree threshold, then l ₂ =2, L=(1,2), continue, otherwise keep l ₁ =1, L=(1) unchanged, end; When l ₂ =2, calculate the support of the third element, if it is greater than the support threshold, then l ₃ =3, L=(1,2,3), repeat this step, otherwise keep it unchanged, and end.

For example, for the browsing pattern sequence 11→1→2→3→12→4→5→6→8→7→9→10, the support threshold s=8. The support degree of the first element AD is 30>8, then l ₁ =1, L=(1); the support degree of the second element 1 is 10>8, then l ₂ =2, L=(1,2); The support degree of the third element 2 is 7<8, then keep l ₂ =2, L=(1,2) unchanged, and end.

Support sequence, for example:

Support sequence, to understand the specific frequency information of a sequence. Calculate the support degree of the first element from the first element. When there are 300 pieces of data in a data set, and the first element is AD, there are 30 pieces. At this time, the support degree of the first element s ₁ =30, S=(30); On the premise that the first element is consistent, there are 10 items whose second element is 1. At this time, the support degree of the second element is s ₂ =10, S=(30,10); on the premise that the first two elements are consistent, There are 7 items with the third element being 2. At this time, the support degree of the third element is s ₃ =7, 7<s, end, S=(30,10) remains unchanged, if the third element satisfies greater than And so on for the support threshold.

Frequency threshold, for example:

The frequency threshold is a value that is more suitable for the size of the current data set, which is debugged through experience. When the support of an element is greater than this value, it is considered that the element is frequent, here let s=8.

Step (4-2-3): Calculate the support degree s _j of the first element of the sequence, if s _j <s, set the sequence s _j =0 and delete the sequence;

For example, if the first elements of 30 pieces of data in the current data set are the same, such as AD, then the support of the first element s _j =30

Calculate the support s _j of the first element of the sequence

Step (4-2-4): Sort the remaining sequences according to the value of the first element from large to small, create a queue G ₁ , G ₂ ,…,G _t corresponding to each first element after sorting, and sort the sequence by category Enter different queues and delete the first element of each sequence;

Step (4-2-5): Update the L and S attributes of the sequence, add l _j+1 =l _j +1, s _j+1 for L and S respectively, the original L=(l ₁ ,l ₂ ,…, l _j ), S=(s ₁ ,s ₂ ,…,s _j ), after updating, it becomes L=(l ₁ ,l ₂ ,…,l _j ,l _j+1 ), S=(s ₁ ,s ₂ ,...,s _j ,s _j+1 );

Step (4-2-6): Repeat steps (4-2-3) and (4-2-4) until the elements in each sequence end with support equal to 0;

Step (4-2-7): Calculate the score F _i =l _i *s _i of each sequence, find the maximum score maxF _i from F _i , the sequence corresponding to the maximum score is a frequent browsing pattern sequence, and output the browsing pattern sequence Q _i , otherwise, it is determined that the sequence is a non-frequent browsing pattern sequence.

As shown in Table 3, the five most common browsing patterns of users obtained by using the DFBP algorithm under the three layouts. Observing the browsing sequence under the L1 layout, it is found that if the user detects an advertisement at the beginning of browsing, the user will first detect the advertisement and then browse in order; if the user does not detect the advertisement at the beginning, the user will browse the URL in order. In the L2 layout, the advertisement is located at the bottom, and users naturally browse in order. In the L3 layout, the advertisement is on the right side. Users usually observe the URL first according to the habit, but usually the advertisement picture on the right side is large, the color is bright, and the stimulus is strong. Therefore, after URL 1 and 2, the user’s attention will be attracted by the advertisement on the right side.

table 3

Observing the three different layouts, it can be seen that under different layouts, the user browsing mode is roughly top-down. And it can also be found that in many cases, the area that the user first looks at is the URL of No. 2, and then backtracks to No. 1. Combined with reality, this situation should be due to the fact that when users use computers to search and browse, there will be a certain time interval when opening a new webpage, and this interval will cause the user's sight to reset to the middle area of the screen. The results are mainly arranged on the left side of the page, so the user first pays attention to the URL at No. 2, and after understanding, the user will read the process in order again.

In this embodiment, the eye tracking method is used to obtain the unconscious attention information of the user in the process of browsing the webpage, and the user's cursor information is obtained through the JavaScript code embedded in the webpage. The present invention proposes a novel, directional and fixed-length directional frequent browsing pattern mining algorithm DFBP for mining frequent sequential patterns, and uses the frequent itemset mining algorithm to mine user cursor behavior patterns. Through eye movement features and cursor features, the analysis of common user behavior patterns in different types of layouts and different types of users in search engine results pages has made important contributions to optimizing page layouts and improving link delivery effects.

Example 2:

The purpose of Embodiment 2 is to provide a computer-readable storage medium.

In order to achieve the above object, the present invention adopts the following technical scheme:

A computer-readable storage medium, in which a plurality of instructions are stored, and the instructions are adapted to be loaded by a processor of a terminal device and perform the following processing:

Receive user information, and classify user types according to the cognitive style in the user information;

receiving at least two different types of test behavior data of users within search engine results pages;

Visually process test behavior data and conduct preliminary qualitative analysis;

The behavior patterns of different types of users and different types of users under the page layout are respectively excavated for analysis, and the user behavior patterns of search result pages are evaluated.

In this embodiment, examples of the computer-readable recording medium include magnetic storage media (for example, ROM, RAM, USB, floppy disk, hard disk, etc.), optical recording media (for example, CD-ROM or DVD), PC interfaces (for example, PCI, PCI-Express, WiFi, etc.), etc. However, aspects of the present disclosure are not limited thereto.

Example 3:

The purpose of Embodiment 3 is to provide a device for evaluating user behavior patterns on a search result page.

In order to achieve the above object, the present invention adopts the following technical scheme:

A device for evaluating user behavior patterns on a search result page, using an Internet terminal device, including a processor and a computer-readable storage medium, the processor is used to implement instructions; the computer-readable storage medium is used to store multiple instructions, and the instructions is adapted to be loaded by the processor and perform the following processing:

Receive user information, and classify user types according to the cognitive style in the user information;

receiving at least two different types of test behavior data of users within search engine results pages;

Visually process test behavior data and conduct preliminary qualitative analysis;

The behavior patterns of different types of users and different types of users under the page layout are respectively excavated for analysis, and the user behavior patterns of search result pages are evaluated.

Those skilled in the art should understand that each module or each step of the present invention described above can be realized by a general-purpose computer device, and optionally, they can be realized by a program code executable by the computing device, thereby, they can be stored in The storage device is executed by the computing device, or they are manufactured as individual integrated circuit modules, or multiple modules or steps among them are manufactured as a single integrated circuit module. The invention is not limited to any specific combination of hardware and software.

Example 4:

The purpose of Embodiment 4 is to provide a system for evaluating user behavior patterns on a search result page.

In order to achieve the above object, the present invention adopts the following technical scheme:

As shown in Figure 9,

A search result page user behavior pattern assessment system, the system is based on the above search result page user behavior pattern assessment method, comprising: a user information collection device, a behavior data collection device and a behavior pattern assessment device;

The user information collection device is used to collect user information and send it to the behavior pattern evaluation device;

The behavior data collection device is used to collect at least two different types of test behavior data of the user in the search engine result page, and send it to the behavior pattern evaluation device;

The behavior pattern evaluation device is used to receive user information, classify user types according to the cognitive style in the user information; receive at least two different types of test behavior data of the user in the search engine result page; visually process the test behavior data and Carry out preliminary qualitative analysis; excavate different user types and different types of user behavior patterns under the page layout for analysis, and evaluate the user behavior patterns of the search results page.

The user information collection device collects basic user information in the form of table statistics, and obtains user cognition information through a mosaic pattern test.

The behavioral data collection device specifically includes:

The eye movement data acquisition module is configured to use the eye movement tracking device to obtain the eye movement data generated during the user's browsing of the search engine result page; in this embodiment, the eye movement tracking device is SMIRED developed by Germany Pusheng Technology Co., Ltd. (Version2.5) eye tracker, the selected sampling frequency is 120Hz.

The cursor data collection module is configured to embed JavaScript code in the search engine result page to obtain cursor data generated during the user's browsing of the search engine result page.

The behavior pattern evaluation device specifically includes:

Preliminary qualitative analysis module: Visualize the obtained data, and conduct preliminary qualitative analysis of the mapping relationship between different factors and links in the search result page under different information sources through the obtained visualized data graph;

The cursor behavior pattern analysis module is configured to use the frequent itemset mining algorithm to mine the common cursor behavior patterns under different user types and different layouts through the captured cursor data, and explain the results in combination with the actual situation;

The eye movement behavior pattern analysis module is configured to use the frequent browsing pattern mining algorithm based on the obtained eye movement data to mine the common browsing patterns of users with time sequence under different user types and different layouts, and analyze the user's cognitive mode and various links The combination relationship among them, and combined with the actual explanation results.

Beneficial effects of the present invention:

(1) A search result page user behavior pattern evaluation method, device and system according to the present invention, obtain at least two different types of test behavior data of the user in the search engine result page, for example, the user's eye movement data and cursor data, mining frequent behavior patterns through at least two information source characteristics generated during the user’s browsing process in the search engine result page, and analyzing the combination mode between page layout, user type, search engine result page and various links, As well as the timing relationship, the present invention is of great significance for improving the link layout mode in the search engine result page, personalized recommendation information, and advertisement delivery effect.

(2) A search result page user behavior pattern evaluation method, device and system according to the present invention collects behavior information of various information sources in the search engine result page, and proposes a novel, directional and fixed-length frequent browsing The pattern mining algorithm mines the frequent sequential patterns in the user's browsing process, and uses the frequent item set mining algorithm to mine the user's cursor behavior pattern, in order to accurately analyze the relationship between various layouts and links under user types, optimize the page layout, and improve user experience. Important reference.

The above descriptions are only preferred embodiments of the present application, and are not intended to limit the present application. For those skilled in the art, there may be various modifications and changes in the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of this application shall be included within the protection scope of this application. Therefore, the present invention will not be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (14)

1. A search result page user behavior pattern evaluation method, is characterized in that, the method comprises: Receive user information, and classify user types according to the cognitive style in the user information; receiving at least two different types of test behavior data of users within search engine results pages; Visually process test behavior data and conduct preliminary qualitative analysis; Mining different user types and different types of user behavior patterns under the page layout for analysis, and evaluating user behavior patterns on the search result page; different types of user behavior patterns include cursor behavior patterns and eye movement browsing patterns; frequent itemset mining algorithms are used to mine Cursor behavior patterns under different user types and page layouts; Using the directional frequent browsing pattern mining algorithm to mine eye movement browsing patterns under different user types and page layouts, the directional frequent browsing pattern mining algorithm is used to mine user directional and fixed-length eye movement browsing patterns under different page layouts to obtain user behavior processes Time series information in . 2. The method according to claim 1, wherein before receiving user information, users who are not suitable for eye movement collection are eliminated according to visual conditions, and a number of users are randomly selected from the remaining users to perform user behavior patterns on the search result page Evaluation. 3. The method of claim 1, wherein, The user information includes basic user information and user cognitive information; The basic user information includes the user's name, gender, age and occupation; The user cognition information is obtained by mosaic graph test method, including cognition style. 4. The method according to claim 1, wherein the test behavior data is information source data generated during the user's browsing process in the engine result page, including but not limited to cursor data and eye movement data; The cursor data is a cursor event triggered by an input device obtained during the user's browsing of the search engine result page; The eye-movement data is eye-tracking information acquired during a user's browsing of a search engine result page. 5. method as claimed in claim 4, is characterized in that, the concrete step of described visual processing test behavior data and carrying out preliminary qualitative analysis comprises: Data visualization of cursor data and eye movement data; Preliminary qualitative analysis of the impact of page layout on the eye movement data. 6. The method according to claim 5, characterized in that, the specific steps of analyzing the different types of behavior patterns of users under different user types and page layouts respectively include: Using the same frequent pattern mining method as the number of evaluation behavior data types, Mining the cursor behavior patterns under different user types and page layouts, based on the cursor data, analyzing the correlation patterns between different links in the search results page and the cursor behaviors between user types and page layouts, as well as the cursor behaviors in links under different user types and page layouts Combination mode of parameters; And mining the eye-movement browsing patterns under different user types and page layouts, according to the eye-movement data, the time-series information in the process of users browsing the search engine result pages is obtained, and the sequence relationship between the links in the process of users browsing the search engine result pages is analyzed. 7. The method according to claim 1, wherein the directional frequent browsing pattern mining method comprises: Obtain the browsing sequence data of the user on the search engine result page according to the test behavior data; Add the adoption length and its corresponding support attribute in the browsing sequence data, and initialize it; Process the support of each browsing sequence to make it zero and get a new sequence; Determine whether the new sequence is a frequent sequence, and output the frequent sequence. 8. The method of claim 7, wherein, The range of the link area is preset, and the time when the user enters the range of each link area is extracted from the test behavior data. 9. The method of claim 7, wherein, The specific steps of obtaining the browsing sequence of the user on the search engine result page according to the test behavior data include: Roughly classifying the test behavior data according to the page layout; Arrange according to the order of time when users enter the area of each link, and each user views each web page corresponding to a piece of browsing sequence data. 10. The method of claim 7, wherein, The specific steps of processing the support of each browsing sequence to make it zero include: Preset support threshold; Calculate the support of the first element of the browsing sequence data, set the support of the first element less than the support threshold to zero and eliminate the sequence; Sort and classify the browsing sequence data according to the value of the first element, create a queue corresponding to it, enter the sequence into different queues by category, and delete the first element of each sequence; Update the adopted length of the sequence and its corresponding support attribute until the elements in each sequence end with support equal to 0. 11. The method of claim 7, wherein, The specific steps for judging whether a new sequence is a frequent sequence are: Calculating the score of the new sequence, the score being the product of the adopted length of the sequence and the corresponding support degree; Sort the scores to determine the maximum score; The new sequence corresponding to the maximum score is a frequent browsing pattern sequence; otherwise, it is determined that the new sequence is a non-frequent browsing pattern sequence. 12. A computer-readable storage medium, wherein a plurality of instructions are stored, wherein the instructions are adapted to be loaded by a processor of a terminal device and perform the following processing: Receive user information, and classify user types according to the cognitive style in the user information; receiving at least two different types of test behavior data of users within search engine results pages; Visually process test behavior data and conduct preliminary qualitative analysis; Mining different user types and different types of user behavior patterns under the page layout for analysis, and evaluating user behavior patterns on the search result page; different types of user behavior patterns include cursor behavior patterns and eye movement browsing patterns; frequent itemset mining algorithms are used to mine Cursor behavior patterns under different user types and page layouts; Using the directional frequent browsing pattern mining algorithm to mine eye movement browsing patterns under different user types and page layouts, the directional frequent browsing pattern mining algorithm is used to mine user directional and fixed-length eye movement browsing patterns under different page layouts to obtain user behavior processes Time series information in . 13. A search result page user behavior mode evaluation device, using Internet terminal equipment, including a processor and a computer-readable storage medium, the processor is used to implement each instruction; the computer-readable storage medium is used to store multiple instructions, characterized in that , the instructions are adapted to be loaded by the processor and perform the following processing: Receive user information, and classify user types according to the cognitive style in the user information; receiving at least two different types of test behavior data of users within search engine results pages; Visually process test behavior data and conduct preliminary qualitative analysis; Mining different user types and different types of user behavior patterns under the page layout for analysis, and evaluating user behavior patterns on the search result page; different types of user behavior patterns include cursor behavior patterns and eye movement browsing patterns; frequent itemset mining algorithms are used to mine Cursor behavior patterns under different user types and page layouts; Using the directional frequent browsing pattern mining algorithm to mine eye movement browsing patterns under different user types and page layouts, the directional frequent browsing pattern mining algorithm is used to mine user directional and fixed-length eye movement browsing patterns under different page layouts to obtain user behavior processes Time series information in . 14. A search result page user behavior pattern evaluation system, a search result page user behavior pattern evaluation method as claimed in any one of claims 1-11, characterized in that it comprises: a user information collection device, a behavior data collection devices and behavioral pattern assessment devices; The user information collection device is used to collect user information and send it to the behavior pattern evaluation device; The behavior data collection device is used to collect at least two different types of test behavior data of the user in the search engine result page, and send it to the behavior pattern evaluation device; The behavior pattern evaluation device is used to receive user information, classify user types according to the cognitive style in the user information; receive at least two different types of test behavior data of the user in the search engine result page; visually process the test behavior data and Carry out preliminary qualitative analysis; excavate different user types and different types of user behavior patterns under the page layout for analysis, and evaluate the user behavior patterns of the search results page.