User operation trajectory analysis method, device, equipment and storage medium

Abstract

The present invention discloses a user operation trajectory analysis method, apparatus, device, and storage medium, which are applied to the field of computer interaction technology. The method comprises: obtaining operational behavior information of users accessing system pages; the operational behavior information includes at least: page information, information about operational components corresponding to the page interactions, and information about regions corresponding to the operational components; constructing a dynamic trajectory sample space based on the operational behavior information; characterizing and stratifying the dynamic trajectory sample space to obtain stratification results; fitting the user operation trajectory based on the stratification results to obtain a target operation trajectory. The present invention characterizes and stratifies the dynamic trajectory sample space and fits the user operation trajectory based on the stratification results to extract high-frequency operational behavior trajectories of different types of users, thereby greatly improving the accuracy of data analysis.

Description

Technical Field

[0001] The present invention relates to the field of computer interaction technology, and in particular to a method, device, equipment and storage medium for analyzing user operation traces. Background Art

[0002] With the development of Internet technology, the analysis of user online behavior trajectory data has gained attention and is being applied in various fields. For example, different layout pages are generated for different types of user groups or used to identify machine clicks. Existing user behavior analysis technology solutions usually adopt the method of designing different versions of pages in the early stage of product development or implementing it after the product is launched by matching user IDs or user's region, gender, behavior operation and other related attributes. This technology cannot analyze the behavior operation trajectory involving multiple pages in a complete user task process. In addition, it matches a behavior operation sample to a pre-set pre-simulated behavior trajectory after hierarchical clustering. There are errors in the integration of this behavior trajectory with the pre-set behavior trajectory, making it difficult to extract different high-frequency behavior paths for different user groups. How to solve this problem? There is currently no effective solution. Summary of the Invention

[0003] In view of this, the present invention is intended to provide a method, apparatus, device and storage medium for analyzing user operation traces.

[0004] The technical embodiment of the embodiment of the present invention is implemented as follows:

[0005] An embodiment of the present invention provides a method for analyzing user operation traces, the method comprising:

[0006] Obtaining operation behavior information of a user accessing a system page; the operation behavior information includes at least: page information, operation component information corresponding to the interaction with the page, and area information corresponding to the operation component;

[0007] Constructing a dynamic trajectory sample space based on the operation behavior information;

[0008] Characterizing and stratifying the dynamic trajectory sample space to obtain a stratification result;

[0009] The user operation trajectory is fitted according to the stratification result to obtain a target operation trajectory.

[0010] In the above solution, the method further includes:

[0011] Obtaining attribute information of the user;

[0012] The user type is determined according to the attribute information.

[0013] In the above solution, constructing a dynamic trajectory sample space based on the operation behavior information includes:

[0014] Determining a sample size of the user operation trajectory based on the user type and the operation behavior information;

[0015] A dynamic trajectory sample space is constructed according to the sample size.

[0016] In the above solution, determining the sample size of the user operation trajectory based on the user type and the operation behavior information includes:

[0017] Determining first operation behavior information corresponding to a visit completed by a first user type based on the user type and the operation behavior information; the first user type is any user type among the user types;

[0018] determining a total number of operations according to the first operation behavior information;

[0019] The first page information, the first operation component information, and the first region information corresponding to the first operation number are combined into a first feature vector; the first operation number is any operation number in the total operation number;

[0020] determining a first sample of the user operation trajectory based on a plurality of the first feature vectors;

[0021] The sample size of the user operation trajectory is determined according to the plurality of first samples.

[0022] In the above solution, the characterization and stratification of the dynamic trajectory sample space to obtain the stratification results includes:

[0023] Obtaining a first characteristic element according to the dynamic trajectory sample space; the first characteristic element represents any one of the characteristic elements of the same type in the plurality of first characteristic vectors;

[0024] Performing cluster analysis on the plurality of first characteristic elements according to the type of the user to obtain a plurality of cluster groups; the cluster groups represent the similarity of the plurality of first characteristic elements;

[0025] The plurality of clusters are grouped as a hierarchical result.

[0026] In the above solution, cluster analysis is performed on the plurality of first characteristic elements according to the type of the user to obtain a plurality of cluster groups, including:

[0027] determining, according to the type of the user, a similarity between any two of the plurality of first characteristic elements;

[0028] Clustering is performed according to the similarity to obtain multiple cluster groups.

[0029] In the above solution, fitting the user operation trajectory according to the stratification result to obtain the target operation trajectory includes:

[0030] Performing time series fitting on the user operation trajectory according to the stratification results to obtain a behavior sequence with the highest probability of page jump corresponding to the user operation;

[0031] Determine a target feature vector based on the behavior sequence; the target feature vector includes target page information, target operation component information, and target area information;

[0032] The spatial influence factor is fitted to the target feature vector to obtain the target operation trajectory.

[0033] In the above solution, performing spatial influence factor fitting on the target feature vector to obtain the target operation trajectory includes:

[0034] The target feature vector is corrected by using preset model parameters to obtain a corrected target feature vector;

[0035] Perform feature vector centroid correction on the corrected target feature vector to obtain a target operation trajectory.

[0036] An embodiment of the present invention provides a user operation trajectory analysis device, comprising: an acquisition unit, a construction unit, a stratification unit, and a fitting unit, wherein:

[0037] The obtaining unit is used to obtain operation behavior information of the user accessing the system page; the operation behavior information at least includes: page information, operation component information corresponding to the interaction with the page, and area information corresponding to the operation component;

[0038] The construction unit is configured to construct a dynamic trajectory sample space based on the operation behavior information;

[0039] The stratification unit is used to perform characteristic stratification on the dynamic trajectory sample space to obtain a stratification result;

[0040] The fitting unit is configured to fit the user operation trajectory according to the stratification result to obtain a target operation trajectory.

[0041] In the above solution, the obtaining unit is further configured to obtain attribute information of the user; and determine the user type according to the attribute information.

[0042] In the above solution, the construction unit is further configured to determine a sample size of the user operation trajectory based on the user type and the operation behavior information; and construct a dynamic trajectory sample space according to the sample size.

[0043] In the above scheme, the construction unit is also used to determine the first operation behavior information corresponding to a visit completed by the first user type based on the user type and the operation behavior information; the first user type is any user type among the user types; the total number of operations is determined according to the first operation behavior information; the first page information, first operation component information, and first area information corresponding to the first number of operations are combined into a first feature vector; the first number of operations is any operation number in the total number of operations; a first sample of the user operation trajectory is determined based on multiple first feature vectors; and the sample size of the user operation trajectory is determined based on multiple first samples.

[0044] In the above scheme, the stratification unit is further used to obtain a first feature element based on the dynamic trajectory sample space; the first feature element represents any one of the feature elements of the same type in the multiple first feature vectors; cluster analysis is performed on the multiple first feature elements according to the type of the user to obtain multiple cluster groups; the cluster grouping represents the similarity of the multiple first feature elements; and the multiple cluster groups are used as stratification results.

[0045] In the above solution, the stratification unit is further configured to determine the similarity between any two of the plurality of first feature elements according to the type of the user; and perform clustering according to the similarity to obtain a plurality of cluster groups.

[0046] In the above scheme, the fitting unit is also used to perform time-series fitting on the user operation trajectory according to the stratification results to obtain a behavior sequence with the highest probability of page jump corresponding to the user operation; determine the target feature vector based on the behavior sequence; the target feature vector includes target page information, target operation component information, and target area information; perform spatial influence factor fitting on the target feature vector to obtain the target operation trajectory.

[0047] In the above solution, the fitting unit is further used to correct the target feature vector using preset model parameters to obtain a corrected target feature vector; and perform feature vector centroid correction on the corrected target feature vector to obtain a target operation trajectory.

[0048] An embodiment of the present invention provides a user operation trajectory analysis device, including a memory and a processor, wherein the memory stores a program that can be run on the processor, and when the processor executes the program, any step of the above-mentioned method is implemented.

[0049] An embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, any step of the above-described method is implemented.

[0050] The present invention provides a method, apparatus, device, and storage medium for analyzing user operation trajectories, wherein the method includes obtaining operational behavior information of users accessing system pages; the operational behavior information includes at least: page information, information about operational components corresponding to interactions with the page, and area information corresponding to the operational components; constructing a dynamic trajectory sample space based on the operational behavior information; characterizing and stratifying the dynamic trajectory sample space to obtain a stratification result; and fitting the user operation trajectory based on the stratification result to obtain a target operation trajectory. The technical solution of the embodiment of the present invention is adopted, adding page variables, characterizing and stratifying the behavioral trajectory routes, and performing trajectory fitting on the operation sequence and operation space based on the stratification results to extract high-frequency operational behavior trajectories of different types of users, thereby greatly improving the accuracy of data analysis and effectively reducing data analysis errors. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is a schematic diagram of a process for implementing a method for analyzing user operation traces according to an embodiment of the present invention;

[0052] Figure 2 A schematic diagram of extracting a user behavior trajectory model in a user operation trajectory analysis method according to an embodiment of the present invention;

[0053] Figure 3 Schematic diagram of the initial state of feature P in the user operation trajectory analysis method according to an embodiment of the present invention;

[0054] Figure 4 This is a schematic diagram of the structure of a user operation trajectory analysis device according to an embodiment of the present invention;

[0055] Figure 5 Schematic diagram of a hardware entity structure of a user operation trajectory analysis device in an embodiment of the present invention. DETAILED DESCRIPTION

[0056] To make the purpose, technical solutions and advantages of the embodiments of the present invention more clear, the specific technical solutions of the invention will be described in further detail below in conjunction with the accompanying drawings in the embodiments of the present invention. The following embodiments are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0057] This embodiment proposes a method for analyzing user operation traces. Figure 1 FIG. 1 is a flow chart illustrating a method for analyzing user operation traces according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0058] Step S101: obtaining operation behavior information of a user accessing a system page; the operation behavior information at least includes: page information, operation component information corresponding to the interaction with the page, and area information corresponding to the operation component.

[0059] It should be noted that obtaining the operational behavior information of a user accessing a system page may be performed by using a preset technology; wherein the preset technology may be determined based on actual conditions and is not limited here. As an example, the preset technology may be a web page embedding technology. The operational behavior information of a user accessing a system page may be the operational behavior information of a user accessing multiple pages in a complete task flow; the operational behavior information may include multiple user operations.

[0060] The operation behavior information includes at least: page information, operation component information corresponding to the page interaction, and area information corresponding to the operation component; wherein, the page information can be understood as the page currently visited by the user, and the system is composed of multiple pages; the page information can be recorded as Page, abbreviated as P; for ease of understanding, here is an example, the first page in the system is page1, recorded as 0001, the second page in the system is page2, recorded as 0010... The operation component information corresponding to the page interaction can be determined according to actual conditions and is not limited here. As an example, the operation component information can be information about human-computer interaction operations such as clicking a button, entering text, and dragging. For ease of understanding, the operation component information can be recorded as a. In actual applications, the components of the human-computer interaction behavior and the corresponding page can be numbered. For example, the click operation is 1, and further, the first button component in the first page is page1_click1, recorded as 0001, and the second button component in the first page is page1_click2, recorded as 0010...; the input operation is 2, and further, the first input component in the first page is page1_input1, and the second input component in the first page is page1_input2, .... The area information corresponding to the operation component can be recorded as L, which can represent the corresponding area of ​​the user operation behavior. The corresponding area can be described by coordinates, for example, it can include (x, y), the screen coordinate value of the corresponding operation area; for example, if the screen is 1920*1080px, it is recorded as 11110000000,10000111000.

[0061] Step S102: constructing a dynamic trajectory sample space based on the operation behavior information.

[0062] It should be noted that constructing a dynamic trajectory sample space based on the operation behavior information can be understood as obtaining the operation pages, operation behaviors, and sequence along the trajectory based on the operation behavior information; and constructing a dynamic trajectory sample space according to the operation pages, operation behaviors, and sequence.

[0063] Step S103: performing feature stratification on the dynamic trajectory sample space to obtain a stratification result.

[0064] It should be noted that the dynamic trajectory sample space is characterized and stratified to obtain the stratification result. A preset algorithm can be used to characterize and stratify the dynamic trajectory sample space to obtain the stratification result. The preset algorithm can be determined according to actual conditions and is not limited here. As an example, the preset algorithm is a hierarchical clustering algorithm. The stratification result can be a classification of the importance and / or abnormality of the operation behavior. As an example, the stratification result can be a high-frequency operation behavior, a secondary operation behavior, an abnormal operation behavior, etc.

[0065] Step S104: fitting the user operation trajectory according to the stratification result to obtain a target operation trajectory.

[0066] Here, fitting the user operation trajectory according to the stratification result to obtain the target operation trajectory can be performed by performing a first fitting and a second fitting on the user operation trajectory according to the stratification result to obtain the target operation trajectory. The first fitting can be an operation timing fitting; the second fitting can be an operation space fitting. Both the operation timing fitting and the operation space fitting can be determined according to actual conditions and are not limited here. As an example, the operation timing fitting can be constructed by using the user's historical trajectory data to construct a feature timing prior condition probability matrix to obtain the behavior sequence with the highest probability; the operation space fitting can be performed by using the operation space information to perform centroid correction based on the operation timing information, so as to achieve the effect of improving the accuracy of the model fitting behavior trajectory curve.

[0067] The user operation trajectory analysis method provided by the embodiment of the present invention adds page variables, characterizes and layers the behavior trajectory routes, performs trajectory fitting on the operation sequence and operation space based on the layering results, and extracts high-frequency operation behavior trajectories of different types of users, thereby greatly improving the accuracy of data analysis and effectively reducing data analysis errors.

[0068] In an optional embodiment of the present invention, the method further includes: obtaining attribute information of the user; and determining the user type according to the attribute information.

[0069] It should be noted that the attribute information can be determined according to actual conditions and is not limited here. As an example, the attribute information can be attribute information representing the user's basic situation, such as the user's region, gender, age, income, etc. In actual applications, the user's region can be distinguished based on the device access IP address, and regional labels can be added to users in different regions. The number of times the user visits the system can be used to distinguish whether the user is a novice user or an experienced user. For example, the novice user can be defined as a first-time visitor; the experienced user can be defined as a user who has used the product for more than half a year and visits more than 2 times a day on average.

[0070] Obtaining attribute information of the user; determining the user type based on the attribute information can be understood as labeling different types of users accessing the system page to identify different user types.

[0071] The embodiment of the present invention can label different types of users in the system and record the sequence of their operation behaviors.

[0072] In an optional embodiment of the present invention, constructing a dynamic trajectory sample space based on the operation behavior information includes: determining a sample size of the user operation trajectory based on the user type and the operation behavior information; and constructing a dynamic trajectory sample space according to the sample size.

[0073] In an embodiment of the present invention, a sample size of the user operation trajectory is determined based on the user type and the operation behavior information; constructing a dynamic trajectory sample space based on the sample size can be to determine users of the same type based on the user type, determine the sample size of the user operation trajectory based on the users of the same type and the corresponding operation behavior information; and construct a dynamic trajectory sample space matrix based on the sample size.

[0074] For ease of understanding, here is an example. Assume that the complete behavior trajectory of a single user in the i-th task process is recorded as T i , T i It can be expressed as the following formula (1), where j is the operation order, 1≤j≤n, n is the total number of operations in a task flow; i is the i-th task flow, 1≤i≤m, and m is the sample size of the user operation behavior trajectory.

[0075]

[0076] In practical applications, determining users of the same type based on the user type can be determining senior user types based on various user types; determining the sample size of the user operation trajectory based on the users of the same type and the corresponding operation behavior information can be determining the sample size of the user operation trajectory based on senior user types and the corresponding operation behavior information; and constructing a dynamic trajectory sample space matrix based on the sample size.

[0077] For ease of understanding, we take m behavior trajectory samples of an experienced user as an example. m behavior trajectory samples can be understood as m complete access task processes of this type of experienced user. m can be determined based on actual conditions and can also be called the sample size of the user's operation behavior trajectory. For example:

[0078] T1=(P1,a1,L1), where P1={P 11 , P 12 ,…P 1j}, a1={a 11 , a 12 ,…a 1j},L1={L 11 , L 12 ,…L 1j};

[0079] T2=(P2,a2,L2), where P2={P 21 , P 22 ,…P 2j}, a2={a 21 , a 22 ,…a 2j},L2={L 21 , L 22 ,…L 2j};

[0080] …

[0081] T m =(P m ,a m ,L m ), where P m ={P m1 , P m2 ,…P mj}, a m ={a m1 , a m2 ,…a mj},L m ={L m1 , L m2 ,…L mj}.

[0082] The dynamic trajectory sample space can be recorded as UBT. The dynamic trajectory sample space of user operation behavior can be constructed according to the sample size as shown in the following formula (2), where m is the sample size of user operation behavior trajectory.

[0083] UBT={(P1,a1,L1),(P2,a2,L2),…,(P m ,a m ,Lm )} (2)

[0084] In the embodiment of the present invention, a page variable is added to the existing operation behavior trajectory description matrix to identify the interactive jump relationship between different pages, thereby more completely describing the cross-page behavior operations in a task flow.

[0085] In an optional embodiment of the present invention, the sample size of the user operation trajectory is determined based on the user type and the operation behavior information, including: determining the first operation behavior information corresponding to a first user type completing a visit based on the user type and the operation behavior information; the first user type is any user type among the user types; determining the total number of operations based on the first operation behavior information; forming a first feature vector with the first page information, first operation component information, and first area information corresponding to the first number of operations; the first number of operations is any operation number in the total number of operations; determining a first sample of the user operation trajectory based on multiple first feature vectors; and determining the sample size of the user operation trajectory based on multiple first samples.

[0086] In an embodiment of the present invention, the first user type is any of the user types described above; the user type can be determined based on actual circumstances and is not limited here. As an example, the user type can be a first-time user, an experienced user, or the like. The first user type can also be determined based on actual circumstances and is not limited here. As an example, the first user type can be an experienced user type, referred to as an experienced user.

[0087] Determining the first operation behavior information corresponding to a visit completed by a first user type based on the user type and the operation behavior information can be understood as determining users of the same type based on the user type, and determining the first operation behavior information corresponding to a visit completed according to the users of the same type and the corresponding operation behavior information.

[0088] The total number of operations is determined according to the first operation behavior information; wherein, the total number of operations can be recorded as n; the specific value of n can be determined according to actual conditions and is not limited here.

[0089] The first operation number is any operation number in the total operation number; wherein, the first operation number can be recorded as j, 1≤j≤n.

[0090] Combining the first page information, first operation component information, and first region information corresponding to the first operation number into a first feature vector can be understood as combining the page information, operation component information, and region information corresponding to one operation in a complete access into a feature vector.

[0091] Determining the first sample of the user operation trajectory based on the multiple first feature vectors can be understood as combining the feature vectors corresponding to each operation in a complete access into a sample.

[0092] Determining the sample size of the user operation trajectory according to the plurality of first samples can be understood as determining the sample size of the user operation trajectory according to each sample in the first user type.

[0093] For ease of understanding, here we can illustrate that the sample size of the user operation trajectory can be recorded as m; the first page information can be recorded as P mj ; The first operation component information can be recorded as a mj ; The first area information can be recorded as L mj ; The first eigenvector can be recorded as T mj , T mj It can be expressed as (P mj , a mj , L mj ); The first sample can be recorded as T m , T m =(P m ,a m ,L m ), where P m ={P m1 , P m2 ,…P mj}, a m ={a m1 , a m2 ,…a mj},L m ={L m1 , L m2 ,…L mj Where j is the operation order, 1≤j≤n, n is the total number of operations in a task flow; i is the i-th task flow, 1≤i≤m, and m is the sample size of the user operation behavior trajectory.

[0094] In practical applications, T mj It can also be abbreviated as T, (P mj , a mj , L mj ) can also be abbreviated as {P, a, L}.

[0095] In an optional embodiment of the present invention, the characterizing and stratifying the dynamic trajectory sample space to obtain a stratification result includes: obtaining a first feature element based on the dynamic trajectory sample space; the first feature element representing any one of the feature elements of the same type in the plurality of first feature vectors; performing cluster analysis on the plurality of first feature elements based on the type of the user to obtain a plurality of cluster groups; the cluster grouping representing the degree of similarity between the plurality of first feature elements; and using the plurality of cluster groups as the stratification result.

[0096] The first characteristic element represents any one of the characteristic elements of the same type in the plurality of first characteristic vectors; for ease of understanding, the first characteristic vector is any one characteristic vector, which can be denoted as {P i 、a i 、L i}; Among them, P i： P1={P 11 , P 12 ,…P 1j}、P2={P 21 , P 22 ,…P 2j}, ..., P m ={P m1 , P m2 ,…P mj};

[0097] a i： a1={a 11 , a 12 ,…a 1j}、a2={a 21 , a 22 ,…a 2j}, …, a m ={a m1 , a m2 ,…a mj};

[0098] L i： L1={L 11 , L 12 ,…L 1j}、L2={L 21 , L 22 ,…L 2j}, …, L m ={L m1 , L m2 ,…L mj};

[0099] Where j is the operation order, 1≤j≤n, and n is the total number of operations in a task flow.

[0100] The first characteristic element can be denoted as P i 、a i 、L i Any one of .

[0101] Obtaining the first characteristic element according to the dynamic trajectory sample space may be based on the dynamic trajectory sample space UBT={(P1, a1, L1), (P2, a2, L2), …, (P i 、a i 、L i ),…,(P m ,a m ,L m )}; Get P i 、a i 、L i Any one of .

[0102] Cluster analysis of the plurality of first characteristic elements based on the user type to obtain the plurality of cluster groups may be performed by clustering first characteristic elements with the same degree of similarity among the plurality of first characteristic elements based on the user type to obtain the plurality of cluster groups. The cluster groups may be based on the importance and / or abnormality of the operation behaviors; as an example, the cluster groups may be high-frequency operation behaviors, minor operation behaviors, abnormal operation behaviors, etc.

[0103] In an optional embodiment of the present invention, clustering analysis is performed on multiple first feature elements according to the type of the user to obtain multiple cluster groups, including: determining the similarity between any two of the multiple first feature elements according to the type of the user; and clustering according to the similarity to obtain multiple cluster groups.

[0104] In an embodiment of the present invention, determining the similarity between any two of the plurality of first feature elements based on the user type may be determining the cosine similarity between any two of the plurality of first feature elements based on the user type; clustering based on the similarity to obtain a plurality of cluster groups may be clustering any two of the first feature elements whose cosine similarity approaches a preset threshold to obtain a plurality of cluster groups; wherein the preset threshold may be determined based on actual circumstances and is not limited herein. As an example, the preset threshold may be 1.

[0105] To facilitate understanding, here is an example of a practical application scenario. Based on the user type classification results, the behavior trajectory routes of similar users can be characterized and stratified using a hierarchical clustering algorithm. The hierarchical clustering algorithm uses the representative agglomerative method AGNES. The specific steps are as follows:

[0106] (1) Initialization: each sample in different features {P, a, L} is treated as a cluster.

[0107] (2) Calculate the clustering of any two clusters in the same feature, find the two clusters with the closest cosine similarity, merge the two clusters and cluster again. Based on the classification in S1, such as the behavior trajectory of senior users, clustering is performed. The current sample space is UBT = {(P1, a1, L1), (P2, a2, L2) ... (P m ,a m ,L m )}, cluster any two samples with the same feature in the classification features {P, a, L}.

[0108] Any cluster in the same feature is represented as follows:

[0109] P i： P1={P 11 , P 12 ,…P 1j}、P2={P 21 , P 22 ,…P 2j}, ..., P m ={P m1 , P m2 ,…P mj},

[0110] a i： a1={a 11 , a 12 ,…a 1j}、a2={a 21 , a 22 ,…a 2j}, …, a m ={a m1 , a m2 ,…a mj},

[0111] L i： L1={L 11 , L 12 ,…L 1j}、L2={L 21 , L 22 ,…L 2j}, …, L m ={L m1 , L m2 ,…L mj},

[0112] Where j is the operation order, 1≤j≤n, and n is the total number of operations in a task flow.

[0113] P iFor example, each sample is considered as a cluster, and the feature P is traversed i For any two clusters in the dataset, the closer the cosine similarity cosθ of the samples is to 1, the more similar the two clusters are. The two clusters with the highest similarity are clustered and merged to generate a new cluster. The cosine similarity is calculated as shown in the following formula (3). If the total number of operations in the task flow of the two samples is different, the subsequent operation order is padded with zeros:

[0114]

[0115] In formula (3), A i 、B i For any two clusters, 1≤i≤n.

[0116] The clustering calculation formula is shown in the following formula (4):

[0117] cos({P1},{P2},{P3},{P4},…,{P m})=max(cos(P1,P2),cos(P1,P3),…cos(P m-1 ,P m )) (4)

[0118] For example, we need to distinguish between two types of behavior trajectories: high-frequency operations and minor operations of senior users. Take 5 samples P1, P2, P3, P4, and P5 as an example. We traverse any two clusters in feature P5 and compare the cosine similarity between P1 and P2, which is recorded as cosθ. 12 , compare the cosine similarity between P1 and P3, recorded as cosθ 13 , compare the cosine similarity between P1 and P4, recorded as cosθ 14 …compared to cosθ 12 、cosθ 13 …cosθ 35 、cosθ 45 , merge the two clusters whose cosine similarity is closest to 1; if cosθ is calculated 12 If the cosine similarity is closest to 1, P1 and P2 are merged into a new cluster P1P2;

[0119] (3) Repeat the above steps. If the cluster contains multiple samples, traverse the samples in the cluster and calculate the cosine similarity and take the average value. Clustering is performed until the number of clusters reaches the specified threshold, and the algorithm is terminated.

[0120] Take P1, P2, P3, P4, and P5 as examples for explanation.

[0121] Compare the cosine similarity between P1 and P3, and between P2 and P3, and take the average value, which is the cosine similarity cosθ between P1P2 and P3. 123 ,

[0122] Compare the cosine similarity between P1 and P4, and between P2 and P4, and take the average value, which is the cosine similarity cosθ between P1P2 and P4. 124 ,

[0123] Compare the cosine similarity between P1 and P5, and between P2 and P5, and take the average value, which is the cosine similarity cosθ between P1P2 and P5. 125 ,

[0124] Compare with cosθ 123 、cosθ 124 、cosθ 125 、cosθ 23 、cosθ 24 …merge the two clusters whose cosine similarity is closest to 1;

[0125] For example, cosθ is calculated 34 The highest, then merge P3 and P4 into P3P4,

[0126] …and so on, the algorithm is terminated when the cluster convergence threshold reaches 2, and the high-frequency operation and secondary operation trajectories of corresponding senior users are distinguished according to the team's expected behavior trajectories.

[0127] For example, the high-frequency operation behavior trajectory finally converges to: P3P4P5, a1a2a3a4, L1L2L3;

[0128] The secondary operation behavior trajectory finally converges to: P1P2, a5, L4L5.

[0129] In an optional embodiment of the present invention, fitting the user operation trajectory according to the stratification result to obtain the target operation trajectory includes: performing time-series fitting on the user operation trajectory according to the stratification result to obtain a behavior sequence with the highest probability of page jump corresponding to the user operation; determining a target feature vector based on the behavior sequence; the target feature vector includes target page information, target operation component information, and target area information; and performing spatial influence factor fitting on the target feature vector to obtain the target operation trajectory.

[0130] In an embodiment of the present invention, performing time-series fitting on the user operation trajectory according to the stratification results to obtain a behavior sequence with the highest probability of page jumps corresponding to the user operation can be performing time-series fitting on the user operation trajectory according to the stratification results using time-series preset conditions to obtain a behavior sequence with the highest probability of page jumps corresponding to the user operation; the time-series preset conditions can be determined based on actual conditions and are not limited here. As an example, the time-series preset conditions can be a time-series prior condition probability matrix, which can specifically be a time-series prior condition probability matrix for feature P, a time-series prior condition probability matrix for feature a, and a time-series prior condition probability matrix for feature L.

[0131] Determining the target feature vector based on the behavior sequence can be determining target feature elements based on the behavior sequence, integrating the target feature elements to determine the target feature vector. The target feature vector includes target page information, target operation component information, and target area information; the target page information can be recorded as P h ; The target operation component information can be recorded as a h ; The target area information can be recorded as L h ; The target feature vector can be recorded as {P h 、a h 、L h}.

[0132] The target feature vector is fitted with a spatial influence factor to obtain a target operation trajectory. The target feature vector can be fitted with a spatial influence factor to obtain a target operation trajectory. The spatial influence factor can be determined based on actual conditions and is not limited here. As an example, the spatial influence factor can be a preset model parameter and / or vector centroid correction.

[0133] In this embodiment of the present invention, user behavior characteristics are fitted to temporal and spatial information, and the spatial information of the trajectory is used to correct the centroid based on the temporal information. This is combined with an error function to improve the robustness of the model. The trajectory is not limited to historical behavior trajectories or pre-set trajectory samples, effectively reducing data analysis errors.

[0134] In an optional embodiment of the present invention, the spatial influence factor fitting of the target feature vector to obtain the target operation trajectory includes: correcting the target feature vector using preset model parameters to obtain a corrected target feature vector; and performing feature vector centroid correction on the corrected target feature vector to obtain the target operation trajectory.

[0135] In this embodiment, the target feature vector is corrected using preset model parameters to obtain a corrected target feature vector; wherein, the preset model parameters can be determined according to actual conditions and are not limited here. As an example, the preset model parameters can be obtained through multiple sample training regressions; the preset model parameters can be denoted as θ; the target feature vector can be denoted as h θ (T).

[0136] The corrected target feature vector is subjected to feature vector centroid correction to obtain a target operation trajectory; the feature vector centroid can be understood as the center point of the trajectory vector.

[0137] For ease of understanding, here is an example to illustrate that user behavior trajectories can be fitted based on the stratification results. For example, the high-frequency operation behavior trajectories of experienced users can be extracted and the sample features {P, a, L} in each category can be fitted in time series to maximize the probability of the feature sequence appearing.

[0138] Specifically, taking feature P as an example, the temporal prior condition probability matrix of feature P is constructed as shown in the following formula (5), where j is the operation order, 1≤j≤n, and n is the total number of operations in a task flow:

[0139]

[0140] For example, if the system contains four pages A, B, C, and D, the probability of A jumping to B is p(AB), the probability of A jumping to C is p(AC), and so on. Then, we can obtain the probability pj of page jumps corresponding to the operation sequence of a task flow in m samples, traverse all sequences, and obtain the probability p(P h )=p(P1)*p(P2|P1)*…p(P j |P j-1 ) The P behavior sequence with the highest probability.

[0141] Let’s take the high-frequency operation behavior of experienced users as an example to illustrate:

[0142] P3={P 31 , P 32 ,…P 3j}、P4={P 41 , P 42 ,…P 4j}、P5={P 51 , P 52 ,…P 5j};

[0143] Calculate P 31 、P 41 、P 51 If the page with the highest landing probability is page A, then p(P1) = p(A);

[0144] Calculate P 32 、P 42 、P 52 The probability of A jumping to B is p(AB), the probability of A jumping to C is p(AC), ..., if p(AB) has the highest probability, then p(P2|P1) = p(AB);

[0145] …

[0146] By analogy, we can obtain the value that makes p(P h )=p(P1)*p(P2|P1)*…p(P j |P j-1) The P behavior sequence with the largest probability is denoted as P h ={P h1 , P h2 ,…P hj};

[0147] Similarly, we can obtain p(a h )=p(a1)*p(a2|a1)*…p(a j |a j-1 ) The a behavior sequence with the largest probability is recorded as a h ={a h1 , a h2 ,…a hj}

[0148] Get p(L h )=p(L1)*p(L2|L1)*…p(L j |L j-1 ) The L behavior sequence with the largest probability, denoted as L h ={L h1 ,L h2 ,…L hj}. The feature {P h 、a h 、L h} is integrated and modified by the model parameter θ, where θ is obtained through multiple sample training regressions. During the regression process, the parameter θ is continuously adjusted so that the error function approaches 0, which is expressed as shown in the following formulas (6) and (7):

[0149]

[0150]

[0151] The user behavior trajectory is affected by both the operation sequence and the operation space. After considering the operation sequence influencing factor, this step further considers the operation space influencing factor, that is, based on the original sequence, the center point of the fitted trajectory vector is corrected to minimize the distance between the fitted operation behavior trajectory curve and the mean of the vector center point of each sample. By utilizing the centroid g spatial information and combining the operation sequence, the accuracy of the model fitting behavior trajectory curve is improved, the data accuracy and the reliability of the fitting value are improved, and the error function calculation is combined to improve the robustness of the model. θ (T) is shown in the following formula (8):

[0152]

[0153] g is the vector center of {P, a, L}, that is, the centroid of the eigenvector, as shown in the following formula (9), where j is the operation order, 1≤j≤n, and n is the total number of operations in a task flow:

[0154]

[0155] The error function is J (θ) , as shown in the following formula (10), J (θ) →0

[0156]

[0157] Among them H θ (T i ) is the model’s assumed output path, T i is the actual output path.

[0158] The user operation trajectory analysis method provided by the embodiment of the present invention labels different types of users in the system, records the operation behavior sequence, characterizes and stratifies the behavior trajectory routes through a hierarchical clustering algorithm, and performs trajectory fitting on the operation time sequence and operation space based on the stratification results to extract the high-frequency operation behavior trajectories of different types of users.

[0159] For ease of understanding, the present invention illustrates a specific application scenario of the user operation trajectory analysis method, and the steps are as follows:

[0160] Step 1: Label different types of users accessing the system. Type refers to the basic attributes of the user, including region, gender, age, income, and other attributes that represent the user's basic situation.

[0161] For ease of understanding, this example illustrates how to distinguish user regions based on device access IP addresses and assign regional tags to users in different regions. Users are distinguished as novice or experienced users based on the number of times they access the system. The definition of novice and experienced users can be determined based on actual circumstances and is not limited here. As an example, a novice user can be a first-time user; an experienced user can be someone who has used the product for more than six months and accesses the system more than twice a day.

[0162] Step 2: Use web tracking technology to extract user operation behavior information when accessing system pages. A user operation behavior is represented as follows:

[0163] T = {P, a, L};

[0164] P represents the page the user is currently visiting. Furthermore, a system consists of multiple pages. The first page in the system is page1, recorded as 0001, the second page in the system is page2, recorded as 0010...

[0165] a represents specific human-computer interaction operations such as clicking a button, entering text, dragging, etc. Furthermore, these interactive behaviors and the components or related areas of the corresponding pages are numbered, for example:

[0166] The click operation is 1. Furthermore, the first button component in the first page is page1_click1, recorded as 0001, and the second button component in the first page is page1_click2, recorded as 0010...

[0167] The input operation input is 2. Furthermore, the first input component in the first page is page1_input1, the second input component in the first page is page1_input2, ...

[0168] L represents the area corresponding to the user's operation behavior, and further includes (x, y), the screen coordinate value of the corresponding operation area; for example, if the screen is 1920*1080px, it is recorded as 11110000000, 10000111000, and so on.

[0169] Step 3: Based on the user operation behavior model described in the second step, further extract the user behavior trajectory model, in which the start event and end event of a task process are customized according to needs.

[0170] In particular, unlike conventional behavior trajectory routes, this behavior trajectory model needs to describe both the linear trajectory and the operation pages, operation behaviors, and sequence along the trajectory. For ease of understanding, here is an example of extracting the user behavior trajectory model in the user operation trajectory analysis method according to an embodiment of the present invention. Figure 2 Schematic diagram of extracting a user behavior trajectory model in the user operation trajectory analysis method according to an embodiment of the present invention. Figure 2 As shown in Figure 1, T1 and T2 are models for extracting user behavior trajectories, both of which describe linear trajectories and the operation pages, operation behaviors, and sequence along the trajectories.

[0171] Construct the user operation behavior dynamic trajectory sample space, which is expressed as follows:

[0172] UBT={(P1,a1,L1),(P2,a2,L2),…,(P m ,a m ,L m )}, 1≤i≤m, m is the sample size of user operation behavior trajectory.

[0173] The complete behavior trajectory of a single user in a task process is T, T i Indicates that a single user iThe complete behavior trajectory in the task process, where j is the operation order, 1≤j≤n, and n is the total number of operations in a task process. i Refer to the previous formula (1).

[0174] For ease of understanding, here is an example, taking m behavior trajectory samples of senior users as an example:

[0175] T1=(P1,a1,L1), where P1={P 11 , P 12 ,…P 1j}, a1={a 11 , a 12 ,…a 1j},L1={L 11 , L 12 ,…L 1j};

[0176] T2=(P2,a2,L2), where P2={P 21 , P 22 ,…P 2j}, a2={a 21 , a 22 ,…a 2j},L2={L 21 , L 22 ,…L 2j};

[0177] …

[0178] T m =(P m ,a m ,L m ), where P m ={P m1 , P m2 ,…P mj}, a m ={a m1 , a m2 ,…a mj},L m ={L m1 , L m2 ,…L mj}.

[0179] Step 4: Based on the classification results of the labels (basic user attributes) in the first step, the behavioral trajectories of similar users are characterized and layered using a hierarchical clustering algorithm. The hierarchical clustering algorithm uses the representative agglomerative method AGNES. The specific steps are as follows:

[0180] (1) Initialization: each sample in different features {P, a, L} is treated as a cluster.

[0181] (2) Calculate the clustering of any two clusters in the same feature, find the two clusters with the closest cosine similarity, merge these two clusters and cluster again. Based on the classification in the first step, for example, clustering the behavior trajectory of senior users, the current sample space can refer to the previous formula (2) and cluster any two samples in the same feature of the classification features {P, a, L}.

[0182] Any cluster in the same feature is represented as follows:

[0183] P i： P1={P 11 , P 12 ,…P 1j}、P2={P 21 , P 22 ,…P 2j}, ..., P m ={P m1 , P m2 ,…P mj},

[0184] a i： a1={a 11 , a 12 ,…a 1j}、a2={a 21 , a 22 ,…a 2j}, …, a m ={a m1 , a m2 ,…a mj},

[0185] L i： L1={L 11 , L 12 ,…L 1j}、L2={L 21 , L 22 ,…L 2j}, …, L m ={L m1 , L m2 ,…L mj},

[0186] Where j is the operation order, 1≤j≤n, and n is the total number of operations in a task flow.

[0187] P i For example, each sample is considered as a cluster, and the feature P is traversed iFor any two clusters in the dataset, the closer the cosine similarity cosθ of the samples is to 1, the more similar the two clusters are. The two clusters with the highest similarity are clustered and merged to generate a new cluster. The cosine similarity calculation refers to the formula (3) above. If the total number of operations in the task flow of the two samples is different, the subsequent operation order is padded with zeros.

[0188] The clustering calculation formula is shown in the previous formula (4).

[0189] For ease of understanding, here is an example. For example, we need to distinguish between two types of behavior trajectories: high-frequency operations and secondary operations of senior users. Take five samples P1, P2, P3, P4, and P5 as an example. Traverse any two clusters in feature P5 and compare the cosine similarity of P1 and P2, which is recorded as cosθ 12 , compare the cosine similarity between P1 and P3, recorded as cosθ 13 , compare the cosine similarity between P1 and P4, recorded as cosθ 14 …compared to cosθ 12 、cosθ 13 …cosθ 35 、cosθ 45 , merge the two clusters whose cosine similarity is closest to 1; if cosθ is calculated 12 If the cosine similarity is closest to 1, P1 and P2 are merged into a new cluster P1P2.

[0190] Referring to the similar method mentioned above, the clustering of a and L is calculated in the same way, which will not be repeated here.

[0191] (3) Repeat the above steps. If the cluster contains multiple samples, traverse the samples in the cluster and calculate the cosine similarity and take the average value. Clustering is performed until the number of clusters reaches the specified threshold, and the algorithm is terminated.

[0192] For ease of understanding, here is an example, taking P1, P2, P3, P4, and P5 as examples. Figure 3 Schematic diagram of the initial state of feature P in the user operation trajectory analysis method according to an embodiment of the present invention, as shown in FIG. Figure 3 shown.

[0193] Compare the cosine similarity between P1 and P3, and between P2 and P3, and take the average value, which is the cosine similarity cosθ between P1P2 and P3. 123 ,

[0194] Compare the cosine similarity between P1 and P4, and between P2 and P4, and take the average value, which is the cosine similarity cosθ between P1P2 and P4. 124 ,

[0195] Compare the cosine similarity between P1 and P5, and between P2 and P5, and take the average value, which is the cosine similarity cosθ between P1P2 and P5. 125,

[0196] Compare with cosθ 123 、cosθ 124 、cosθ 125 、cosθ 23 、cosθ 24 …merge the two clusters whose cosine similarity is closest to 1;

[0197] For example, cosθ is calculated 34 The highest, then merge P3 and P4 into P3P4,

[0198] …and so on, the algorithm is terminated when the cluster convergence threshold reaches 2, and the high-frequency operation and secondary operation trajectories of corresponding senior users are distinguished according to the team's expected behavior trajectories.

[0199] Referring to the similar method mentioned above, the clustering of a and L is calculated in the same way, which will not be repeated here.

[0200] As an example, the high-frequency operation behavior trajectory finally converges to: P3P4P5, a1a2a3a4, L1L2L3;

[0201] The secondary operation behavior trajectory finally converges to: P1P2, a5, L4L5.

[0202] Step 5: Fit user behavior trajectories based on the stratification results of the fourth step above. For example, extract the high-frequency operation behavior trajectories of senior users and fit the sample features {P, a, L} in each category in time series to maximize the probability of the feature sequence appearing.

[0203] Specifically, taking feature P as an example, the temporal prior condition probability matrix of feature P can be constructed by referring to the above formula (5), where j is the operation order, 1≤j≤n, and n is the total number of operations in a task process:

[0204] For example, if the system contains four pages A, B, C, and D, the probability of A jumping to B is p(AB), the probability of A jumping to C is p(AC), and so on. The probability p of page jumping corresponding to the operation sequence of a task flow in m samples is obtained. j , traverse all sequences and obtain the value that makes p(P h )=p(P1)*p(P2|P1))*…p(P j |P j-1 ) The P behavior sequence with the highest probability.

[0205] Let's take the high-frequency operation behavior of experienced users in S4 as an example to illustrate:

[0206] P3={P 31 , P 32 ,…P3j}、P4={P 41 , P 42 ,…P 4j}、P5={P 51 , P 52 ,…P 5j},

[0207] Calculate P 31 、P 41 、P 51 If the page with the highest landing probability is page A, then p(P1)=p(A)

[0208] Calculate P 32 、P 42 、P 52 The probability of A jumping to B is p(AB), the probability of A jumping to C is p(AC), ..., if p(AB) has the highest probability, then p(P2|P1)=p(AB)

[0209] …

[0210] By analogy, we can obtain the value that makes p(P h )=p(P1)*p(P2|P1))*…p(P j |P j-1 ) The P behavior sequence with the largest probability is denoted as P h ={P h1 , P h2 ,…P hj}

[0211] Similarly, we can obtain p(a h )=p(a1)*p(a2|a1))*…p(a j |a j-1 ) The a behavior sequence with the largest probability is recorded as a h ={a h1 , a h2 ,…a hj}

[0212] Get p(L h )=p(L1)*p(L2|L1))*…p(L j |L j-1 ) The L behavior sequence with the largest probability, denoted as L h ={L h1 , L h2 ,…L hj}.

[0213] Step 6: This step will feature {P h 、a h, L} are integrated and modified by the model parameter θ, where θ is obtained through multiple sample training regressions. During the regression process, the parameter θ is continuously adjusted so that the error function approaches 0, which can be referred to the previous formulas (6) and (7).

[0214] The user behavior trajectory is affected by both the operation sequence and the operation space. The operation sequence influencing factor was considered in the fifth step above. This step further considers the operation space influencing factor, that is, based on the original sequence, the center point of the fitted trajectory vector is corrected to minimize the distance between the fitted operation behavior trajectory curve and the mean of the vector center points of each sample. By utilizing the centroid g spatial information and combining the operation sequence, the accuracy of the model fitting behavior trajectory curve is improved, the data accuracy and the reliability of the fitting value are improved, and the error function calculation is combined to improve the robustness of the model. θ (T) can refer to the above formula (8).

[0215] g is the vector center of {P, a, L}, i.e., the centroid of the eigenvector, which can be shown in the previous formula (9). Where j is the operation order, 1≤j≤n, and n is the total number of operations in a task flow:

[0216] The error function is J (θ) , you can refer to the previous formula (10). (θ) →0,H θ (T) is the model’s assumed output path, T i is the actual output path.

[0217] In practical applications, H θ (T) is the target operation trajectory.

[0218] The embodiments of the present invention have the following beneficial technical effects:

[0219] First: Analyze cross-page jumps and behavioral trajectories within a page, and vectorize cross-page behavioral trajectories. The behavioral trajectory model simultaneously describes the linear trajectory as well as the operation pages, operation behaviors, and sequence along the trajectory, constructing a complete dynamic behavioral trajectory model in a conversation process.

[0220] Second: Accurately fit the high-frequency access paths in user operation behaviors. Use historical user trajectory data to construct a feature time series prior condition probability matrix. Based on the operation time series information, use the operation space information to perform centroid correction, thereby improving the accuracy of the model fitting behavior trajectory curve. At the same time, the model robustness is improved through error function calculation. This invention is not limited to historical behavior trajectories or pre-set trajectory samples, reducing data analysis errors and effectively improving data accuracy and fitting value reliability.

[0221] This embodiment provides a user operation trajectory analysis device. Figure 4 FIG. 1 is a schematic diagram showing the structure of a user operation trajectory analysis device according to an embodiment of the present invention. Figure 4 As shown, the apparatus 200 includes: an obtaining unit 201, a constructing unit 202, a stratification unit 203 and a fitting unit 204, wherein:

[0222] The obtaining unit 201 is used to obtain operation behavior information of a user accessing a system page; the operation behavior information at least includes: page information, operation component information corresponding to the interaction with the page, and area information corresponding to the operation component;

[0223] The construction unit 202 is configured to construct a dynamic trajectory sample space based on the operation behavior information;

[0224] The stratification unit 203 is used to perform feature stratification on the dynamic trajectory sample space to obtain a stratification result;

[0225] The fitting unit 204 is configured to fit the user operation trajectory according to the stratification result to obtain a target operation trajectory.

[0226] In other embodiments, the obtaining unit 201 is further configured to obtain attribute information of the user; and determine the user type according to the attribute information.

[0227] In other embodiments, the constructing unit 202 is further configured to determine a sample size of the user operation trajectory based on the user type and the operation behavior information; and construct a dynamic trajectory sample space according to the sample size.

[0228] In other embodiments, the construction unit 202 is also used to determine the first operation behavior information corresponding to a visit completed by the first user type based on the user type and the operation behavior information; the first user type is any user type among the user types; the total number of operations is determined according to the first operation behavior information; the first page information, first operation component information, and first area information corresponding to the first number of operations are combined into a first feature vector; the first number of operations is any operation number in the total number of operations; a first sample of the user operation trajectory is determined based on multiple first feature vectors; and the sample size of the user operation trajectory is determined based on multiple first samples.

[0229] In other embodiments, the stratification unit 203 is further configured to obtain a first feature element based on the dynamic trajectory sample space; the first feature element represents any one of the feature elements of the same type in the multiple first feature vectors; cluster analysis is performed on the multiple first feature elements according to the type of the user to obtain multiple cluster groups; the cluster groups represent the degree of similarity of the multiple first feature elements; and the multiple cluster groups are used as stratification results.

[0230] In other embodiments, the stratification unit 203 is further configured to determine the similarity between any two of the plurality of first feature elements according to the type of the user; and perform clustering according to the similarity to obtain a plurality of cluster groups.

[0231] In other embodiments, the fitting unit 204 is further used to perform time-series fitting on the user operation trajectory according to the stratification results to obtain a behavior sequence with the highest probability of page jump corresponding to the user operation; determine a target feature vector based on the behavior sequence; the target feature vector includes target page information, target operation component information, and target area information; and perform spatial influence factor fitting on the target feature vector to obtain a target operation trajectory.

[0232] In other embodiments, the fitting unit 204 is further configured to correct the target feature vector using preset model parameters to obtain a corrected target feature vector; and perform feature vector centroid correction on the corrected target feature vector to obtain a target operation trajectory.

[0233] The description of the above device embodiment is similar to the description of the above method embodiment and has similar beneficial effects as the method embodiment. For technical details not disclosed in the device embodiment of the present invention, please refer to the description of the method embodiment of the present invention for understanding.

[0234] It should be noted that, in the embodiment of the present invention, if the above-mentioned user operation trajectory analysis method is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical embodiment of the embodiment of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling a control server (which can be a personal computer, server, or network server, etc.) to execute all or part of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as a U disk, a mobile hard disk, a read-only memory (ROM), a magnetic disk or an optical disk. In this way, the embodiment of the present invention is not limited to any specific combination of hardware and software.

[0235] Correspondingly, an embodiment of the present invention provides a user operation trajectory analysis device, including a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the program, the steps in the control method provided in the above embodiment are implemented.

[0236] Correspondingly, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the control method provided in the above embodiment are implemented.

[0237] It should be noted that the description of the above storage medium and device embodiments is similar to the description of the above method embodiments and has similar beneficial effects as the method embodiments. For technical details not disclosed in the storage medium and device embodiments of the present invention, please refer to the description of the method embodiments of the present invention for understanding.

[0238] It should be noted that Figure 5 FIG. 1 is a schematic diagram of a hardware entity structure of a user operation trajectory analysis device according to an embodiment of the present invention. Figure 5 As shown, the hardware entity of the user operation trajectory analysis device 300 includes: a processor 301 and a memory 303 . Optionally, the user operation trajectory analysis device 300 may further include a communication interface 302 .

[0239] It is understood that the memory 303 can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM); the magnetic surface memory can be a magnetic disk memory or a magnetic tape memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), synchronous static random access memory (SSRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct rambus random access memory (DRRAM).The memory 303 described in the embodiments of the present invention is intended to include, but is not limited to, these and any other suitable types of memories.

[0240] The methods disclosed in the above embodiments of the present invention can be applied to or implemented by processor 301. Processor 301 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by hardware integrated logic circuits in processor 301 or by software instructions. Processor 301 may be a general-purpose processor, a digital signal processor (DSP), or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, etc. Processor 301 can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of the present invention can be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module may be located in a storage medium located in memory 303. Processor 301 reads information from memory 303 and, in conjunction with its hardware, completes the steps of the above method.

[0241] In an exemplary embodiment, the device may be implemented by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers (MCUs), microprocessors, or other electronic components to perform the aforementioned method.

[0242] In the several embodiments provided by the present invention, it should be understood that the disclosed methods and devices can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as: multiple units or components can be combined, or can be integrated into another observation quantity, or some features can be ignored or not executed. In addition, the communication connection between the various components shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0243] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of this embodiment.

[0244] Those skilled in the art will understand that all or part of the steps of implementing the above-mentioned method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above-mentioned method embodiment; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROMs), magnetic disks, optical disks, and other media that can store program codes.

[0245] Alternatively, if the above-mentioned integrated unit of the embodiment of the present invention is implemented in the form of a software functional unit and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical embodiment of the embodiment of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for enabling a device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program code, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0246] The user operation trajectory analysis method, device, equipment and storage medium described in the examples of the present invention are only taken as examples of the embodiments of the present invention, but are not limited to this. As long as the user operation trajectory analysis method, device, equipment and storage medium are involved, they are within the scope of protection of the present invention.

[0247] It should be understood that "one embodiment" or "an embodiment" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present invention. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention. The serial numbers of the above-mentioned embodiments of the present invention are for description only and do not represent the advantages and disadvantages of the embodiments.

[0248] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0249] The methods disclosed in the several method embodiments provided by the present invention can be arbitrarily combined without conflict to obtain new method embodiments.

[0250] The features disclosed in several product embodiments provided by the present invention can be arbitrarily combined without conflict to obtain new product embodiments.

[0251] The features disclosed in several method or device embodiments provided by the present invention can be arbitrarily combined without conflict to obtain new method embodiments or device embodiments.

[0252] The above description is merely an embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for analyzing user operation trajectory, characterized in that: The method comprises: Obtaining operation behavior information of a user accessing a system page; the operation behavior information includes at least: page information, operation component information corresponding to the interaction with the page, and area information corresponding to the operation component; Constructing a dynamic trajectory sample space based on the operation behavior information; Characterizing and stratifying the dynamic trajectory sample space to obtain a stratification result; Fitting the user operation trajectory according to the stratification result to obtain a target operation trajectory; The method further comprises: Obtaining attribute information of the user; determining the user type according to the attribute information; The constructing of a dynamic trajectory sample space based on the operation behavior information includes: Determining a sample size of the user operation trajectory based on the user type and the operation behavior information; A dynamic trajectory sample space is constructed according to the sample size.

2. The method according to claim 1, characterized in that The determining of the sample size of the user operation trajectory based on the user type and the operation behavior information includes: Determining first operation behavior information corresponding to a visit completed by a first user type based on the user type and the operation behavior information; the first user type is any user type among the user types; determining a total number of operations according to the first operation behavior information; The first page information, the first operation component information, and the first region information corresponding to the first operation number are combined into a first feature vector; the first operation number is any operation number in the total operation number; determining a first sample of the user operation trajectory based on a plurality of the first feature vectors; The sample size of the user operation trajectory is determined according to the plurality of first samples.

3. The method according to claim 2, characterized in that The characterizing and stratifying the dynamic trajectory sample space to obtain a stratification result includes: Obtaining a first characteristic element according to the dynamic trajectory sample space; the first characteristic element represents any one of the characteristic elements of the same type in the plurality of first characteristic vectors; Performing cluster analysis on the plurality of first characteristic elements according to the type of the user to obtain a plurality of cluster groups; the cluster groups represent the similarity of the plurality of first characteristic elements; The plurality of clusters are grouped as a hierarchical result.

4. The method according to claim 3, characterized in that The performing cluster analysis on the plurality of first characteristic elements according to the type of the user to obtain a plurality of cluster groups includes: determining, according to the type of the user, a similarity between any two of the plurality of first characteristic elements; Clustering is performed according to the similarity to obtain multiple cluster groups.

5. The method according to claim 1, characterized in that The fitting of the user operation trajectory according to the stratification result to obtain a target operation trajectory includes: Performing time series fitting on the user operation trajectory according to the stratification results to obtain a behavior sequence with the highest probability of page jump corresponding to the user operation; Determine a target feature vector based on the behavior sequence; the target feature vector includes target page information, target operation component information, and target area information; The spatial influence factor is fitted to the target feature vector to obtain the target operation trajectory.

6. The method according to claim 5, characterized in that The performing spatial influence factor fitting on the target feature vector to obtain a target operation trajectory includes: Correcting the target feature vector using preset model parameters to obtain a corrected target feature vector; Perform feature vector centroid correction on the corrected target feature vector to obtain a target operation trajectory.

7. A user operation trajectory analysis device, characterized in that: The device comprises: an obtaining unit, a constructing unit, a stratification unit and a fitting unit, wherein: The obtaining unit is used to obtain operation behavior information of the user accessing the system page; the operation behavior information at least includes: page information, operation component information corresponding to the interaction with the page, and area information corresponding to the operation component; The construction unit is configured to construct a dynamic trajectory sample space based on the operation behavior information; The stratification unit is used to perform characteristic stratification on the dynamic trajectory sample space to obtain a stratification result; The fitting unit is configured to fit the user operation trajectory according to the stratification result to obtain a target operation trajectory; The obtaining unit is further configured to obtain attribute information of the user; determining the user type according to the attribute information; The construction unit is further configured to determine a sample size of the user operation trajectory based on the user type and the operation behavior information; A dynamic trajectory sample space is constructed according to the sample size.

8. A user operation trajectory analysis device, comprising a memory and a processor, wherein the memory stores a program that can be run on the processor, characterized in that: When the processor executes the program, the steps in the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images