Behavior path generation method and device, equipment and storage medium

Abstract

The application relates to the technical field of artificial intelligence, and discloses a behavior path generation method, device, equipment and storage medium. The method comprises the following steps: acquiring user behavior track data; constructing a user portrait based on the user behavior track data; clustering the user portrait by using a clustering algorithm to obtain a plurality of portrait clusters; calculating the similarity between the user portraits in each portrait cluster by using a collaborative filtering algorithm, and constructing a strategy graph based on the user portraits and the similarity between the user portraits; and mining the strategy graph by using a recommendation model to obtain an optimal behavior path. The application improves the efficiency and quality of obtaining the optimal behavior path.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, and particularly relates to a behavior path generation method and device, equipment and a storage medium. BACKGROUND

[0002] To cope with the increasingly complex user needs of users, the corresponding functions of each system or platform are also increasingly complex, the corresponding function modules are also increasingly large, the operation process of the entire function module is also increasingly complex, and the difficulty for operating personnel or users to master the system is also increasingly large. New operating personnel or new users have a high learning cost. In the prior art, the system operation guide is often set by manual operation or the reproducible process in the process steps executed by existing operating personnel is obtained for popularization and application, but the processing efficiency is not high. Therefore, how to improve the efficiency of operation guide generation has become a problem to be solved. SUMMARY

[0003] The present application provides a behavior path generation method, device, equipment and storage medium to solve the problem of low operation guide generation efficiency in each platform or system.

[0004] To solve the above problems, the present application provides a behavior path generation method, comprising:

[0005] Obtaining user behavior trajectory data;

[0006] Constructing a user portrait based on the user behavior trajectory data;

[0007] Clustering the user portrait by using a clustering algorithm to obtain a plurality of portrait clusters;

[0008] Calculating the similarity between the user portraits in each portrait cluster by using a collaborative filtering algorithm, and constructing a strategy graph based on the user portraits and the similarity between the user portraits;

[0009] Mining the strategy graph by using a recommendation model to obtain an optimal behavior path.

[0010] Further, after obtaining the optimal behavior path, the method further comprises:

[0011] Constructing a corresponding frequent pattern tree by using an Fp-growth algorithm based on each portrait cluster;

[0012] Judging the fitting degree of the optimal behavior path and the frequent pattern tree;

[0013] If the fitting degree is greater than or equal to a preset value, outputting the optimal behavior path to the front end and prompting;

[0014] If the fit degree is less than the preset value, the optimal behavior path is inserted into the frequent pattern tree, and the step of mining the strategy graph by a recommendation model to obtain the optimal behavior path is performed again.

[0015] Further, the obtaining of the user behavior trajectory data comprises:

[0016] The user behavior trajectory data is obtained by front-end data burying, and the user behavior trajectory data comprises operation data and stay data.

[0017] Further, the constructing of the user portrait based on the user behavior trajectory data comprises:

[0018] According to the operation data and the stay data, the user is tagged by using a preset label;

[0019] The user portrait is constructed based on the user label.

[0020] Further, after the clustering of the user portraits by using a clustering algorithm to obtain a plurality of portrait clusters, the method further comprises:

[0021] According to the stay data of the user portraits in each portrait cluster, a trajectory mulberry graph corresponding to the portrait cluster is made;

[0022] Based on the trajectory mulberry graph, the weight of each label in the user portrait is determined;

[0023] The calculation of the similarity between the user portraits in each portrait cluster by using a collaborative filtering algorithm comprises:

[0024] The user portraits are converted to obtain a corresponding score matrix;

[0025] The score matrix is edited based on the weight of each label;

[0026] The similarity between each edited score matrix is calculated by using a collaborative filtering algorithm to obtain the similarity between the user portraits.

[0027] Further, the mining of the strategy graph by a recommendation model to obtain the optimal behavior path comprises:

[0028] The recommendation model mines the relationship between nodes in the strategy graph based on the similarity between the nodes to obtain the optimal behavior path, wherein the recommendation model is trained based on a latent semantic model.

[0029] Further, the mining of the strategy graph by a recommendation model comprises:

[0030] The recommendation model mines the strategy graph according to the following formula:

[0031] wherein r ui represents the predicted similarity of whether the user u performs the behavior i, represents the time cost of scanning the operation behavior vector of the user u, q i represents the operation behavior vector of the behavior i, p u,k represents the relationship between the preferred behavior of the user u and the hidden class k, q i,k represents the relationship between the behavior i and the hidden class k, k = 1, 2, …, n.

[0032] To solve the above problems, the application further provides a behavior path generation device, the device comprising:

[0033] An acquisition module configured to acquire user behavior trajectory data;

[0034] An image construction module configured to construct a user image based on the user behavior trajectory data;

[0035] A clustering module configured to cluster the user images by using a clustering algorithm to obtain a plurality of image clusters;

[0036] A graph construction module configured to calculate the similarity between the user images in each of the image clusters by using a collaborative filtering algorithm, and construct a strategy graph based on the user images and the similarity between the user images;

[0037] A recommendation module configured to mine the strategy graph by using a recommendation model to obtain an optimal behavior path.

[0038] To solve the above problems, the application further provides a computer device, comprising:

[0039] at least one processor; and

[0040] a memory in communication connection with the at least one processor; wherein

[0041] The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the behavior path generation method as described above.

[0042] To solve the above problems, the application further provides a non-volatile computer readable storage medium, the computer readable storage medium stores computer readable instructions, and the computer readable instructions are executed by a processor to implement the behavior path generation method as described above.

[0043] According to the behavior path generation method, device, equipment and storage medium provided by the embodiment of the application, compared with the prior art, at least the following beneficial effects are achieved.

[0044] By obtaining user behavior trajectory data, constructing a user portrait based on the user behavior trajectory data, clustering the user behavior portrait by using a clustering algorithm, obtaining a plurality of portrait clusters, classifying the types of behavior trajectories, facilitating subsequent derivation of operation instructions corresponding to each cluster, calculating the similarity between the user portraits in each portrait cluster by using a collaborative filtering algorithm, and constructing a strategy map based on the user portraits and the similarity between the user portraits, subsequent data mining is facilitated, the strategy map is mined by using a recommendation model, and the optimal behavior path, i.e., the operation instruction, corresponding to each portrait cluster is obtained, the efficiency and quality of obtaining the operation instruction are improved, and the obtained operation instruction is more suitable for actual use. BRIEF DESCRIPTION OF DRAWINGS

[0045] In order to more clearly illustrate the schemes in the application, the drawings needed in the description of the embodiments of the application will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0046] Figure 1 The flowchart of the behavior path generation method provided by an embodiment of the application is shown in the figure.

[0047] Figure 2 The flowchart of the behavior path generation method provided by another embodiment of the application is shown in the figure.

[0048] Figure 3 The module diagram of the behavior path generation device provided by an embodiment of the application is shown in the figure.

[0049] Figure 4 The structure diagram of the computer equipment of an embodiment of the application is shown in the figure. DETAILED DESCRIPTION

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the application belongs; the terms used in the specification of the application are only for the purpose of describing specific embodiments and are not intended to limit the application; the specification and claims of the application and the above description of the drawings, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. The specification and claims of the application or the above description of the drawings, the terms "first", "second", etc. are used to distinguish different objects, not to describe a specific order.

[0051] Reference to“an embodiment” means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment. The appearances of the phrase“in one embodiment” in various places in the specification are not necessarily referring to the same embodiment nor are separate or alternative embodiments mutually exclusive of other embodiments. As will be apparent to those of ordinary skill in the art, embodiments described herein can be combined with other embodiments.

[0052] The application provides a behavior path generation method. Referring to Figure 1 , a flowchart of the behavior path generation method provided by an embodiment of the application is shown. Figure 1 , a flowchart of the behavior path generation method provided by an embodiment of the application is shown.

[0053] In the embodiment, the behavior path generation method comprises:

[0054] S1, obtaining user behavior trajectory data;

[0055] Specifically, to obtain the behavior trajectory of a user on a system or platform, the behavior trajectory is, for example, what is clicked on a certain interface of the system, how long the user stays, etc.

[0056] Further, the obtaining of the user behavior trajectory data comprises:

[0057] The user behavior trajectory data is obtained by front-end data burying, and the user behavior trajectory data comprises operation data and stay data.

[0058] Specifically, the behavior trajectory data of a user on a corresponding platform or system is collected by full data burying and custom data burying. Data burying is a common data collection method for website analysis, and data burying is a good private deployment data collection method. The user behavior trajectory data comprises operation data and stay data. The operation data is, for example, the operation condition of the user on the platform or system, such as clicking, and the corresponding click time. The stay data is the stay time of the user after jumping to a new page after the operation condition such as clicking. The operation data and the stay data are also simply counted, such as the maximum number of clicks on something, the longest stay time on a certain page, etc.

[0059] For example, after a user logs in to a system or platform, the user behavior trajectory data is obtained by performing a series of operations. XX user clicks on xx in the morning and stays for xx duration, clicks on xx at noon and stays for xx duration. The simply counted data is, for example, the longest stay duration on xx, the shortest stay duration on xx, the maximum number of clicks on xx, the minimum number of clicks on xx, the maximum number of clicks on xx content at a certain time point, the minimum number of clicks on xx content at a certain time point, etc.

[0060] More specifically, as the system, new inquiry sheet; into the system, open the insurance tracking page, query inquiry sheet, click copy out sheet; into the system, enter information input box; click IM, answer the question robot, enter the input box template, etc.

[0061] By burying data on the system or platform, the behavior trajectory data of the user can be quickly and accurately obtained, providing data support for subsequent processing steps.

[0062] S2, based on the user behavior trajectory data, constructing a user portrait;

[0063] Specifically, according to the operation data and stay data in the above-mentioned user behavior trajectory data, a user portrait is constructed.

[0064] Further, the user portrait is constructed based on the user behavior trajectory data, comprising:

[0065] According to the operation data and stay data, the user is tagged with a preset label;

[0066] Based on the user label, a user portrait is constructed.

[0067] Specifically, according to the corresponding preset label obtained based on the operation data, such as operation data: entering the system, new inquiry sheet, corresponding preset label is new inquiry sheet, and according to the stay data corresponding to the operation data, that is, the stay time in the new inquiry sheet interface, as a score, based on the corresponding preset label and score for tagging; according to all the user labels of the user, a user portrait is constructed;

[0068] Further, the user operation data and stay data obtained every day can be used for tagging and constructing a user portrait, so that the user portrait obtained is more targeted, and the processing scene or application scene is more concentrated.

[0069] By tagging the user according to the operation data and stay data, and constructing a user portrait based on the user label, the data is preprocessed, and the subsequent processing steps are improved.

[0070] S3, using a clustering algorithm to cluster the user portrait, obtaining a plurality of portrait clusters;

[0071] Specifically, the user portrait is clustered by a clustering algorithm, realizing the classification of users, that is, according to the label in the user portrait, each user portrait is divided into different application scenes, so as to obtain a plurality of portrait clusters.

[0072] The clustering algorithm can be hierarchical, such as nearest neighbor method, farthest neighbor method, intra-group clustering method, inter-group clustering method, Ward clustering method, positive binary method, and coarse clustering algorithm; or partitioning, such as graph theory algorithm, K-means algorithm, and fuzzy C-means; or based on density and grid, such as GDILC algorithm, SGC algorithm, GCHL algorithm, TFCTMO algorithm, and ST-DBSCAN; and the selection is free according to needs, which is not limited herein.

[0073] The multiple portrait clusters and the corresponding user portraits in the portrait clusters are obtained, and the corresponding classification storage can be performed by using a Hadoop system.

[0074] S4, similarity between the user portraits in each portrait cluster is calculated by a collaborative filtering algorithm, and a strategy map is constructed based on the user portraits and the similarity between the user portraits;

[0075] Specifically, the similarity between the user portraits in each portrait cluster is calculated by the collaborative filtering algorithm, and the calculation is further performed based on the labels and the corresponding scores of the user portraits; after all the user portraits in a portrait cluster are calculated, the user portraits are connected as nodes and the similarity between the user portraits is connected as lines, so that the strategy map is constructed. The cosine similarity or Pearson correlation coefficient between the user portraits is calculated by the collaborative filtering algorithm, so that the corresponding similarity is obtained.

[0076] Further, after the user portraits are clustered by the clustering algorithm to obtain multiple portrait clusters, the method further includes:

[0077] According to the stay data of the user portraits in each portrait cluster, a trajectory mulberry graph corresponding to the portrait cluster is made;

[0078] Based on the trajectory mulberry graph, the weight of each label in the user portrait is determined;

[0079] The calculation of the similarity between the user portraits in each portrait cluster by the collaborative filtering algorithm includes:

[0080] The user portraits are converted to obtain a corresponding score matrix;

[0081] The score matrix is edited based on the weight of each label;

[0082] The similarity between each edited score matrix is calculated by the collaborative filtering algorithm to obtain the similarity between the user portraits.

[0083] Specifically, according to the stay data of the user portrait in each of the portrait clusters, a trajectory mulberry graph corresponding to the portrait cluster is made, the trajectory mulberry graph is arranged according to the time corresponding to the stay interface in the stay data, and the longer the time corresponding to the stay interface, the greater the proportion it occupies; according to the proportion occupied by each stay interface, the weight of each label in the user portrait is determined.

[0084] According to the label in the user portrait, it is converted into a score matrix, such as a score matrix containing 3 column data elements, in a specific scheme this value can reach dozens or dozens, a user's label is a new inquiry order, and the corresponding score is 1; the corresponding score of the input information is 5. The new inquiry order corresponds to column A in the score matrix, and the input information corresponds to column C in the score matrix, that is, the user score matrix is (1, 0, 5).

[0085] Since the weight of each label is determined above, the weight of each label is directly multiplied and edited according to the score matrix, such as the score matrix with 3 column data elements mentioned above, the weight of the first column data element is 0.5, the weight of the second column data element is 0.3, and the weight of the third column data element is 0.2; therefore, the edited score matrix is (0.5, 0, 1).

[0086] Through a collaborative filtering algorithm, the cosine similarity between each edited score matrix is calculated to obtain the similarity between each of the user portraits, such as another user's edited score matrix is (5, 0.9, 0), and the cosine similarity is calculated as follows:

[0087] By introducing the weight according to the stay data in the user portrait before calculating the similarity, the weight is used to re-edit the user's score matrix, and the similarity is calculated according to the edited score matrix, so as to improve the accuracy of the similarity and make it more close to the actual use.

[0088] S5, through the recommendation model, the strategy graph is mined to obtain an optimal behavior path.

[0089] Specifically, the recommendation model is used to mine the association relationship of each node in the strategy graph to obtain an optimal behavior path, that is, an operation guide, to help users quickly use the system or platform.

[0090] Further, as shown in Figure 2 After the optimal behavior path is obtained, it further includes:

[0091] S6, based on each of the portrait clusters, an Fp-growth algorithm is used to construct a corresponding frequent pattern tree;

[0092] S7, judge the fitting degree of the optimal behavior path and the frequent pattern tree;

[0093] S8, if the fitting degree is greater than or equal to a preset value, output the optimal behavior path to the front end and prompt;

[0094] S9, if the fitting degree is less than the preset value, insert the optimal behavior path into the frequent pattern tree, and perform the step of mining the strategy graph to obtain the optimal behavior path by the recommendation model again.

[0095] Specifically, according to the user portrait data in each portrait cluster, a corresponding frequent pattern tree is constructed by using Fp-growth algorithm;

[0096] The Fp-growth algorithm adopts a divide-and-conquer strategy. After the first scan, the frequent item set in the user portrait data in each portrait cluster is compressed into a frequent pattern tree, while the association information is still retained. Then the tree is divided into some conditional libraries, each of which is related to a frequent item set with a length of 1. Then these conditional libraries are mined. When the amount of original data is very large, the method of division can be combined to make the tree can be put into the main memory.

[0097] The construction process of the tree is as follows: scan the data set to get the count of all frequent 1-item sets. Then delete the items with support less than the threshold, put the 1-item frequent set into the item header table, and arrange them in descending order of support. Scan the data set, delete the non-frequent 1-item set from the original data set, and arrange the transactions in descending order of support.

[0098] Read the sorted transaction set and insert it into the frequent pattern tree in the order of sorting. The nodes in the front of the sorting are parent nodes, and the nodes in the back are child nodes. If there is a common parent node, the corresponding common parent node count is incremented by 1. After insertion, if a new node appears, the node header table corresponding to the node will be linked to the new node through the node chain table. Until all data is inserted into the frequent pattern tree, the construction of the frequent pattern tree is completed.

[0099] Judge the fitting degree of the optimal behavior path and the frequent pattern tree, that is, judge whether the main steps of the optimal behavior path are searched in the frequent pattern tree to ensure that the optimal behavior path obtained is operable and reproducible. The specific fitting degree calculation is to calculate the proportion of the continuous steps in the optimal behavior path that match the frequent pattern tree successfully to the total steps in the optimal behavior path;

[0100] If the fitting degree is greater than or equal to a preset value, output the optimal behavior path to the front-end system or platform, and prompt through the corresponding foreground on the system or platform page. Specifically, it can be prompted through flashing or arrow pointing;

[0101] If the fitting degree is less than the preset value, the optimal behavior path is inserted into the frequent pattern tree, and the step of mining the strategy graph by the recommendation model to obtain the optimal behavior path is performed again.

[0102] The frequent pattern tree constructed by the Fp-growth algorithm is used to judge the fitting degree of the optimal behavior path, so as to check the optimal behavior path, ensure the operability and reproducibility of the obtained optimal behavior path, and ensure the usability of the optimal behavior path.

[0103] Further, the mining of the strategy graph by the recommendation model to obtain the optimal behavior path comprises:

[0104] The recommendation model mines the relationship between nodes in the strategy graph based on the similarity between the nodes to obtain the optimal behavior path, wherein the recommendation model is trained based on a latent semantic model.

[0105] The user portrait and the latent class in the strategy graph and the relationship between the operation behavior and the latent class are mined by the recommendation model, so as to determine the optimal behavior path, and each portrait cluster will obtain a corresponding optimal behavior path, i.e., each portrait cluster has a corresponding optimal behavior path to achieve a certain operation purpose.

[0106] The latent semantic model first appeared in the field of text mining, and was used to find the implicit semantics of text. The core idea is to associate the best behavior path strategy of the user through implicit features, to establish the relationship between the user and the latent class and the relationship between the behavior and the latent class through matrix decomposition, and finally to obtain the optimal relationship of the user to the behavior. The behavior of the user in the recommendation system is divided into explicit feedback and implicit feedback. The latent semantic model solves the rating prediction problem on the explicit feedback data and achieves good accuracy. The implicit feedback data refers to a data set that only has positive samples, i.e., which behavior path strategy is the best and which behavior path strategy is the worst.

[0107] The recommendation model trained based on the latent semantic model is used to mine the strategy graph, so as to obtain an optimal behavior path that is close to the actual situation and improves the processing efficiency and the quality of the optimal behavior path.

[0108] Further, the mining of the strategy graph by the recommendation model comprises:

[0109] The recommendation model mines the strategy graph according to the following formula:

[0110] wherein, r ui represents the predicted similarity of whether the user u performs the behavior i, represents the time cost of the user u to derive the operation behavior vector, q i represents the operation behavior vector of the behavior i, p u,k represents the relationship between the preferred behavior of the user u and the hidden class k, q i,k represents the relationship between the behavior i and the hidden class k, k = 1, 2, …, n.

[0111] The recommendation model improves the processing efficiency by mining the strategy graph using the above formula to obtain the optimal behavior path.

[0112] It should be emphasized that, in order to further ensure the privacy and security of the data, all data of the user portrait data and the behavior trajectory data can also be stored in a node of a blockchain.

[0113] The blockchain referred to in the present application is a new application mode of distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm and other computer technologies. Blockchain, in essence, is a decentralized database, which is a series of data blocks associated using cryptographic methods, each data block contains information of a batch of network transactions, used to verify the validity (anti-fake) of the information and generate the next block. The blockchain can include a blockchain underlying platform, a platform product service layer, and an application service layer, etc.

[0114] The behavior path generation method of the embodiment obtains user behavior trajectory data, constructs a user portrait based on the user behavior trajectory data, clusters the user behavior portrait using a clustering algorithm, obtains a plurality of portrait clusters, classifies the types of behavior trajectories, facilitates subsequent derivation of operation instructions corresponding to each cluster, calculates the similarity between the user portraits in each portrait cluster through a collaborative filtering algorithm, and constructs a strategy graph based on the user portraits and the similarity between the user portraits, facilitating subsequent data mining. The strategy graph is mined by a recommendation model to obtain the optimal behavior path corresponding to each portrait cluster, i.e., the operation instruction, which improves the efficiency and quality of obtaining the operation instruction, and the obtained operation instruction is more suitable for actual use.

[0115] The embodiment also provides a behavior path generation device, as shown in Figure 3 Figure 2 is a functional module diagram of the behavior path generation device of the present application.

[0116] The behavior path generation apparatus 100 described in the present application can be installed in an electronic device. According to the functions implemented, the behavior path generation apparatus 100 can include an acquisition module 101, a portrait construction module 102, a clustering module 103, a graph construction module 104, and a recommendation module 105. The modules described in the present application can also be referred to as units, which refer to a series of computer program segments that can be executed by an electronic device processor and can complete a fixed function, which are stored in the memory of the electronic device.

[0117] In the present embodiment, the functions of each module / unit are as follows:

[0118] The acquisition module 101 is configured to acquire user behavior trajectory data.

[0119] Further, the acquisition module 101 includes a trajectory data acquisition submodule.

[0120] The trajectory data acquisition submodule is configured to obtain user behavior trajectory data by embedding point data in the front end, wherein the user behavior trajectory data includes operation data and stay data.

[0121] By embedding point data on the system or platform through the trajectory data acquisition submodule, user behavior trajectory data can be quickly and accurately obtained, providing data support for subsequent processing steps.

[0122] The portrait construction module 102 is configured to construct a user portrait based on the user behavior trajectory data.

[0123] Further, the portrait construction module 102 includes a tagging submodule and a construction submodule.

[0124] The tagging submodule is configured to tag users with a preset label based on the operation data and stay data.

[0125] The construction submodule is configured to construct a user portrait based on the user tag.

[0126] Through the cooperation of the tagging submodule and the construction submodule, the user is tagged based on the operation data and stay data, and a user portrait is constructed based on the user tag, which realizes the preprocessing of data and improves the subsequent processing steps.

[0127] The clustering module 103 is configured to cluster the user portraits using a clustering algorithm to obtain a plurality of portrait clusters.

[0128] The graph construction module 104 is configured to calculate the similarity between the user portraits in each portrait cluster using a collaborative filtering algorithm, and construct a strategy graph based on the user portraits and the similarity between the user portraits.

[0129] Further, the behavior path generation apparatus 100 further comprises a mapping module and a weight determination module;

[0130] The mapping module is configured to generate a trajectory mulberry graph corresponding to each portrait cluster according to the stay data of the user portraits in the portrait cluster.

[0131] The weight determination module is configured to determine the weight of each label in the user portrait based on the trajectory mulberry graph.

[0132] The graph construction module 104 comprises a conversion submodule, an editing submodule and a similarity calculation submodule.

[0133] The conversion submodule is configured to convert the user portrait to obtain a corresponding score matrix.

[0134] The editing submodule is configured to edit the score matrix based on the weight of each label.

[0135] The similarity calculation submodule is configured to calculate the similarity between each edited score matrix by a collaborative filtering algorithm to obtain the similarity between each user portrait.

[0136] Through the cooperation of the mapping module, the weight determination module, the conversion submodule, the editing submodule and the similarity calculation submodule, the weight is introduced according to the stay data in the user portrait before calculating the similarity, so that the score matrix of the user is re-edited by the weight, the similarity is calculated according to the edited score matrix, thereby improving the accuracy of the similarity and being closer to the actual use.

[0137] The recommendation module 105 is configured to mine the strategy graph by a recommendation model to obtain an optimal behavior path.

[0138] Further, the behavior path generation apparatus 100 further comprises a tree construction module, a fitting degree judgment module and a corresponding processing module.

[0139] The tree construction module is configured to construct a corresponding frequent pattern tree by using a Fp-growth algorithm based on each portrait cluster.

[0140] The fitting degree judgment module is configured to judge the fitting degree of the optimal behavior path and the frequent pattern tree.

[0141] The corresponding processing module is configured to output the optimal behavior path to the front end and prompt if the fitting degree is greater than or equal to a preset value, and if the fitting degree is less than the preset value, insert the optimal behavior path into the frequent pattern tree and execute the step of mining the strategy graph by the recommendation model to obtain the optimal behavior path again.

[0142] Through cooperation of the tree construction module, the fitting degree judgment module and the corresponding processing module, the frequent pattern tree obtained by using the Fp-growth algorithm is used to judge the fitting degree of the optimal behavior path, the optimal behavior path is checked, and it is ensured that the obtained optimal behavior path is operable and reproducible, and the usability of the optimal behavior path is ensured.

[0143] Further, the recommendation module 105 includes a first mining submodule;

[0144] The first mining submodule is configured to mine the relationship between nodes in the strategy graph based on the similarity between the nodes by using the recommendation model, and obtain the optimal behavior path, wherein the recommendation model is obtained based on a latent semantic model.

[0145] By using the recommendation model obtained based on the latent semantic model, the strategy graph is mined by the first mining submodule, the optimal behavior path close to the actual situation is obtained, and the processing efficiency and the quality of the optimal behavior path are improved.

[0146] Further, the recommendation module 105 includes a second mining submodule;

[0147] The second mining submodule is configured to mine the strategy graph according to the following formula by using the recommendation model:

[0148] wherein, r ui represents the predicted similarity of whether the user u performs the behavior i, represents the time cost of scanning the user u to obtain the operation behavior vector, q i represents the operation behavior vector of the behavior i, p u,k represents the relationship between the preferred behavior of the user u and the latent class k, q i,k represents the relationship between the behavior i and the latent class k, k=1, 2, …, n.

[0149] By using the above formula, the strategy graph is mined by the recommendation model in the second mining submodule, the optimal behavior path is obtained, and the processing efficiency is improved

[0150] By adopting the above device, the behavior path generation device 100, through the cooperation of the acquisition module 101, the portrait construction module 102, the clustering module 103, the graph construction module 104 and the recommendation module 105, acquires user behavior trajectory data, constructs a user portrait based on the user behavior trajectory data, clusters the user behavior portrait by using a clustering algorithm, obtains a plurality of portrait clusters, classifies the types of behavior trajectories, facilitates subsequent derivation of operation instructions corresponding to each cluster, calculates the similarity between the user portraits in each portrait cluster by using a collaborative filtering algorithm, and constructs a strategy graph based on the user portraits and the similarity between the user portraits, facilitates subsequent data mining, mines the strategy graph by using a recommendation model, obtains the optimal behavior path corresponding to each portrait cluster, that is, the operation instruction, improves the efficiency and quality of obtaining the operation instruction, and the obtained operation instruction is more suitable for actual use.

[0151] The embodiment of the present application also provides a computer device. For details, please refer to Figure 4 , Figure 4 The basic structure block diagram of the computer device of the embodiment is shown in the figure.

[0152] The computer device 4 includes a memory 41, a processor 42 and a network interface 43 which are connected to each other through a system bus. It should be pointed out that only the computer device 4 with components 41-43 is shown in the figure, but it should be understood that it is not required to implement all the shown components, and more or fewer components can be alternatively implemented. Among them, those skilled in the art can understand that the computer device here is a device capable of automatically performing numerical calculation and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0153] The computer device can be a desktop computer, a notebook computer, a palm computer and a cloud server, etc. The computer device can interact with the user through a keyboard, a mouse, a remote controller, a touchpad or a voice control device, etc.

[0154] The memory 41 includes at least one type of readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., SD or DX memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or a memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the computer device 4. Of course, the memory 41 can also include both the internal storage unit and the external storage device of the computer device 4. In this embodiment, the memory 41 is generally used to store an operating system and various application software installed on the computer device 4, such as computer readable instructions of the behavior path generation method, etc. In addition, the memory 41 can also be used to temporarily store various data that have been output or will be output.

[0155] The processor 42 can be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor 42 is generally used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to run computer readable instructions or process data stored in the memory 41, such as computer readable instructions of the behavior path generation method.

[0156] The network interface 43 can include a wireless network interface or a wired network interface, and is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0157] The embodiment is realized by a processor executing computer readable instructions stored in a memory, and the steps of the behavior path generation method of the above embodiment are realized by obtaining user behavior trajectory data, constructing a user portrait based on the user behavior trajectory data, clustering the user behavior portrait using a clustering algorithm to obtain a plurality of portrait clusters, classifying the types of behavior trajectories, facilitating subsequent derivation of operation instructions corresponding to each cluster, calculating the similarity between the user portraits in each portrait cluster using a collaborative filtering algorithm, and constructing a strategy map based on the user portraits and the similarity between the user portraits, facilitating subsequent data mining, mining the strategy map using a recommendation model to obtain optimal behavior paths corresponding to each portrait cluster, i.e., operation instructions, improving the efficiency and quality of obtaining operation instructions, and the obtained operation instructions are more suitable for actual use.

[0158] The embodiment of the present application also provides a computer readable storage medium storing computer readable instructions, which can be executed by at least one processor to enable the at least one processor to perform the steps of the behavior path generation method as described above, by obtaining user behavior trajectory data, constructing a user portrait based on the user behavior trajectory data, clustering the user behavior portrait using a clustering algorithm to obtain a plurality of portrait clusters, classifying the types of behavior trajectories, facilitating subsequent derivation of operation instructions corresponding to each cluster, calculating the similarity between the user portraits in each portrait cluster using a collaborative filtering algorithm, and constructing a strategy map based on the user portraits and the similarity between the user portraits, facilitating subsequent data mining, mining the strategy map using a recommendation model to obtain optimal behavior paths corresponding to each portrait cluster, i.e., operation instructions, improving the efficiency and quality of obtaining operation instructions, and the obtained operation instructions are more suitable for actual use.

[0159] From the above description of the embodiments, those skilled in the art can clearly understand that the above embodiment method can be realized by means of software and a general hardware platform as required, and of course can also be realized by hardware, but in many cases the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium (such as a ROM / RAM, a magnetic disk, or an optical disk), and includes a plurality of instructions for causing a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the methods described in the various embodiments of the present application.

[0160] The behavior path generation device, computer device, and computer readable storage medium of the above embodiment of the present application have the same technical effects as the behavior path generation method of the above embodiment, and are not expanded here.

[0161] Obviously, the above-described embodiments are only some embodiments but not all embodiments of the present application, the preferred embodiments of the present application are shown in the drawings, but do not limit the patent scope of the present application. The present application can be implemented in many different forms, and conversely, the purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent replacements to some technical features therein. Any equivalent structure made by using the content of the specification and drawings, directly or indirectly applied to other related technical fields, is also within the patent protection scope of the present application.

Claims

1. A method for generating behavior paths, characterized in that, The method includes: Obtain user behavior trajectory data; Based on the user behavior trajectory data, a user profile is constructed; The user profiles are clustered using a clustering algorithm to obtain multiple profile clusters; Based on the dwell data of user profiles in each of the aforementioned profile clusters, a trajectory mulberry map corresponding to each profile cluster is created. The trajectory mulberry map is arranged according to the time corresponding to the dwell interface in the dwell data. The longer the dwell interface is, the larger its proportion. Based on the proportion of each dwell interface, the weight of each tag in the user profile is determined. The similarity between user profiles in each profile cluster is calculated using a collaborative filtering algorithm, and a strategy graph is constructed based on the user profiles and the similarity between them. The optimal behavioral path is obtained by mining the policy graph using a recommendation model. The acquisition of user behavior trajectory data includes: User behavior trajectory data is obtained by embedding data at the front end, and the user behavior trajectory data includes operation data and dwell data; The step of calculating the similarity between user profiles in each profile cluster using a collaborative filtering algorithm includes: The user profile is transformed to obtain the corresponding rating matrix; The scoring matrix is ​​edited based on the weight of each of the aforementioned tags; The similarity between each edited rating matrix is ​​calculated using a collaborative filtering algorithm to obtain the similarity between each user profile. The step of mining the policy graph using a recommendation model to obtain the optimal behavioral path includes: The recommendation model mines the relationships between nodes based on the similarity between nodes in the strategy graph to obtain the optimal behavior path, or mines the relationship between user profiles and latent classes, and operation behaviors and latent classes in the strategy graph through the recommendation model to determine the optimal behavior path. The recommendation model is trained based on a latent semantic model.

2. The behavior path generation method according to claim 1, characterized in that, After obtaining the optimal behavior path, the following is also included: Based on each of the aforementioned image clusters, the corresponding frequent pattern tree is constructed using the Fp-growth algorithm; Determine the degree of fit between the optimal behavior path and the frequent pattern tree; If the fit is greater than or equal to a preset value, the optimal behavior path is output to the front end and a prompt is given. If the fit is less than the preset value, the optimal behavior path is inserted into the frequent pattern tree, and then the step of mining the strategy graph through the recommendation model to obtain the optimal behavior path is executed again.

3. The behavior path generation method according to claim 1, characterized in that, The acquisition of user behavior trajectory data includes: User behavior trajectory data is obtained by embedding data at the front end. The user behavior trajectory data includes operation data and dwell data.

4. The behavior path generation method according to claim 3, characterized in that, The process of constructing a user profile based on the user behavior trajectory data includes: Based on the operation data and dwell data, user tags are assigned using preset tags; Based on the user tags, a user profile is constructed.

5. The behavior path generation method according to claim 1, characterized in that, The step of mining the policy graph using a recommendation model includes: The recommendation model mines the policy graph according to the following formula: Where, r ui This represents the predicted similarity between whether user u performs behavior i and the predicted similarity. q represents the time taken to scan user u to obtain the operation behavior vector. i p represents the operation vector of behavior i. u,k q represents the relationship between user u's preference behavior and the latent class k. i,k Let i represent the relationship between behavior i and the latent class k, where k = 1, 2, ..., n.

6. A behavior path generation device, characterized in that, The device includes: The acquisition module is used to acquire user behavior trajectory data; the user behavior trajectory data is obtained by embedding data on the front end, and the user behavior trajectory data includes operation data and dwell data. The user profile building module is used to build user profiles based on the user behavior trajectory data. The clustering module is used to cluster the user profiles using a clustering algorithm to obtain multiple profile clusters; The mapping module is used to create a trajectory mulberry map corresponding to each of the user profiles in each profile cluster based on the dwell data of the user profiles in each profile cluster. The trajectory mulberry map is arranged according to the time corresponding to the dwell interface in the dwell data. The longer the dwell interface is, the larger the proportion it occupies. The weight of each tag in the user profile is determined according to the proportion of each dwell interface. The graph construction module is used to calculate the similarity between user profiles in each of the profile clusters through a collaborative filtering algorithm, and to construct a strategy graph based on the user profiles and the similarity between the user profiles. The map construction module includes a transformation submodule, an editing submodule, and a similarity calculation submodule; The conversion submodule is used to convert the user profile to obtain the corresponding rating matrix; The editing submodule is used to edit the scoring matrix based on the weight of each of the tags; The similarity calculation submodule is used to calculate the similarity between each edited rating matrix through a collaborative filtering algorithm, thereby obtaining the similarity between each user profile. The recommendation module is used to mine the policy graph through a recommendation model to obtain the optimal behavior path; The recommendation module includes a first mining submodule; The first mining submodule is used by the recommendation model to mine the relationship between nodes based on the similarity between nodes in the strategy graph to obtain the optimal behavior path, or to mine the relationship between user profiles and latent classes, and operation behaviors and latent classes in the strategy graph through the recommendation model to determine the optimal behavior path, wherein the recommendation model is trained based on a latent semantic model.

7. A computer device, characterized in that, The computer device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores computer-readable instructions, and the processor, when executing the computer-readable instructions, implements the behavior path generation method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions that, when executed by a processor, implement the behavior path generation method as described in any one of claims 1 to 5.

Images