A mining method, device, and storage medium for association rules between applications

Abstract

The present disclosure provides a method, device and storage medium for mining association rules between applications. The present disclosure calculates the evaluation usage rate of applications based on time segments as the minimum support, filters out all applications that are greater than the minimum support to form a frequent item set L1, performs pairwise combinations of applications based on L1, calculates the credibility between applications to obtain a credibility set L2, and filters out items from L2 whose credibility from the previous application to the next application is greater than the product of the minimum support of the next application and a preset constant, forming a set R2 that expresses the association rules between applications. Further, a graph is constructed based on R2 and community division is performed using a community division algorithm. The present disclosure can actively push applications to users who use related applications more based on the mined association rules and community divisions, and can focus on operation and maintenance of core applications in the identified application community, thereby improving the intelligence and flexibility of association rule mining and improving the efficiency of the application maintenance system.

Description

Technical Field

[0001] The present disclosure relates to the field of Internet and communication technology, and in particular to a method, device, and storage medium for mining association rules between applications. Background Art

[0002] With the rapid development of information technology, the internet industry has accumulated a vast amount of network traffic data. However, simply relying on database queries and statistical methods makes it difficult to discover potential insights. Therefore, there is an urgent need for more intelligent methods to mine more valuable information. Data mining is a multidisciplinary field that integrates the latest research findings from database technology, artificial intelligence, machine learning, statistics, knowledge engineering, information retrieval, and other technologies. Big data mining of network traffic is undoubtedly one of the most important areas.

[0003] With the rapid development of various internet applications, the network is enabling people to do more and more things, and people's dependence on the network is growing. More and more tasks can be solved through online applications, and application traffic is experiencing explosive growth. When users use applications, they often follow certain patterns. By observing their application behavior, we can understand user characteristics and perform operations such as classification, clustering, and association, which can then be used to implement prediction and recommendation systems.

[0004] Association analysis is the task of finding relationships within large datasets. These relationships can take two forms: frequent itemsets or association rules. Frequent itemsets are collections of items that frequently appear together, while association rules suggest a strong relationship between two items. The Apriori algorithm is a classic data mining algorithm for mining frequent itemsets and association rules. It compresses the search space by leveraging prior properties of frequent itemsets.

[0005] Due to the various defects of existing association rule mining methods between applications, it is not possible to analyze and mine the association regulations between applications well. The association rules of applications with low usage rates are even more difficult to discover. Moreover, for applications with interdependent relationships, the core applications among these applications cannot be identified, resulting in low overall maintenance efficiency. Summary of the Invention

[0006] In view of this, the present disclosure provides a method and apparatus for mining association rules between applications, which are used to solve technical problems such as inaccurate mining of existing application association rules and inability to identify core applications in an application community.

[0007] Based on an embodiment of the present disclosure, the present disclosure provides a method for mining association rules between applications, the method comprising:

[0008] In multiple time periods of the analysis duration, the application usage information is grouped and counted by user identification code to obtain the set of applications used by each user in each time period;

[0009] Calculate the usage rate of each application in different time periods, and take the average usage rate of each application in different time periods as the minimum support of each application;

[0010] For each time period, calculate the support of each application in the time period, and filter out the applications with support greater than their own minimum support in the time period to form the item set L1;

[0011] Combine the items in L1 in pairs to form a candidate item set C2 that expresses the sequential application relationship between the two applications. Calculate the credibility of each item in C2 to obtain the credibility set L2 that expresses the sequential application relationship between the two applications.

[0012] Items whose credibility from the previous application to the next application is greater than the product of the minimum support of the next application and a preset constant are filtered out from L2 to obtain a set R2 expressing association rules between applications.

[0013] Based on the embodiment of the present disclosure, the method further includes:

[0014] The two applications before and after each item in the set R2 of association rules are respectively used as points in the graph, and the credibility of the two applications before and after is used as the weight of the edge between the two points to construct an application association graph;

[0015] A community partitioning algorithm is used to perform community partitioning on the application association graph to obtain application communities.

[0016] Based on the embodiment of the present disclosure, the method further includes:

[0017] The sum of the edge weights of the nodes corresponding to each application in each application community is calculated, and the application with the largest sum of the edge weights in each application community is determined as the core application of the application community.

[0018] Based on the embodiment of the present disclosure, further, before calculating the minimum support degree of each application, the method further includes: filtering out application data with a usage rate greater than a preset threshold.

[0019] Based on the embodiment of the present disclosure, further, the community division algorithm is the Louvain algorithm, and the preset constant is a constant greater than 1.2.

[0020] Based on an embodiment of the present disclosure, the present disclosure further provides a device for mining association rules between applications, the device comprising:

[0021] A group statistics module is used to group and count application usage information by user identification code within multiple time periods of the analysis duration to obtain the set of applications used by each user in each time period;

[0022] The minimum support calculation module is used to calculate the usage rate of each application in different time periods and take the average usage rate of each application in different time periods as the minimum support of each application;

[0023] The application screening module is used to calculate the support of each application in each time period, and screen out the applications with a support greater than the minimum support of the application in the time period to form the item set L1;

[0024] The credibility calculation module is used to combine the items in L1 in pairs to form a candidate item set C2 that expresses the relationship between the two applications in order, and calculate the credibility of each item in C2 to obtain the credibility set L2 that expresses the relationship between the two applications in order;

[0025] The association rule determination module is used to filter out items from L2 whose credibility from the previous application to the next application is greater than the product of the minimum support of the next application and a preset constant, and obtain a set R2 expressing the association rules between applications.

[0026] Based on the embodiment of the present disclosure, the device further includes:

[0027] A community partitioning module is configured to construct an application association graph by treating the two preceding and following applications of each item in the set of association rules R2 as points in a graph and the credibility of the two preceding and following applications as the weight of the edge between the two points; and to perform community partitioning on the application association graph using a community partitioning algorithm to obtain application communities;

[0028] The core application identification module is used to calculate the sum of the edge weights of the nodes corresponding to each application in each application community, and determine the application with the largest sum of the edge weights in each application community as the core application of the application community.

[0029] Based on the embodiment of the present disclosure, the device further includes:

[0030] The minimum support calculation module is further configured to filter out application data with a usage rate greater than a preset threshold before calculating the minimum support of each application.

[0031] Based on the embodiments of the present disclosure, the present disclosure also provides a storage medium, which stores a computer program. When the computer program in the storage medium is read and executed by a processor, it is used to complete the steps and functions of the method for mining association rules between any applications provided by the above-mentioned present disclosure.

[0032] The present disclosure calculates the evaluation usage rate of applications based on time segments as the minimum support, filters out all applications that are greater than the minimum support to form a frequent item set L1, performs pairwise combinations of applications based on L1, calculates the credibility between applications to obtain a credibility set L2, and filters out items from L2 whose credibility from the previous application to the next application is greater than the product of the minimum support of the next application and a preset constant, forming a set R2 that expresses the association rules between applications. Further, a graph is constructed based on R2 and a community division algorithm is used to perform community division. The present disclosure can actively push applications to users who use more related applications based on the mined association rules and community divisions, and can focus on the operation and maintenance of core applications in the identified application community, thereby improving the intelligence and flexibility of association rule mining and improving the efficiency of the application maintenance system. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments of the present disclosure or the description of the prior art. Obviously, the drawings described below are only some embodiments recorded in the present disclosure. For ordinary technicians in this field, other drawings can also be obtained based on these drawings of the embodiments of the present disclosure.

[0034] Figure 1 A flow chart of a method for mining association rules between applications provided by an embodiment of the present disclosure;

[0035] Figure 2 A structural diagram of a device for mining association rules between applications provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0036] The terms used in the embodiments of the present disclosure are intended only to describe specific embodiments and are not intended to limit the embodiments of the present disclosure. The singular forms "a," "an," "the," and "the" used in the embodiments of the present disclosure and the claims are intended to include the plural forms, unless the context clearly indicates otherwise. The term "and / or" used in the present disclosure refers to any or all possible combinations of one or more of the associated listed items.

[0037] It should be understood that although the terms first, second, third, etc. may be used to describe various information in the embodiments of the present disclosure, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of the embodiments of the present disclosure, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" used may be interpreted as "at the time of" or "when" or "in response to determining".

[0038] One of the main purposes of the embodiments of the present disclosure is to provide a method for mining association rules between applications, which is an improvement on the method for mining association rules between applications based on the Apriori algorithm. The Apriori algorithm is a classic data mining algorithm for mining frequent item sets and association rules. The algorithm uses the a priori property of frequent item sets, that is, all non-empty subsets of frequent item sets must also be frequent. The Apriori algorithm uses an iterative method called layer-by-layer search, in which k item sets are used to explore the (k+1) item set.

[0039] In this disclosure, the meanings of some basic terms used in this disclosure are explained as follows:

[0040] Item and item set: Let appset = {app1, app2, ..., app m} is a collection of all applications, where app i The subscript i represents the i-th application. The set of items is called an itemset appset, and the itemset containing k items is called an application k-itemset, denoted as k-appset.

[0041] Support: The proportion of records containing the item set in the dataset to the total records, which is used to measure the frequency of a set in the original data. For example, the i-th application app i The support level is the ratio of users using the application to the total number of users, which is expressed as follows:

[0042]

[0043] Confidence: used to measure the likelihood of a rule occurring, such as when a user is using an app i Then use the app n The credibility is equal to app i and app n The support of two applications divided by the app i The support used alone is expressed as follows:

[0044]

[0045] This disclosed embodiment first utilizes a traffic collector to preprocess network traffic. This preprocessing includes network traffic collection, data matching and filtering, and user-based network application data merging. Based on this preprocessing, this disclosed embodiment proposes a method for determining the minimum support of association rules between applications within a preset analysis duration, using time-based grouping. This method then designs a framework for analyzing the associations between users and applications, encompassing data processing, data analysis, data mining, and application community segmentation.

[0046] Figure 1 A flowchart of a method for mining association rules between applications provided in an embodiment of the present disclosure includes:

[0047] Step 101: In multiple time periods of the analysis duration, group and count application usage information by user identification code to obtain a set of applications used by each user in each time period.

[0048] One of the purposes of the present disclosure is to analyze the correlation between two applications used simultaneously by a user within a certain time range. In order to more accurately determine the association rules between applications, the present disclosure proposes a solution of segmenting the analysis time and then performing segmented statistics. The analysis time is used to determine the time range for analyzing application association rules, for example, it can be 1 day, 1 week, etc. The division of time periods can be based on user living habits or application usage habits in different geographical environments, or divided in fixed time units, etc. The division of time periods needs to take into account the analysis accuracy and efficiency of association rules. If the time period is too short, the correlation between certain applications may be very small, and if the time period is too long, the application set may be too large, making the analysis inefficient.

[0049] Step 102: Calculate the usage rate of each application in different time periods, and use the average value of the usage rate of each application in different time periods as the minimum support degree of each application.

[0050] This step calculates the usage rate of each application across different time periods based on the aggregate data of applications used by each user during each time period within the analysis duration. The average usage rate for each application across these time periods is then used as the minimum support for each application. This dynamic average is more flexible and intelligent than a fixed threshold, and it can prevent low-usage applications from being easily discovered.

[0051] Step 103: For each time period, calculate the support of each application in the time period, and filter out the applications with support greater than their own minimum support in the time period to form the item set L1.

[0052] Step 104: Combine the items in L1 in pairs to form a candidate item set C2 that expresses the relationship between the two applications. Calculate the credibility of each item in C2 to obtain a credibility set L2 that expresses the relationship between the two applications.

[0053] Step 105: Filter out items from L2 whose credibility from the previous application to the next application is greater than the product of the minimum support of the next application and a preset constant, and obtain a set R2 expressing association rules between applications.

[0054] In order to ensure the credibility of the association rules when mining the association rules between two applications, the disclosed embodiment further determines whether the two applications have a strong association rule based on the calculated credibility of the two applications being used together and the relationship between the credibility and the minimum support of the latter application. Only when the credibility of the two applications being used successively is greater than the product of the minimum support of the latter application and a preset constant, are the two applications considered to have a strong association rule, thereby further ensuring the credibility of the association rule.

[0055] In one embodiment of the present disclosure, in order to identify and divide application communities, mine community relationships between applications, and identify core applications in application communities, the method further includes:

[0056] Step 106: Take the two preceding and following applications of each item in the association rule set R2 as points in the graph, and use the credibility of the two preceding and following applications as the weight of the edge between the two points to construct an application association graph.

[0057] Step 107: Use a community partitioning algorithm to perform community partitioning on the application association graph to obtain application communities.

[0058] This step processes the application association graph through a community partitioning algorithm to obtain one or more application communities. The application community reflects the community relationship between a group of applications. The application store can recommend applications in the same community to users based on the application community, thereby improving the intelligence and user experience of similar application stores.

[0059] Step 108: Calculate the sum of the edge weights of the nodes corresponding to each application in each application community, and determine the application with the largest sum of edge weights in each application community as the core application of the application community.

[0060] By identifying the core applications in the application community, we can discover the core applications in the application community and focus on maintaining the core applications, which can improve the availability and stability of the operation and maintenance system, and enhance maintenance efficiency and user experience.

[0061] Based on the above embodiments of the present disclosure, it can be seen that the embodiments of the present disclosure have improved the existing application association rule mining method. The present disclosure calculates the application usage rate in different time periods and automatically sets the support threshold for each application, thereby improving the intelligence and adaptability of the application association rule mining method and solving the problem that the traditional Apriori algorithm cannot take into account all situations. In addition, when judging whether it is a strong association rule, it is compared with the minimum support of the following application to further ensure credibility. The embodiments of the present disclosure further use the association rules between the mined applications as points and edges in the graph algorithm, and use the community partitioning algorithm to obtain the application community graph, and transform closely related applications into the same community. The application association rules can predict the user's behavior in accessing applications and recommend a certain application that may be needed to the user. The application community graph can actively push applications in the same community to users who use this community application.

[0062] The steps of the method for mining association rules between applications provided by the present disclosure are described in detail below in conjunction with specific embodiments.

[0063] Step 201: Use traffic collection technology to collect user network traffic data and obtain application usage information therefrom. The application usage information at least includes application usage information such as user identification code and usage time.

[0064] This step first uses a traffic collector to obtain network data and stores the data in a database. Then, Python is used to process the data, including filtering, grouping, and integration.

[0065] Network traffic data is collected using deep packet inspection technology in traffic collectors. This technology not only captures data such as the source IP address, destination IP address, source port, destination port, and protocol within traffic packets, but also performs content inspection and deep decoding of application-layer data. However, the data obtained from deep packet inspection is often redundant, so it is sufficient to extract only the required application usage information. This information includes user identification codes / user IDs, application names, usage time, traffic volume, and packet counts. This data is then stored in a data warehouse to form tables for application correlation analysis.

[0066] Association analysis data table

[0067] Field Field meaning user_id User identification code apply_name Application Name log_time Usage time flow flow pack Number of packages app_class Application Category

[0068] Step 202: In multiple time periods of the analysis duration, group and count the acquired application usage information by user identification code to obtain a set of applications used by each user in each time period.

[0069] In one embodiment of the present disclosure, the analysis duration is set to one day, and the 24-hour day is divided into seven time periods. From midnight to five o'clock is the late night period; from six to eight o'clock is the morning period; from nine to eleven o'clock is the morning period; from twelve to thirty-three o'clock is the noon period; from fourteen to seventeen o'clock is the afternoon period; from eighteen to nineteen o'clock is the evening period; and from eight to twenty-three o'clock is the night period.

[0070] Time period division

[0071]

[0072]

[0073] For some popular applications, such as WeChat, the applications themselves are relatively independent and are often used alone. Therefore, the traffic data of these applications needs to be filtered out from the traffic data collected by the collector. Otherwise, the frequent itemsets mined will almost entirely consist of these popular applications, which is not conducive to discovering relationships between other applications with low usage, and the mined rules are also meaningless. Therefore, in one embodiment of the present disclosure, by calculating the usage rate of each application, application data with a usage rate greater than a preset threshold is filtered out, for example, application data with a usage rate greater than 1.5 times the interquartile range is filtered out, where the interquartile range is the difference between the third quartile and the first quartile.

[0074] The processed data is grouped and counted according to different user identification codes in different time periods, and the applications used by the users are merged to obtain the application sets used by each user in each time period. These sets are used as input for association rule mining.

[0075] Step 203: Calculate the usage rate of each application in different time periods, and use the average value of the usage rate of each application in different time periods as the minimum support degree of each application.

[0076] The Apriori algorithm uses an iterative method called layer-by-layer search, where k itemsets are used to explore (k+1) itemsets. The disclosed embodiment searches only for frequent 2-itemsets. First, by scanning the collected data set, accumulating the counts of each item, and collecting items that meet the minimum support, a set of frequent 1-itemsets is found, denoted as L1. Then, L1 is used to find a set of frequent 2-itemsets, L2, and finally the association rule R2 is applied.

[0077] This step is to calculate the usage rate of each application in different time periods based on the application set data used by each user in each time period within the total analysis time range, and take the average usage rate of each application in different time periods as the minimum support of each application.

[0078]

[0079] Among them, user_num is the total number of users in period t, user_app m _n is the time period for using the app m The number of users.

[0080] Step 204: For each time period, obtain the support of each application in the time period, and filter out the applications with support greater than their own minimum support in the time period to form a frequent 1-item set L1.

[0081] In this step, a candidate set C1 is constructed for each time period. C1 is a one-dimensional candidate item set. C1 is obtained by counting and calculating the number of applications in each time period. Each element of C1 contains information about the application identifier and application support. For example:

[0082]

[0083] Then, scan C1 in each time period to determine whether each application in the time period is greater than its minimum support, and form the applications with greater than its minimum support in each time period into a frequent 1-item set L1.

[0084]

[0085] Step 205: Combine the elements in L1 in pairs to form a candidate item set C2 that expresses the relationship between the two applications. Calculate the credibility of each item in C2 to obtain the credibility set L2 that expresses the relationship between the two applications.

[0086] In this step, first, the elements in L1 are combined into candidate set C2. Each item in C2 includes the application identifiers of two applications and the support information of the combination of the two applications:

[0087]

[0088] Collection (app a , app b ) Regardless of the order of application, there is an order relationship between applications when calculating the two-item set L2. For each item in C2, the following two credibility measures are calculated:

[0089] and

[0090]

[0091] After calculating the credibility of each item in C2, the set L2 is obtained.

[0092]

[0093] Each item in L2 contains the association rules and credibility information between two applications.

[0094] Step 206: Filter out items from L2 whose credibility from the previous application to the next application is greater than the product of the minimum support of the next application and a preset constant, to form a set R2 expressing the association regulation between the two applications.

[0095] Scan L2 to determine whether the credibility of each item is greater than the product of the minimum support of the next application in the item and a preset constant. The items that are greater than the product of the minimum support of the next application and the preset constant constitute the set R2 that expresses the association regulation between the two applications.

[0096]

[0097] Here, K is a preset constant. In one embodiment of the present disclosure, K is a constant value greater than 1.2, for example, 1.4.

[0098] Through the above steps, the association rules of users accessing applications in different time periods can be obtained. The user's application access behavior can be predicted based on the association rules, and applications with associated relationships can be recommended to users.

[0099] Step 207: Take the two preceding and following applications of each item in the association rule set R2 as points in the graph, and use the credibility of the two preceding and following applications as the weight of the edge between the two points to construct an application association graph.

[0100] This step associates each app in the rule set R2 a , app b As a point, As a point app a Daodian app b The weights of the edges are used to construct the application association graph G. The purpose of constructing the application association graph is to divide the application communities.

[0101] Step 208: Use the application association graph G as the input of the community division algorithm to obtain different application communities based on application association relationships. Applications within a community are closely connected, while applications between communities are sparsely connected.

[0102] Modularity is used to measure the quality of a community network division. It can be simply understood as the sum of all edge weights within the community minus the sum of edge weights connected to the community. It is defined as follows:

[0103]

[0104]

[0105] Among them, A ij represents the weight of the edge between node i and node j; ki =∑ j A ij represents the sum of the weights (degrees) of all edges connected to node i; c i represents the community to which node i belongs; represents the sum of the weights of all edges.

[0106] The Louvain algorithm is a modularity-based community discovery algorithm. It performs well in terms of efficiency and effectiveness, and can discover hierarchical community structures. Its optimization goal is to maximize the modularity of the entire community network. One embodiment of the present disclosure utilizes the Louvain algorithm to implement community partitioning between applications.

[0107] The process of using the Louvain algorithm to implement community division between applications is as follows:

[0108] 1) Treat each node in the graph as an independent community, and the number of communities is the same as the number of nodes;

[0109] 2) For each node i, try to assign node i to the community of each of its neighboring nodes in turn, calculate the modularity change ΔQ before and after the assignment, and record the neighboring node with the largest ΔQ. If maxΔQ>0, assign node i to the community of the neighboring node with the largest ΔQ, otherwise it remains unchanged.

[0110] 3) Repeat 2) until the communities to which all nodes belong no longer change;

[0111] 4) Compress the graph, compressing all nodes in the same community into a new node. The edge weights between nodes in the community are converted into the weights of the new node ring, and the edge weights between communities are converted into the edge weights between new nodes. The new node ring means that after the graph is compressed, the points in the same community are used as new nodes. The weight at this time is the sum of the weights of the compressed points in the same community.

[0112] 5) Repeat 1)-4) until the modularity of the entire graph no longer changes.

[0113] 6) Each community will be k i The node i with the largest degree is recorded as the community core, where k i =∑ j A ij represents the sum of the weights (degrees) of all edges connected to node i, A ij Represents the weight of the edge between node i and node j. Then the community partition result is returned as follows:

[0114] {Community 1: [(community core, application a), (community members, application b), ...];

[0115] Community 2: [(community core, application m), (community members, application n), …];

[0116] …}

[0117] The application community division results are obtained in this way, which can focus on the operation and maintenance of core applications in the application community, and can also recommend applications in the same community to users who use this community application.

[0118] Figure 2 This is a schematic diagram of the structure of a device for mining association rules between applications provided in one embodiment of the present disclosure. The various functional modules in the device can be implemented in the form of software modules or hardware units. The functions of the various modules of the device correspond to the various steps in the method for mining association rules between applications provided by the embodiment of the present disclosure. The various modules in the device can be executed on one hardware device, or different hardware devices can respectively complete one or more steps or module functions in the method provided by the present disclosure. The device 200 includes: a group statistics module 201, a minimum support calculation module 202, an application screening module 203, a credibility calculation module 204, and an association rule determination module 205.

[0119] The grouping statistics module 201 is used to group and count application usage information by user identification code in multiple time periods of the analysis duration to obtain the set of applications used by each user in each time period;

[0120] The minimum support calculation module 202 is used to calculate the usage rate of each application in different time periods, and use the average of the usage rates of each application in different time periods as the minimum support of each application;

[0121] Application screening module 203, for each time period, calculating the support of each application in the time period, screening out the applications with support greater than the minimum support of the application in the time period, and forming the itemset L1;

[0122] Credibility calculation module 204 is used to combine items in L1 in pairs to form a candidate item set C2 expressing the sequential application relationship between the two applications, calculate the credibility of each item in C2, and obtain a credibility set L2 expressing the sequential application relationship between the two applications;

[0123] The association rule determination module 205 is configured to filter out from L2 items whose credibility from the previous application to the next application is greater than the product of the minimum support of the next application and a preset constant, and obtain a set R2 expressing association rules between applications.

[0124] To achieve application community division, in one embodiment of the present disclosure, the apparatus further includes:

[0125] The community partitioning module 206 is configured to construct an application association graph by treating the two preceding and following applications of each item in the set of association rules R2 as points in a graph and using the credibility of the two preceding and following applications as the weight of the edge between the two points; and to perform community partitioning on the application association graph using a community partitioning algorithm to obtain application communities.

[0126] To achieve core application identification, in one embodiment of the present disclosure, the device also includes: a core application identification module 207, which is used to calculate the sum of the edge weights of the nodes corresponding to each application in each application community, and determine the application with the largest sum of the edge weights in each application community as the core application of the application community.

[0127] To filter out frequently used independent applications, in one embodiment of the present disclosure, the minimum support calculation module 202 is further configured to filter out application data with a usage rate greater than a preset threshold before calculating the minimum support of each application.

[0128] In another embodiment of the present disclosure, a storage medium is also provided. The storage medium is located in a device having a processor and a bus structure. The storage medium can be a volatile storage medium or a non-volatile storage medium. A computer program is stored in the storage medium. When the computer program in the storage medium is read and executed by the processor, it can be used to complete the steps and functions of the method for mining association rules between applications provided in the above-mentioned embodiment of the present disclosure.

[0129] The foregoing is merely an embodiment of the present disclosure and is not intended to limit the present disclosure. Various modifications and variations are possible for those skilled in the art. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present disclosure are intended to be included within the scope of the claims of the present disclosure.

Claims

1. A method for mining association rules between applications, characterized in that: The method comprises: In multiple time periods of the analysis duration, the application usage information is grouped and counted by user identification code to obtain the set of applications used by each user in each time period; Based on the obtained application set used by each user in each time period, the usage rate of each application in different time periods is calculated, and the average usage rate of each application in different time periods is used as the minimum support of each application; For each time period, calculate the support of each application in the time period, and filter out the applications with support greater than their own minimum support in the time period to form the item set L1; Combine the items in L1 in pairs to form a candidate item set C2 that expresses the relationship between the two applications. Calculate the credibility of each item in C2 to obtain the credibility set L2 that expresses the relationship between the two applications. Filter out from L2 the items whose credibility from the previous application to the next application is greater than the product of the minimum support of the next application and a preset constant, and obtain the set R2 expressing the association rules between the applications; Calculating the support level of each application in the time period includes: calculating, for each application, the ratio of the number of users who used the application in the time period to the total number of users as the support level of the application; The calculation of the credibility of each item in C2 includes: calculating the support degree of the first application and the second application being used successively divided by the support degree of the first application being used alone, where the first application is the application used first in the item and the second application is the application used later in the item.

2. The method according to claim 1, characterized in that The method further comprises: The two applications before and after each item in the set R2 of association rules are respectively used as points in the graph, and the credibility of the two applications before and after is used as the weight of the edge between the two points to construct an application association graph; A community partitioning algorithm is used to perform community partitioning on the application association graph to obtain application communities.

3. The method according to claim 2, characterized in that The method further comprises: The sum of the edge weights of the nodes corresponding to each application in each application community is calculated, and the application with the largest sum of the edge weights in each application community is determined as the core application of the application community.

4. The method according to claim 1, wherein Before calculating the minimum support degree of each application, the method further includes filtering out application data with a usage rate greater than a preset threshold.

5. The method according to claim 2, characterized in that The community division algorithm is the Louvain algorithm, and the preset constant is 1.

4.

6. A device for mining association rules between applications, characterized in that: The device comprises: A group statistics module is used to group and count application usage information by user identification code within multiple time periods of the analysis duration, and obtain the set of applications used by each user in each time period; A minimum support calculation module is used to calculate the usage rate of each application in different time periods based on the obtained application set used by each user in each time period, and to take the average usage rate of each application in different time periods as the minimum support of each application; The application screening module is used to calculate the support of each application in each time period, and screen out the applications with a support greater than the minimum support of the application in the time period to form the item set L1; The credibility calculation module is used to combine the items in L1 in pairs to form a candidate item set C2 that expresses the relationship between the two applications in order, and calculate the credibility of each item in C2 to obtain the credibility set L2 that expresses the relationship between the two applications in order; An association rule determination module is used to filter out from L2 the items whose credibility from the previous application to the next application is greater than the product of the minimum support of the next application and a preset constant, and obtain a set R2 expressing the association rules between the applications; Calculating the support level of each application in the time period includes: calculating, for each application, the ratio of the number of users who used the application in the time period to the total number of users as the support level of the application; The calculation of the credibility of each item in C2 includes: calculating the support degree of the first application and the second application being used successively divided by the support degree of the first application being used alone, where the first application is the application used first in the item and the second application is the application used later in the item.

7. The device according to claim 6, characterized in that The device further comprises: A community partitioning module is configured to construct an application association graph by treating the two preceding and following applications of each item in the set of association rules R2 as points in a graph and the credibility of the two preceding and following applications as the weight of the edge between the two points; and to perform community partitioning on the application association graph using a community partitioning algorithm to obtain application communities; The core application identification module is used to calculate the sum of the edge weights of the nodes corresponding to each application in each application community, and determine the application with the largest sum of the edge weights in each application community as the core application of the application community.

8. The device according to claim 6, characterized in that The device further comprises: The minimum support calculation module is further configured to filter out application data with a usage rate greater than a preset threshold before calculating the minimum support of each application.

9. The device according to claim 7, characterized in that The community division algorithm is the Louvain algorithm, and the preset constant is 1.

4.

10. A storage medium storing a computer program, wherein: When the computer program in the storage medium is read and executed by the processor, it is used to complete the steps and functions of the method according to any one of claims 1 to 5.

Images