User presence behavior prediction method and device

Abstract

The embodiments of the present specification provide a method and device for predicting user presence behavior, wherein a method for predicting user presence behavior includes: calculating the behavior similarity between a target user and a sampled user based on application access data and behavior trajectory data, and determining at least one sampled user as a similar user of the target user; predicting the presence behavior of the similar users in a target time period; assigning weights to the similar users according to the behavior similarity, and calculating prediction scores of similar users whose presence behaviors belong to the same behavior category based on the assigned prediction weights; and determining the predicted presence behavior of the target user in the target time period based on the prediction score.

Description

Technical Field

[0001] This document relates to the field of data processing technology, and in particular to a method and device for predicting user presence behavior. Background Art

[0002] With the rapid development of mobile Internet and the rapid growth of Internet users, more and more information transmission and advertising rely on various applications of mobile Internet. Users can access applications to obtain corresponding data information. In this process, a huge amount of data with user time and space attributes is generated. Usually, massive user data can be used for data processing to explore potential data patterns and provide users with better services. Summary of the invention

[0003] One or more embodiments of the present specification provide a method for predicting user presence behavior, including: calculating the behavioral similarity between a target user and a sampled user based on application access data and behavioral trajectory data, and determining at least one sampled user as a similar user of the target user. Predicting the presence behavior of the similar users in a target time period. Assigning weights to the similar users according to the behavioral similarity, and calculating the predicted scores of similar users whose presence behaviors belong to the same behavior category based on the assigned prediction weights. Determining the predicted presence behavior of the target user in the target time period based on the prediction score.

[0004] One or more embodiments of the present specification provide a user behavior prediction device, including: a similarity calculation module, configured to calculate the behavioral similarity between a target user and a sampled user based on application access data and behavioral trajectory data, and determine at least one sampled user as a similar user of the target user. A presence behavior prediction module, configured to predict the presence behavior of the similar users in a target time period. A prediction score calculation module, configured to assign weights to the similar users according to the behavioral similarity, and calculate the prediction scores of similar users whose presence behaviors belong to the same behavior category based on the assigned prediction weights. A presence behavior determination module, configured to determine the predicted presence behavior of the target user in the target time period based on the prediction score.

[0005] One or more embodiments of the present specification provide a user behavior prediction device, including: a processor; and a memory configured to store computer executable instructions, wherein when the computer executable instructions are executed, the processor: calculates the behavioral similarity between a target user and a sampled user based on application access data and behavioral trajectory data, and determines at least one sampled user as a similar user of the target user. Predict the presence behavior of the similar users in a target time period. Assign weights to the similar users according to the behavioral similarity, and calculate the prediction score of similar users whose presence behavior belongs to the same behavior category based on the assigned prediction weights. Determine the predicted presence behavior of the target user in the target time period based on the prediction score.

[0006] One or more embodiments of the present specification provide a storage medium for storing computer executable instructions, which implement the following process when executed by a processor: Calculate the behavioral similarity between a target user and a sampled user based on application access data and behavioral trajectory data, and determine at least one sampled user as a similar user of the target user. Predict the presence behavior of the similar users in a target time period. Assign weights to the similar users according to the behavioral similarity, and calculate the predicted scores of similar users whose presence behaviors belong to the same behavior category based on the assigned prediction weights. Determine the predicted presence behavior of the target user in the target time period based on the prediction score. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] In order to more clearly illustrate one or more embodiments of the present specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present specification. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative labor.

[0008] Figure 1 A processing flow chart of a method for predicting user presence behavior provided in one or more embodiments of this specification;

[0009] Figure 2 A processing flow chart of a method for predicting user presence behavior applied to a service push scenario provided by one or more embodiments of this specification;

[0010] Figure 3 A schematic diagram of a user behavior prediction device provided in one or more embodiments of this specification;

[0011] Figure 4 A schematic diagram of the structure of a user behavior prediction device provided in one or more embodiments of this specification. DETAILED DESCRIPTION

[0012] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below in conjunction with the drawings in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of them. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this document.

[0013] This specification provides an embodiment of a method for predicting user presence behavior:

[0014] Reference Figure 1 , which shows a processing flow chart of a method for predicting user presence behavior provided by this embodiment, referring to Figure 2 , which shows a processing flow chart of a user presence behavior prediction method applied to a service push scenario provided by this embodiment.

[0015] Reference Figure 1 The user presence behavior prediction method provided in this embodiment specifically includes steps S102 to S108.

[0016] Step S102: Calculate the behavior similarity between the target user and the sampled users based on the application access data and the behavior trajectory data, and determine at least one sampled user as a similar user to the target user.

[0017] In the user presence behavior prediction method provided by the present embodiment, the presence behavior of the target user in the target time period is first predicted. If the prediction result obtained by the prediction is unpredictable, the predicted presence behavior of the target user is determined with the help of the predicted presence behavior of similar users in the target time period. Specifically, in the process of predicting the presence behavior of similar users in the target time period, a weight is assigned to each similar user according to the behavior similarity, and the prediction score of similar users whose predicted presence behavior belongs to the same behavior category is calculated. The predicted presence behavior of the target user in the target time period is further determined based on the prediction score. If the predicted presence behavior belongs to the presence state, a service message is pushed to the target user in the target time period.

[0018] By introducing the prediction of the presence behavior of the target user in the target time period and using the prediction of the presence behavior of similar users in the target time period to complete the joint processing mechanism of the presence behavior prediction of the target user, high accuracy and high reliability of the presence behavior prediction of the target user can be achieved, and the confidence and timeliness of the prediction results can be improved.

[0019] In the actual document service push scenario, users can perform real-name registration by uploading document information such as identity credentials. However, the operation process of uploading document information is often cumbersome, and users cannot complete the task of uploading document information without carrying their documents. Since users are in the target field (such as residence), they generally have more free time. In this case, if it can be predicted that the user is in the target field or target route, service messages can be pushed to the user in a targeted manner, avoiding waste of data resources, increasing the probability of users triggering service information, and improving user conversion efficiency.

[0020] In this embodiment, the application access data and the behavior trajectory data are generated based on historical access records, such as when a user uploads the geographical location information of the location when accessing the application at the (x, y) position at time T; the behavior similarity represents the degree of similarity between the on-site behaviors of two users; in actual applications, users can upload their own geographical location information when accessing the application. Here, users who access the application and upload geographical location information are historical users. A certain number of users are extracted from the historical users as samples for behavior similarity calculation. The certain number of users are sampled users, and the target users are users for whom on-site behavior prediction for the target time period is required.

[0021] The presence behavior includes the presence state or the absence state.

[0022] In the process of service processing, the presence state refers to the user's location meeting the location required for the current service processing. Conversely, when the user's location does not meet the location required for the current service processing, the user's presence behavior is determined to be non-presence. For example, in the process of service message push, the user's frequency of application access is higher when he is at home. In this case, the message push is effective. Therefore, the presence state includes the user's location at home. At the same time, when the user is in some specific places (such as restaurants, taking public transportation on the way home from get off work), the frequency of application access is also higher. Therefore, the presence state also includes the user's location at a place or route where the application access frequency meets the access conditions.

[0023] In specific implementation, in order to improve the accuracy of the prediction result, the behavior similarity between the target user and the sampled user can be calculated in the process of determining the predicted presence behavior of the target user based on the predicted presence behavior of similar users in the target time period. In an optional implementation provided by this embodiment, in the process of calculating the behavior similarity between the target user and the sampled user, the following operations are performed:

[0024] Extracting the sampled user from the application access data and behavior trajectory data of historical users;

[0025] constructing a behavior feature vector according to the application access data and behavior trajectory data of the target user, and constructing a behavior feature vector according to the application access data and behavior trajectory data of the sampled user;

[0026] The behavior feature vector of the target user and the behavior feature vector of the sampled user are input into a similarity algorithm to calculate the behavior similarity, and the behavior similarity between the target user and the sampled user is output.

[0027] For example, in the process of determining whether to send a real-name registration to user u, whether to send a service message push to user u can be determined by judging whether user u is in the target field during the 19:00-22:00 time period. In this case, a behavior feature vector constructed based on the application access data and behavior trajectory data of user u can be expressed as: "Ua-Features = {f1, f2, f3, ..., fn}", and a behavior feature vector constructed based on the application access data and behavior trajectory data of the sampled users of user u can be expressed as: "Ubx-Features = {fx1, fx2, fx3, ..., fxn}, where x is the user number of each sampled user, x = {1b, 2b, 3b, ..., nb}", and the obtained behavior feature vector of user u and the behavior feature vector of the sampled user are input into the similarity algorithm, and the behavior similarity between user u and the sampled user can be output.

[0028] The similarity algorithm can be calculated by calculating the distance between the feature vectors of the target user and similar users. If the distance is small, the similarity is large; if the distance is large, the similarity is small. The similarity between the target user and similar users can also be measured by measuring the difference in direction between the two feature vectors. Specifically, in the process of similarity calculation, the following three candidate similarity algorithms are provided:

[0029] (1) Euclidean Distance

[0030] For example, the constructed behavior feature vector of user u is: Ua-Features = {f1, f2, f3, ..., fn}, and the constructed behavior feature vector of any sampled user is: Ubx-Features = {fx1, fx2, fx3, ..., fxn}, x = {1b, 2b, 3b, ..., nb}, then the Euclidean distance, that is, the behavior similarity between user u and any sampled user, is:

[0031]

[0032] (2) Manhattan Distance

[0033] For example, the constructed behavior feature vector of user u is: Ua-Features = {f1, f2, f3, ..., fn}, and the constructed behavior feature vector of any sampled user is: Ubx-Features = {fx1, fx2, fx3, ..., fxn}, then the Manhattan distance, that is, the behavior similarity between user u and any sampled user, is:

[0034] (3) Cosine distance

[0035] For example, the constructed behavior feature vector of user u is: Ua-Features = {f1, f2, f3, ..., fn}, and the constructed behavior feature vector of any sampled user is: Ubx-Features = {fx1, fx2, fx3, ..., fxn}, then the cosine distance, that is, the behavior similarity between user u and any sampled user, is:

[0036] In addition, in addition to the above-mentioned implementation method of using Euclidean distance, Manhattan distance and cosine distance to calculate similarity, other similarity algorithms may also be used, which are not specifically limited in this embodiment.

[0037] Since the target user's presence behavior is divided into a presence state and a non-presence state, in order to promote the accurate generation of prediction results and make the target user's predicted presence behavior closer to the actual presence behavior, which is helpful for the subsequent targeted push of service messages, in the process of constructing the behavior feature vector according to the application access data and behavior trajectory data of the sampled users, the sampled users can be classified to generate positive sample users and negative sample users. The positive sample users are the presence sample users, and the negative sample users are the non-presence sample users.

[0038] In an optional implementation manner provided by this embodiment, in the process of constructing a behavior feature vector according to the application access data and behavior trajectory data of the sampled user, the following operations are performed:

[0039] Extracting positive sample users and negative sample users from the application access data and behavior trajectory data of the sampled users;

[0040] Extracting time feature elements and presence behavior elements through the application access data and behavior trajectory data of the positive sample user and the application access data and behavior trajectory data of the negative sample user;

[0041] A behavior feature vector of the sampled user is constructed according to the time feature element and the presence behavior element.

[0042] The time feature element includes at least one of the following: working day, weekend, holiday, time period, date range, frequency; the presence behavior element includes presence or absence.

[0043] It should be noted that, based on the extraction of time feature elements and presence behavior feature elements, the time feature elements are used to generate time behavior features, and the presence behavior elements are used to generate presence behavior features, and then the time feature vector is constructed through the time behavior features, and the presence behavior feature vector is constructed through the presence behavior features, and then the time feature vector and the presence behavior feature vector are integrated to form a behavior feature vector; it is also possible to use the time feature elements to generate time behavior features, and use the presence behavior elements to generate presence behavior features, and then integrate the time behavior features and the presence behavior features to form user behavior features, and quantize the user behavior features to generate a behavior feature vector; it is also possible to first integrate the time feature elements and the presence behavior elements to form user behavior elements, and then use the user behavior elements to generate user behavior features, and then construct a behavior feature vector through the user behavior features.

[0044] For example, the user behavior feature formed by integrating the time behavior feature and the presence behavior feature may be: a certain sampled user is present at 21:00 on Monday, and a certain sampled user is not present at 12:00 on Saturday.

[0045] Similarly, in the above process of constructing a behavior feature vector based on the target user's application access data and behavior trajectory data, time feature elements and presence behavior elements can be extracted from the target user's application access data and behavior trajectory data, and then the target user's behavior feature vector can be constructed based on the time feature elements and presence behavior elements.

[0046] It should be added that, in addition to extracting time feature elements and presence behavior elements through the application access data and behavior trajectory data of the positive sample users and the application access data and behavior trajectory data of the negative sample users, it is also possible to extract either the time feature elements or the presence behavior elements through the application access data and behavior trajectory data of the positive sample users and the application access data and behavior trajectory data of the negative sample users.

[0047] After the behavior similarity calculation is completed, similar users of the target user may be further determined. Specifically, in an optional implementation provided by this embodiment, in the process of determining at least one sampled user as a similar user of the target user, the following operations are performed:

[0048] Selecting a sampled user that matches the geographic location of the target user from the sampled users;

[0049] The sampled users obtained by screening are sorted in descending order of the behavior similarity, and the sampled users whose sorting positions are before the preset positions are determined as the similar users.

[0050] Continuing with the above example, based on the behavioral similarity calculation using the behavioral feature vector of user u: "Ua-Features = {f1, f2, f3, ..., fn}" and the behavioral feature vector of the sampled users of user u: "Ubx-Features = {fx1, fx2, fx3, ..., fxn}", similar users of the target user can be determined based on the behavioral similarity. In the process of determining similar users, sampled users matching the city location of user u are screened out from the sampled users, and sampled users that do not meet the city location conditions are eliminated. Then, the screened sampled users are sorted in descending order according to the calculated behavioral similarity, and the sampled users whose sorting positions are before the preset positions are determined as similar users, expressed as: "U-similarity = {u1, u2, ..., un}", so as to increase the reliability of user data and achieve high efficiency of the prediction process.

[0051] Furthermore, in order to save prediction time and improve the timeliness of prediction results, so that users can receive service message push in a shorter time, before predicting the presence behavior of similar users in the target time period, the presence behavior of the target user in the target time period can be predicted based on the target user's application access data and behavior trajectory data. The prediction results include presence status, non-presence status or null, which means that the presence behavior of the target user in the target time period cannot be predicted.

[0052] In an optional implementation manner provided by this embodiment, after calculating the behavior similarity between the target user and the historical users according to the application access data and the behavior trajectory data, and determining at least one historical user as a similar user of the target user, the following operations are performed:

[0053] Predicting the presence behavior of the target user in the target time period;

[0054] Determining whether the prediction result of the target user's presence behavior in the target time period is empty;

[0055] If it is not empty, the prediction result is used as the predicted presence behavior of the target user in the target time period;

[0056] If it is empty, execute step S104.

[0057] In the process of predicting the presence behavior of the target user in the target time period, in order to improve the accuracy and reliability of the prediction results, the following three specific implementation methods of prediction are provided:

[0058] In a first optional implementation manner provided by this embodiment, during the process of predicting the presence behavior of the target user in the target time period, the following operations are performed:

[0059] Reading the target user's last access location information from the target user's application access data and behavior trajectory data, and calculating the distance between the access location in the last access location information and the target field;

[0060] A prediction result of the target user's presence behavior in the target time period is determined based on the calculated distance.

[0061] The last visited location information includes the last visited location information, the last two visited location information, and the like.

[0062] Specifically, in the process of determining the predicted result of the presence behavior of the target user in the target time period based on the calculated distance, if the calculated distance is lower than the preset distance threshold, it is determined that the predicted presence behavior of the similar user in the target time period is in the presence state; if the calculated distance is higher than the preset distance threshold, it is determined that the predicted presence behavior of the similar user in the target time period is in the non-presence state; if the predicted result cannot be obtained through calculation, it is determined that the predicted presence behavior of the similar user in the target time period is empty, and the operation of determining whether the predicted result obtained by the prediction is empty is continued.

[0063] For example, from the application access data and behavior trajectory data of user a, the last access location information read from the 9:00-10:00 time period includes an access location that is a supermarket near the target field of user a. The distance between the supermarket location and the target field location is calculated to be 10m. It is judged that the calculated distance (10m) is less than the preset distance threshold (20m), and the predicted presence behavior of user a in the 9:00-10:00 time period is determined to be in the presence state, and the prediction result is not empty.

[0064] In a second optional implementation manner provided by this embodiment, during the process of predicting the presence behavior of the target user in the target time period, the following operations are performed:

[0065] Acquire the historical access time and historical location information of the target user through the application access data and behavior trajectory data of the target user;

[0066] Based on the historical access time and the historical location information, count the effective time periods during which the target user is in the presence state, and calculate the effective frequency of the target user being in the presence state in each time period in the effective time period;

[0067] A prediction result of the presence behavior of the target user in the target time period is determined based on the effective frequency.

[0068] For example, the historical access time and historical location information of user b in the past month are obtained through the application access data and behavior trajectory data of user b, and the effective time periods of user b's presence are counted as 12:00-13:30, 19:00-22:00, and 23:00-8:30 the next day. The effective frequency of being in the presence state in the 12:00-13:30 time period is calculated to be 10, the effective frequency of being in the presence state in the 19:00-22:00 time period is calculated to be 50, and the effective frequency of being in the presence state in the 23:00-8:30 the next day is calculated to be 20. Through comparison, it is determined that the predicted presence behavior of user b in the 12:00-13:30 time period belongs to the non-presence state, and the prediction result is not empty.

[0069] It should be added that in addition to obtaining the target user's historical access time and historical location information through the target user's application access data and behavior trajectory data, one of the historical access time and historical location information can also be obtained through the target user's application access data and behavior trajectory data.

[0070] In a third optional implementation manner provided by this embodiment, during the process of predicting the presence behavior of the target user in the target time period, the following operations are performed:

[0071] Classifying the target users into corresponding group characteristic categories based on the user characteristics of the target users;

[0072] According to the presence behavior characteristics corresponding to the group characteristic category, a prediction result of the presence behavior of the target user in the target time period is determined.

[0073] For example, the user characteristics of user c include a 25-year-old female programmer. Based on the user characteristics of user c, user c is divided into the corresponding group characteristic category. According to the presence behavior characteristics corresponding to the group characteristic category, no prediction result of user c's presence behavior in the time period of 13:00-16:00 is obtained, indicating that the prediction result is empty.

[0074] In addition, in addition to the three above-mentioned implementation methods for predicting the predicted presence behavior of the target user in the target time period, other prediction methods may be used to obtain prediction results using more accurate prediction methods and improve the confidence of the prediction results.

[0075] It should be added that the process of predicting the target user's presence behavior in the target time period can also be performed before the behavior similarity between the target user and the sampled user is calculated based on the application access data and the behavior trajectory data, and at least one sampled user is determined to be a similar user to the target user. The method for predicting the user's presence behavior includes:

[0076] Predict the target user's presence behavior during the target time period;

[0077] Determine whether the prediction result of the target user's presence behavior in the target time period is empty;

[0078] If it is not empty, the prediction result is used as the predicted presence behavior of the target user in the target time period;

[0079] If it is empty, the behavior similarity between the target user and the sampled users is calculated based on the application access data and behavior trajectory data, and at least one sampled user is determined to be a similar user to the target user; the presence behavior of similar users in the target time period is predicted; similar users are weighted according to the behavior similarity, and the predicted scores of similar users whose presence behaviors belong to the same behavior category are calculated based on the assigned prediction weights; the predicted presence behavior operation of the target user in the target time period is determined based on the prediction score.

[0080] Step S104: predicting the presence behavior of the similar users in the target time period.

[0081] In the specific implementation, in the process of predicting the presence behavior of similar users in the target time period, the following three implementation methods are provided:

[0082] (1) Read the previous access location information of similar users from the application access data and behavior trajectory data of similar users, and calculate the distance between the access location in the previous access location information and the target field;

[0083] The prediction result of the presence behavior of similar users in the target time period is determined according to the calculated distance.

[0084] Specifically, if the calculated distance is lower than the preset distance threshold, it is determined that the predicted presence behavior of the similar user in the target time period is in the presence state, which means that the prediction result is not empty; if the calculated distance is higher than the preset distance threshold, it is determined that the predicted presence behavior of the similar user in the target time period is in the non-presence state, which means that the prediction result is not empty; if the prediction result cannot be obtained through calculation, it is determined that the predicted presence behavior of the similar user in the target time period is empty.

[0085] (2) Obtain historical access time and historical location information of similar users through application access data and behavior trajectory data of similar users;

[0086] Based on historical access time and historical location information, statistics are kept of effective time periods in which similar users are present, and effective frequencies of being present in each time period in the effective time period are calculated;

[0087] Determine the prediction results of the presence behavior of similar users in the target time period based on the effective frequency.

[0088] (3) Classifying similar users into corresponding group characteristic categories based on their user characteristics;

[0089] According to the presence behavior characteristics corresponding to the group feature categories, the prediction results of the presence behavior of similar users in the target time period are determined.

[0090] In actual applications, in the process of predicting the presence behavior of similar users in the target time period, in addition to the three implementation methods provided above, a prediction model can also be used to achieve high accuracy and high reliability of the prediction results. The specific operation process is as follows:

[0091] The model input data of similar users is extracted from the application access data and behavior trajectory data of similar users, and the model input data is input into the behavior prediction model to predict the presence behavior of similar users in the target time period, and the prediction results of the presence behavior of similar users are obtained; wherein the behavior prediction model is generated based on the training of positive sample user data and negative sample user data.

[0092] The model input data may be a standardized sequence obtained by converting application access data and behavior trajectory data, or a standardized sequence obtained by converting on-site behavior characteristics, or the constructed behavior feature vector of similar users may be directly input into the prediction model for prediction.

[0093] In addition, since the predicted results may be empty, in view of this, after predicting the presence behavior of similar users in the target time period, the predicted results of the presence behavior of similar users can be judged. If the predicted results are all empty, no processing is performed to improve data processing efficiency, avoid waste of data resources, reduce the execution steps of the prediction, and promote the flexible execution of the prediction process. In the case where the presence behavior of similar users in the target time period cannot be predicted (the predicted results are all empty), it is very likely that the presence behavior of target users with similarities to the similar users in the target time period cannot be predicted. By detecting whether the presence behavior of similar users is predictable, it is determined whether the predicted results are all empty.

[0094] In an optional implementation provided by this embodiment, after predicting the presence behavior of similar users in the target time period, the following operation is performed: determining whether the predicted results of the presence behavior of the similar users in the target time period are all empty; if so, determining that the predicted presence behavior of the target user in the target time period is empty; if not, executing step S106.

[0095] Step S106, weights are assigned to the similar users according to the behavior similarities, and prediction scores of similar users whose behaviors on-site belong to the same behavior category are calculated based on the assigned prediction weights.

[0096] The above judgment determines whether the prediction results of the presence behaviors of similar users are all empty. On this basis, if the prediction results are not empty, weights are assigned to similar users according to the behavior similarity, and the prediction scores of similar users whose presence behaviors belong to the same behavior category are calculated based on the assigned prediction weights.

[0097] In an optional implementation provided by this embodiment, in the process of assigning weights to similar users according to behavior similarity and calculating predicted scores of similar users whose behaviors on-site belong to the same behavior category according to the assigned prediction weights, the following operations are performed:

[0098] Allocating weights to the similar users according to the behavior similarities, and removing similar users whose prediction results are empty from the similar users;

[0099] According to the prediction weights, the presence scores of similar users whose behavior categories after elimination are in the presence state are calculated, and the absence scores of similar users whose behavior categories after elimination are in the absence state are calculated.

[0100] Optionally, the weight of the prediction weight decreases in sequence according to the sorting order, and the weight difference between two adjacent prediction weights is equal. By assigning weights to the sorted similar users in a decreasing order and with equal weight differences, and calculating the prediction scores of similar users in the presence state and the absence state respectively, the prediction effect is improved, and the presence behavior data of a large number of similar users is used to compare the presence behavior of the target user, so as to improve the data utilization and mine the potential data rules of massive data to improve the user experience.

[0101] For example, the top 10 similar users of user u are U-similarity = {u1, u2, ..., u10}, and the similar users are assigned to {u1, u2, ..., u10}->{10, 9, ..., 1} in reverse order of 10 according to the behavioral similarity, and the prediction results of the top 10 similar users are {presence, presence, presence, presence, presence, absence, absence, absence, absence, empty, empty} respectively. Two similar users with empty prediction results are eliminated, and the prediction results are expressed as {presence, presence, presence, presence, presence, absence, absence, absence, absence}. According to the assigned prediction weights, the presence score of the similar users in the presence state is calculated to be 10+9+8+7+6=40, and the absence score of the similar users in the absence state is calculated to be 5+4+3=12.

[0102] Step S108: determining the predicted presence behavior of the target user in the target time period based on the predicted score.

[0103] Based on the above-mentioned weight allocation to similar users according to behavior similarity, and the prediction scores of similar users whose presence behaviors belong to the same behavior category are calculated according to the allocated prediction weights, this embodiment determines the predicted presence behavior of the target user in the target time period based on the prediction scores. The prediction score includes the presence score and the non-presence score mentioned above, and the predicted presence behavior includes at least one of the following: presence state, non-presence state, and empty.

[0104] In specific implementation, in the process of determining the predicted presence behavior of the target user in the target time period based on the predicted score, the following operations are performed:

[0105] If the presence score is higher than the absence score, taking the presence status as the predicted presence behavior of the target user in the target time period;

[0106] If the presence score is lower than the absence score, the absence state is determined to be the predicted presence behavior of the target user in the target time period.

[0107] Continuing with the above example, according to the assigned prediction weights, the presence score of similar users in the presence state is calculated as 10+9+8+7+6=40, and the absence score of similar users in the absence state is calculated as 5+4+3=12. Since the presence score 40> the absence score 12, it is determined that the predicted presence behavior of user u in the 19:00-22:00 time period belongs to the presence state.

[0108] During the specific implementation process, in order to push service messages to the target users in a targeted manner based on the predicted presence behavior of the target users, reduce the failure rate of service message push, increase the probability of users triggering service efficiency, and thereby increase the user conversion rate and the operational efficiency of the service provider, service messages can be pushed to the target users based on the presence status as the predicted presence behavior of the target users in the target time period.

[0109] In an optional implementation manner provided by this embodiment, after the presence status is used as the predicted presence behavior of the target user in the target time period, the following operations are performed:

[0110] A service message is pushed to the target user in the target time period according to the service information of the service to be processed.

[0111] Continuing with the above example, if the predicted presence behavior of user u during the 19:00-22:00 time period is determined to be present, a service message push regarding real-name registration can be sent to user u during the 19:00-22:00 time period, so that user u can complete the task of uploading identity credentials and other document information by triggering the service message, thereby avoiding the invalidity of the service message push.

[0112] The following uses the application of a user presence behavior prediction method provided by this embodiment in a service push scenario as an example to further illustrate the user presence behavior prediction method provided by this embodiment. Figure 2 ,The user presence behavior prediction method applied to the service push scenario specifically includes the following steps.

[0113] Step S202: extracting sample users from the application access data and behavior trajectory data of historical users.

[0114] Step S204: constructing a behavior feature vector according to the application access data and behavior trajectory data of the target user, and constructing a behavior feature vector according to the application access data and behavior trajectory data of the sampled user.

[0115] In this process, positive sample users and negative sample users are extracted from the application access data and behavior trajectory data of the sampled users, and time feature elements and presence behavior elements are extracted through the application access data and behavior trajectory data of the positive sample users and the application access data and behavior trajectory data of the negative sample users. The behavior feature vector of the sampled users is constructed based on the time feature elements and presence behavior elements, and the time feature elements and presence behavior elements are extracted from the application access data and behavior trajectory data of the target users, and the behavior feature vector of the target user is constructed based on the time feature elements and presence behavior elements.

[0116] Step S206: Input the behavior feature vector of the target user and the behavior feature vector of the sampled user into a similarity algorithm to calculate the behavior similarity, and output the behavior similarity between the target user and the sampled user.

[0117] Step S208: Filter the sampled users whose geographical locations match the target user from the sampled users.

[0118] Step S210 , sorting the sampled users obtained by screening in descending order of behavior similarity, and determining the sampled users whose sorting positions are before the preset positions as similar users.

[0119] Step S212, reading the target user's last visited location information from the target user's application access data and behavior trajectory data, and calculating the distance between the access location in the last visited location information and the target field.

[0120] Step S214: determining a prediction result of the target user's presence behavior in the target time period based on the calculated distance.

[0121] Step S216, determining whether the prediction result of the target user's presence behavior in the target time period is unpredictable;

[0122] If not, the prediction result is used as the predicted presence behavior of the target user in the target time period;

[0123] If so, execute steps S218 to S222.

[0124] Step S218: classify similar users into corresponding group characteristic categories based on the user characteristics of similar users.

[0125] Step S220, determining the prediction result of the presence behavior of similar users in the target time period according to the presence behavior characteristics corresponding to the group characteristic categories.

[0126] Step S222, determining whether the prediction results of the presence behaviors of similar users are all unpredictable;

[0127] If so, it is determined that the predicted presence behavior of the target user in the target time period is unpredictable;

[0128] If not, execute steps S224 to S230.

[0129] Step S224, weights are assigned to similar users according to behavior similarity, and similar users whose prediction results are unpredictable are removed from the similar users.

[0130] Step S226, calculating the presence scores of similar users whose behavior categories after elimination are in the presence state, and calculating the absence scores of similar users whose behavior categories after elimination are in the absence state, according to the prediction weights.

[0131] Among them, the weights of the prediction weights decrease in descending order according to the sorting order, and the weight difference between two adjacent prediction weights is equal.

[0132] Step S228: If the presence score is higher than the absence score, the presence status is used as the predicted presence behavior of the target user in the target time period.

[0133] Step S230: Push a service message of quantified carbon saving index to the target user in the target time period according to the carbon saving service information.

[0134] In summary, the user presence behavior prediction method provided by the present embodiment first extracts sample users from the application access data and behavior trajectory data of historical users, constructs a behavior feature vector according to the application access data and behavior trajectory data of the target user, and constructs a behavior feature vector according to the application access data and behavior trajectory data of the sampled user, then inputs the behavior feature vector of the target user and the behavior feature vector of the sampled user into a similarity algorithm to calculate the behavior similarity, and outputs the behavior similarity between the target user and the sampled user, then selects sample users whose geographic locations match the target user from the sampled users, sorts the sampled users obtained by screening in descending order of behavior similarity, and determines the sampled users whose sorting positions are before a preset position as similar users;

[0135] Secondly, the target user's presence behavior in the target time period is predicted, and then it is determined whether the prediction result obtained by the prediction is empty. If it is not empty, the prediction result is used as the predicted presence behavior of the target user in the target time period. If it is empty, the presence behavior of similar users in the target time period is predicted, and then it is determined whether the prediction results of the presence behavior of similar users are all empty. If so, it is determined that the predicted presence behavior of the target user in the target time period is empty. If not, similar users are weighted according to the behavioral similarity, and similar users with empty prediction results are eliminated from the similar users. According to the prediction weight, the presence score of similar users whose behavior category is present after elimination is calculated, and the non-presence score of similar users whose behavior category is non-present after elimination is calculated;

[0136] Finally, if the presence score is higher than the absence score, the presence status is used as the predicted presence behavior of the target user in the target time period, and then the service message is pushed to the target user in the target time period based on the service information of the service to be processed. In this way, the accuracy and reliability of the prediction of the user's presence behavior are improved, the confidence and timeliness of the prediction results are improved, and targeted service information is pushed to the user, avoiding the ineffectiveness of service message push and the waste of data resources, while increasing the probability of users triggering service information, thereby improving the user's conversion efficiency.

[0137] An embodiment of a user presence behavior prediction device provided in this specification is as follows:

[0138] In the above-mentioned embodiment, a user presence behavior prediction method is provided, and correspondingly, a user presence behavior prediction device is also provided, which will be described below with reference to the accompanying drawings.

[0139] Reference Figure 3 , which shows a schematic diagram of a user presence behavior prediction device provided by this embodiment.

[0140] Since the device embodiment corresponds to the method embodiment, the description is relatively simple, and the relevant parts can refer to the corresponding description of the method embodiment provided above. The device embodiment described below is only illustrative.

[0141] This embodiment provides a user presence behavior prediction device, including:

[0142] A similarity calculation module 302 is configured to calculate the behavior similarity between the target user and the sampled users according to the application access data and the behavior trajectory data, and determine at least one sampled user as a similar user of the target user;

[0143] The presence behavior prediction module 304 is configured to predict the presence behavior of the similar users in the target time period;

[0144] The prediction score calculation module 306 is configured to assign weights to the similar users according to the behavior similarities, and calculate the prediction scores of the similar users whose behaviors on the scene belong to the same behavior category according to the assigned prediction weights;

[0145] The presence behavior determination module 308 is configured to determine the predicted presence behavior of the target user in the target time period based on the prediction score.

[0146] An embodiment of a user presence behavior prediction device provided in this specification is as follows:

[0147] Corresponding to the user presence behavior prediction method described above, based on the same technical concept, one or more embodiments of this specification also provide a user presence behavior prediction device, which is used to execute the user presence behavior prediction method provided above. Figure 4 A schematic diagram of the structure of a user presence behavior prediction device provided in one or more embodiments of this specification.

[0148] This embodiment provides a user presence behavior prediction device, including:

[0149] like Figure 4 As shown, the user presence behavior prediction device may have relatively large differences due to different configurations or performances, and may include one or more processors 401 and memory 402, and the memory 402 may store one or more storage applications or data. Among them, the memory 402 may be a short-term storage or a persistent storage. The application stored in the memory 402 may include one or more modules (not shown in the figure), and each module may include a series of computer executable instructions in the user presence behavior prediction device. Furthermore, the processor 401 may be configured to communicate with the memory 402 to execute a series of computer executable instructions in the memory 402 on the user presence behavior prediction device. The user presence behavior prediction device may also include one or more power supplies 403, one or more wired or wireless network interfaces 404, one or more input / output interfaces 405, one or more keyboards 406, etc.

[0150] In a specific embodiment, the user presence behavior prediction device includes a memory and one or more programs, wherein the one or more programs are stored in the memory, and the one or more programs may include one or more modules, and each module may include a series of computer executable instructions for the user presence behavior prediction device, and the one or more programs are configured to be executed by one or more processors, including computer executable instructions for performing the following:

[0151] Calculate the behavior similarity between the target user and the sampled users based on the application access data and the behavior trajectory data, and determine at least one sampled user as a similar user of the target user;

[0152] Predicting the presence behavior of the similar users in a target time period;

[0153] Assigning weights to the similar users according to the behavior similarities, and calculating predicted scores of similar users whose behaviors on-site belong to the same behavior category according to the assigned prediction weights;

[0154] A predicted presence behavior of the target user in the target time period is determined based on the predicted score.

[0155] An embodiment of a storage medium provided in this specification is as follows:

[0156] Corresponding to the user presence behavior prediction method described above, based on the same technical concept, one or more embodiments of this specification also provide a storage medium.

[0157] The storage medium provided in this embodiment is used to store computer executable instructions, and the computer executable instructions implement the following process when executed by a processor:

[0158] Calculate the behavior similarity between the target user and the sampled users based on the application access data and the behavior trajectory data, and determine at least one sampled user as a similar user of the target user;

[0159] Predicting the presence behavior of the similar users in a target time period;

[0160] Assigning weights to the similar users according to the behavior similarities, and calculating predicted scores of similar users whose behaviors on-site belong to the same behavior category according to the assigned prediction weights;

[0161] A predicted presence behavior of the target user in the target time period is determined based on the predicted score.

[0162] It should be noted that the embodiment of the storage medium in this specification and the embodiment of the user presence behavior prediction method in this specification are based on the same inventive concept, so the specific implementation of this embodiment can refer to the implementation of the aforementioned corresponding method, and the repeated parts will not be repeated.

[0163] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0164] In the 1930s, improvements to a technology could be clearly distinguished as hardware improvements (for example, improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the method flow). However, with the development of technology, many improvements to the method flow today can be regarded as direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that an improvement in a method flow cannot be implemented using a hardware entity module. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system on a PLD by programming it themselves, without having to ask a chip manufacturer to design and produce a dedicated integrated circuit chip. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages ​​and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.

[0165] The controller can be implemented in any appropriate manner, for example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (such as software or firmware) that can be executed by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of controllers include, but are not limited to, the following microcontrollers: ARC625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in a purely computer-readable program code manner, the controller can be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, this controller can be considered as a hardware component, and the devices included therein for implementing various functions can also be regarded as structures within the hardware component. Or even, the devices for implementing various functions can be regarded as both software modules for implementing the method and structures within the hardware component.

[0166] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0167] For the convenience of description, the above devices are described in terms of functions and are divided into various units. Of course, when implementing the embodiments of this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0168] It should be understood by those skilled in the art that one or more embodiments of this specification may be provided as a method, system or computer program product. Therefore, one or more embodiments of this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0169] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable user behavior prediction device to produce a machine, so that the instructions executed by the processor of the computer or other programmable user behavior prediction device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0170] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable user behavior prediction device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0171] These computer program instructions can also be loaded onto a computer or other programmable user behavior prediction device, so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0172] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0173] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0174] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0175] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0176] One or more embodiments of the present specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. One or more embodiments of the present specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0177] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0178] The above description is only an embodiment of this document and is not intended to limit this document. For those skilled in the art, this document may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this document should be included in the scope of the claims of this document.

Claims

1. A method for predicting user presence behavior, include: Calculate the behavior similarity between the target user and the sampled users based on the application access data and the behavior trajectory data, and determine at least one sampled user as a similar user of the target user; Predicting the presence behavior of the similar users in a target time period; Determining whether the prediction results of the presence behaviors of the similar users in the target time period are all empty; If not, weights are assigned to the similar users according to the behavior similarities, and prediction scores of similar users whose behaviors on-site belong to the same behavior category are calculated based on the assigned prediction weights; A predicted presence behavior of the target user in the target time period is determined based on the predicted score.

2. The method for predicting user presence behavior according to claim 1, wherein the behavior similarity between the target user and the sampled user is calculated based on the application access data and the behavior trajectory data. include: Extracting the sampled user from the application access data and behavior trajectory data of historical users; constructing a behavior feature vector according to the application access data and behavior trajectory data of the target user, and constructing a behavior feature vector according to the application access data and behavior trajectory data of the sampled user; The behavior feature vector of the target user and the behavior feature vector of the sampled user are input into a similarity algorithm to calculate the behavior similarity, and the behavior similarity between the target user and the sampled user is output.

3. The user presence behavior prediction method according to claim 2, wherein the behavior feature vector is constructed based on the application access data and behavior trajectory data of the sampled user. include: Extracting positive sample users and negative sample users from the application access data and behavior trajectory data of the sampled users; Extracting time feature elements and / or presence behavior elements through the application access data and behavior trajectory data of the positive sample user and the application access data and behavior trajectory data of the negative sample user; A behavior feature vector of the sampled user is constructed according to the time feature element and / or the presence behavior element.

4. The method for predicting user presence behavior according to claim 1, wherein determining at least one sampled user as a similar user to the target user, include: Selecting a sampled user that matches the geographic location of the target user from the sampled users; The sampled users obtained by screening are sorted in descending order of the behavior similarity, and the sampled users whose sorting positions are before the preset positions are determined as the similar users.

5. The method for predicting user presence behavior according to claim 1, after the step of calculating the behavior similarity between the target user and the sampled users based on the application access data and the behavior trajectory data, and determining at least one sampled user as a similar user of the target user is performed, and before the step of predicting the presence behavior of the similar users in the target time period is performed, further include: Predicting the presence behavior of the target user in the target time period; Determining whether the prediction result of the target user's presence behavior in the target time period is empty; If it is not empty, use the prediction result of the target user's presence behavior during the target time period as the predicted presence behavior of the target user during the target time period; If it is empty, perform the step of predicting the presence behavior of the similar user during the target time period.

6. The user presence behavior prediction method according to claim 5, wherein the predicting the presence behavior of the target user during the target time period comprises: Read the previous access location information of the target user from the application access data and behavior track data of the target user, and calculate the distance between the access location in the previous access location information and the target field; Determine the prediction result of the presence behavior of the target user during the target time period based on the calculated distance; Or, Obtain the historical access time and / or historical location information of the target user through the application access data and behavior track data of the target user; Statistically calculate the effective time periods when the target user is in the presence state based on the historical access time and / or the historical location information, and calculate the effective frequencies of being in the presence state in each time period during the effective time periods; Determine the prediction result of the presence behavior of the target user during the target time period based on the effective frequencies; Or, Classify the target user into the corresponding group feature category based on the user characteristics of the target user; Determine the prediction result of the presence behavior of the target user during the target time period according to the presence behavior characteristics corresponding to the group feature category.

7. The user presence behavior prediction method according to claim 1, if the execution result after performing the step of judging whether the prediction results of the presence behaviors of the similar users during the target time period are all empty is yes, perform the following operations: Determine that the predicted presence behavior of the target user during the target time period is empty.

8. The user presence behavior prediction method according to claim 7, wherein the weights of the prediction weights decrease in sequence according to the sorting order, and the weight differences between adjacent two prediction weights are equal.

9. The user presence behavior prediction method according to claim 7, wherein the similar users are weighted according to the behavior similarity, and the prediction scores of the similar users whose predicted presence behaviors belong to the same behavior category are calculated according to the assigned prediction weights, comprises: Weight the similar users according to the behavior similarity, and exclude the similar users with empty prediction results from the similar users; According to the prediction weights, calculate the presence scores of the similar users with the presence state as the behavior category after exclusion, and calculate the non-presence scores of the similar users with the non-presence state as the behavior category after exclusion.

10. The user presence behavior prediction method according to claim 9, wherein the predicted presence behavior of the target user during the target time period is determined based on the prediction scores, comprises: If the presence score is higher than the non-presence score, use the presence state as the predicted presence behavior of the target user during the target time period; If the presence score is lower than the absence score, the absence state is determined to be the predicted presence behavior of the target user in the target time period.

11. The method for predicting user presence behavior according to claim 10, after the sub-step of taking the presence status as the predicted presence behavior of the target user in the target time period is performed, include: A service message is pushed to the target user in the target time period according to the service information of the service to be processed.

12. The user presence behavior prediction method according to claim 11, wherein the presence status includes at least one of the following: the user is at a residence, the user is at a place or route where the application access frequency meets the access condition.

13. A user behavior prediction device, include: a similarity calculation module, configured to calculate the behavior similarity between the target user and the sampled users according to the application access data and the behavior trajectory data, and determine at least one sampled user as a similar user of the target user; A presence behavior prediction module, configured to predict the presence behavior of the similar users in a target time period; A prediction score calculation module is configured to determine whether the prediction results of the presence behaviors of the similar users in the target time period are all empty; If not, weights are assigned to the similar users according to the behavior similarities, and prediction scores of similar users whose behaviors on-site belong to the same behavior category are calculated based on the assigned prediction weights; The presence behavior determination module is configured to determine the predicted presence behavior of the target user in the target time period based on the prediction score.

14. A user behavior prediction device, include: processor; and a memory configured to store computer executable instructions that, when executed, cause the processor to: Calculate the behavior similarity between the target user and the sampled users based on the application access data and the behavior trajectory data, and determine at least one sampled user as a similar user of the target user; Predicting the presence behavior of the similar users in a target time period; Determining whether the prediction results of the presence behaviors of the similar users in the target time period are all empty; If not, weights are assigned to the similar users according to the behavior similarities, and prediction scores of similar users whose behaviors on-site belong to the same behavior category are calculated based on the assigned prediction weights; A predicted presence behavior of the target user in the target time period is determined based on the predicted score.

15. A storage medium for storing computer executable instructions, wherein the computer executable instructions, when executed by a processor, implement the following process: Calculate the behavior similarity between the target user and the sampled users based on the application access data and the behavior trajectory data, and determine at least one sampled user as a similar user of the target user; Predicting the presence behavior of the similar users in a target time period; Determining whether the prediction results of the presence behaviors of the similar users in the target time period are all empty; If not, weights are assigned to the similar users according to the behavior similarities, and prediction scores of similar users whose behaviors on-site belong to the same behavior category are calculated based on the assigned prediction weights; A predicted presence behavior of the target user in the target time period is determined based on the predicted score.

Images