Information recommendation method and system

Abstract

An embodiment of the present invention provides an information recommendation method and system, which relates to the field of activity operations. The method includes: determining a first related user whose feature similarity with the first user is higher than a first threshold based on a first user who participates in a first activity to be recommended; determining a second activity similar to the first activity, and determining a second related user whose feature similarity with the second user is higher than a second threshold and who has not participated in the first activity based on a second user who participates in the second activity; increasing the priority of the first activity corresponding to each related user; and recommending information about at least one activity corresponding to each related user according to the priority of at least one activity corresponding to each related user. The information of the first activity can be dynamically displayed to necessary users, thereby expanding the scope of recommended users and improving operational capabilities; displaying corresponding activities according to the priority level of each activity on the user interface meets the personalized needs of different users.

Description

Technical Field

[0001] The present invention relates to the field of information recommendation, and in particular to an information recommendation method and system. Background Art

[0002] Currently, in the Internet industry, the process of recommending activity information can create short-term user value and help products improve long-term value.

[0003] However, existing technologies for recommending activity information typically use methods such as setting time and pre-defined user groups to prominently display activity information on websites or software. This pre-defined user boundaries prevent dynamic display of activity information to the necessary users. Furthermore, when multiple activity information is available for the same user, manual calculation and adjustment of the display priority of each activity are required, which is time-consuming, labor-intensive, and unintelligent. This method of recommending activity information cannot effectively support the distribution of recommended activity information, nor can it be applied to various business scenarios. Summary of the Invention

[0004] The embodiments of the present invention provide an information recommendation method and system, which can solve the technical problems in the prior art of limited user range for active information recommendation and difficulty in adjusting user priority.

[0005] To achieve the above objectives, an embodiment of the present invention provides an information recommendation method, comprising:

[0006] Determining, based on a first user who has participated in a first activity to be recommended, a first related user whose feature similarity with the first user is greater than a first threshold;

[0007] Determining a second activity similar to the first activity, and determining, based on a second user who participated in the second activity, a second related user whose feature similarity with the second user exceeds a second threshold and who has not participated in the first activity; wherein the features include user behavioral features with respect to the first activity and / or the second activity;

[0008] For the first related user and the second related user, raising the priority of the first activity corresponding to each related user; the priority indicates the level at which the current activity is displayed to the corresponding user;

[0009] According to the priority of at least one activity corresponding to each of the first related user and the second related user, information of the at least one activity corresponding to each of the first related user and the second related user is recommended to each related user; wherein the at least one activity includes the first activity.

[0010] In another aspect, an embodiment of the present invention provides an information recommendation system, including:

[0011] a related user determination unit configured to determine, based on a first user who participated in a first activity to be recommended, a first related user whose feature similarity with the first user exceeds a first threshold; determine a second activity similar to the first activity, and, based on a second user who participated in the second activity, determine a second related user whose feature similarity with the second user exceeds a second threshold and who has not participated in the first activity; wherein the features include user behavioral characteristics with respect to the first activity and / or the second activity;

[0012] a priority adjustment unit, configured to increase the priority of the first activity corresponding to each of the first related user and the second related user; the priority indicates a level at which the current activity is displayed to the corresponding user;

[0013] A recommendation unit is configured to recommend information about the at least one activity corresponding to each of the first related user and the second related user according to a priority of the at least one activity corresponding to each of the first related user and the second related user; wherein the at least one activity includes the first activity.

[0014] The above technical solution has the following beneficial effects: it identifies relevant users based on the specific circumstances of a specific first activity, enabling dynamic display of information about the first activity to necessary users; it expands the scope of recommended users and improves operational capabilities; and it avoids the limitations of recommending the first activity by prominently displaying information about the activity in a website or software by setting a time or pre-defined user group. Furthermore, when a user is identified as a relevant user, the corresponding display priority is directly determined, and the corresponding activities are displayed on the user interface according to their priority level, avoiding the time-consuming and laborious problem of manually calculating and adjusting the display priority of each activity. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0016] Figure 1 This is a flow chart of an information recommendation method according to an embodiment of the present invention;

[0017] Figure 2 This is a schematic diagram of the structure of an information recommendation system according to an embodiment of the present invention;

[0018] Figure 3 is a structural diagram of another information recommendation system according to an embodiment of the present invention;

[0019] Figure 4 is an example of how similar neighbors are calculated. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0021] like Figure 1 As shown, in combination with an embodiment of the present invention, an information recommendation method is provided, comprising:

[0022] S101: Determine, based on a first user who has participated in a first activity to be recommended, a first related user whose feature similarity with the first user is higher than a first threshold;

[0023] S102: Determine a second activity similar to the first activity, and based on a second user who participated in the second activity, determine a second related user whose feature similarity with the second user exceeds a second threshold and who has not participated in the first activity; wherein the feature includes a user's behavioral features with respect to the first activity and / or the second activity;

[0024] S103: For the first related user and the second related user, increase the priority of the first activity corresponding to each related user; the priority indicates the level at which the current activity is displayed to the corresponding user;

[0025] S104: Recommending information about the at least one activity corresponding to each of the first related user and the second related user to each related user according to the priority of the at least one activity; wherein the at least one activity includes the first activity.

[0026] By determining the relevant users based on the specific circumstances of the specific first activity, the information of the first activity can be dynamically displayed to the necessary users; this avoids the limitation of recommending the first activity by guiding the information of the activity in a prominent position on the website or software by setting the time, defining the user group in advance, etc. Moreover, when the user is determined to be a relevant user, the corresponding display priority is directly determined, and the corresponding activities are displayed on the user interface according to the priority level of each activity, avoiding the time-consuming and laborious problem of manually calculating and adjusting the display priority of each activity. Therefore, the method for recommending and issuing activity information in the embodiment of the present invention can be applied to various business scenarios and is easy to apply.

[0027] Preferably, S101: determining, based on a first user who has participated in a first activity to be recommended, a first related user whose feature similarity with the first user is higher than a first threshold, specifically includes:

[0028] S1011: Users who participate in the first activity within a first time period adjacent to the current time are regarded as first users, and similarity is calculated between the user behavior of each first user and the user behavior of each first user to be recommended in the feature library to obtain the feature similarity corresponding to each first user to be recommended, and the first users to be recommended whose feature similarity is higher than a first threshold are regarded as first related users. The first users to be recommended are users who participate in activities other than the first activity.

[0029] Users with similar behaviors are more likely to like the same activity. Therefore, the similarity of user behaviors is fully utilized, and users who participated in the first activity in the first time period adjacent to the current time are regarded as the first users. Based on the first users who participated in the first activity to be recommended, the behavior of the first users is fully utilized, and the necessary users can be matched, so that more accurate necessary users can be found.

[0030] Preferably, in S1011, the step of calculating the similarity between the user behavior of each first user and the user behavior of each first user to be recommended in the feature library to obtain the feature similarity corresponding to each first user to be recommended specifically includes:

[0031] S10111-1: Mapping the user behavior of each first user in participating in each activity into a point in an N-dimensional Euclidean space;

[0032] S10111-2: Mapping the user behavior of each first to-be-recommended user in participating in each activity into a point in an N-dimensional Euclidean space;

[0033] S10111-3: Calculate the Euclidean distance between each point of the first user in the N-dimensional Euclidean space and each point of the first to-be-recommended user in the N-dimensional Euclidean space to obtain the Euclidean distance between the first user and the first to-be-recommended user;

[0034] S10111-4: Determine the feature similarity corresponding to each of the first to-be-recommended users based on the Euclidean distance; the feature similarity refers to the reciprocal of the sum of 1 and the Euclidean distance.

[0035] The dimensions of the N-dimensional Euclidean space are equal to the total number of activities, with a one-to-one correspondence between all dimensions and all activities. The coordinate value in each dimension is the sum of the behavior coefficients of the corresponding user participating in the activity corresponding to the dimension, and N is greater than or equal to 2. The Euclidean distance in two-dimensional space is the true distance between two points.

[0036] Calculate the Euclidean distance between each of the first users in the N-dimensional Euclidean space and the points of each of the first users to be recommended in the N-dimensional Euclidean space to obtain the Euclidean distance between the first user and the first users to be recommended. The smaller the Euclidean distance, the greater the similarity. The Euclidean distance can be used to accurately determine the feature similarity corresponding to each of the first users to be recommended.

[0037] Preferably, S102: determining, based on the second user who participated in the second activity, a second related user whose feature similarity with the second user is higher than a second threshold and who has not participated in the first activity, specifically includes:

[0038] S1021: Users who participate in the second activity within a second time period adjacent to the current time are regarded as second users, and similarity is calculated between the user behavior of each second user and the user behavior of each second user to be recommended in the feature library to obtain the feature similarity corresponding to each second user to be recommended, and the second users to be recommended whose feature similarity is higher than a second threshold and who have not participated in the first activity, as well as the second users who have not participated in the first activity, are regarded as second related users. The second users to be recommended are users who participate in activities other than the second activity.

[0039] By making full use of the similarity of activities, users who participated in the second activity within a second time period adjacent to the current time are regarded as second users, and second related users are determined based on the second users. The second related users are more likely to like the first activity, and more accurate necessary users can be found.

[0040] Preferably, in S1021, the step of calculating the similarity between the user behavior of each second user and the user behavior of each second user to be recommended in the feature library to obtain the feature similarity corresponding to each second user to be recommended specifically includes:

[0041] S1021-1: Mapping the user behavior of each second user participating in each activity into a point in an N-dimensional Euclidean space;

[0042] S1021-2: Mapping the user behavior of each second to-be-recommended user in participating in each activity into a point in an N-dimensional Euclidean space;

[0043] S1021-3: Calculate the Euclidean distance between each second user in the N-dimensional Euclidean space and each second user to be recommended in the N-dimensional Euclidean space to obtain the Euclidean distance between the second user and the second user to be recommended;

[0044] S1021-4: Determine the feature similarity corresponding to each of the second users to be recommended based on the Euclidean distance; the feature similarity refers to the reciprocal of the sum of 1 and the Euclidean distance.

[0045] The dimensions of the N-dimensional Euclidean space are equal to the total number of activities, with a one-to-one correspondence between all dimensions and all activities. The coordinate value in each dimension is the sum of the behavior coefficients of the corresponding user participating in the activity corresponding to the dimension, and N is greater than or equal to 2. The Euclidean distance in two-dimensional space is the true distance between two points.

[0046] Calculate the points of each second user in the N-dimensional Euclidean space and the points of each second user to be recommended in the N-dimensional Euclidean space to obtain the Euclidean distance between the second user and the second user to be recommended; the smaller the Euclidean distance, the greater the similarity. The Euclidean distance can accurately determine the feature similarity corresponding to each first user to be recommended.

[0047] Preferably, S103: for the first related user and the second related user, raising the priority of the first activity corresponding to each related user specifically includes:

[0048] S1031: When the first relevant user is determined, the first priority cumulative value corresponding to the first relevant user is determined; the first priority cumulative value and the priority initial value corresponding to the first activity are added to obtain the priority value of the first activity corresponding to the first relevant user; wherein, before the first activity is issued, the initial priority and the corresponding priority initial value are configured for the first activity; when the user is determined as a relevant user, the corresponding display priority is directly determined, and the corresponding activities are displayed on the user interface according to the priority level of each activity, thereby avoiding the time-consuming and labor-intensive problem of manually calculating and adjusting the display priority of each activity.

[0049] S1032: When the second relevant user is determined, determine the second priority cumulative value corresponding to the second relevant user; add the second priority cumulative value to the initial priority value to obtain the priority value of the first activity corresponding to the second relevant user; wherein, the first priority cumulative value and the second priority cumulative value are updated regularly; directly determine the corresponding display priority when the user is determined as a relevant user, and display the corresponding activities according to the priority level of each activity on the user's interface, thereby avoiding the time-consuming and labor-intensive problem of manually calculating and adjusting the display priority of each activity.

[0050] S1034: When a target user is simultaneously determined as a related user by multiple first activities, a target first activity is configured for the target user from the multiple first activities, and the configured target first activity is displayed to the target user with the highest priority, and the configured target first activity does not participate in the priority adjustment.

[0051] Preferably, the information recommendation method further includes:

[0052] S105: Obtain the user behavior of each user for all activities, and store the obtained user behavior in a distributed publish-subscribe queue; asynchronous consumption avoids congestion.

[0053] S106: Asynchronously obtain the user behavior from the distributed publish-subscribe queue, store the user behavior in a feature library, and reserve the user behavior for activity recommendation.

[0054] S107: Based on the user as a unit, the corresponding behavior label and corresponding behavior coefficient are marked for the user behavior, and the behavior coefficient is a normalized value. The data of each behavior are unified in the same value range. For a certain activity that a certain user is interested in, the values of each behavior are weighted and summed, so that the overall preference obtained by the weighted sum is more accurate.

[0055] S107: The user behavior is labeled with a corresponding behavior label and a corresponding behavior coefficient, wherein the behavior coefficient is a normalized value, specifically including:

[0056] S1071: Processing the behavior coefficient, so that the processed behavior coefficient is within a preset value range of the behavior coefficient;

[0057] S1072: The behavior coefficient is normalized by dividing it by the maximum value of the behavior coefficient within the behavior category to obtain the user's behavior normalization value. The data of each behavior is unified in the same value range. For a certain activity that a user is interested in, the values of each behavior are weighted and summed, so that the overall preference obtained by the weighted sum is more accurate.

[0058] Preferably, the information recommendation method further includes:

[0059] S108: Before the first activity is issued, the data of the first activity is stored in the database, and the first-level cache and the second-level cache are set at the same time; that is, when configuring a recommendation information, the data of the recommendation information is stored in the Mysql database, and the first-level cache (i.e., memory) and the second-level cache are also set. When data is added or modified, the second-level cache is also updated at the same time.

[0060] S109: When the user retrieves data for the first activity, the user first accesses the first-level cache. If the first-level cache does not contain the data for the first activity, the user accesses the second-level cache. If the second-level cache does not contain the data for the first activity, the user accesses the database. This prevents a sudden increase in pressure on the MySQL database, leading to a possible crash, due to cache failure after the first-level cache fails.

[0061] like Figure 2 As shown, in combination with an embodiment of the present invention, an information recommendation system is provided, including:

[0062] The related user determination unit 21 is configured to determine, based on a first user who participated in a first activity to be recommended, a first related user whose feature similarity with the first user exceeds a first threshold; determine a second activity similar to the first activity, and, based on a second user who participated in the second activity, determine a second related user whose feature similarity with the second user exceeds a second threshold and who has not participated in the first activity; wherein the features include user behavioral characteristics with respect to the first activity and / or the second activity;

[0063] The priority adjustment unit 22 is configured to increase the priority of the first activity corresponding to each of the first related user and the second related user; the priority indicates the level at which the current activity is displayed to the corresponding user;

[0064] The recommendation unit 23 is configured to recommend information about the at least one activity corresponding to each of the first related user and the second related user according to a priority of the at least one activity respectively corresponding to the first related user and the second related user; wherein the at least one activity includes the first activity.

[0065] Preferably, the related user determination unit 21 includes:

[0066] The first related user determination subunit is used to select users who participated in the first activity within a first time period adjacent to the current time as first users, calculate the similarity between the user behavior of each first user and the user behavior of each first user to be recommended in the feature library, obtain the feature similarity corresponding to each first user to be recommended, and select the first users to be recommended whose feature similarity is higher than a first threshold as first related users. The first users to be recommended are users who participated in activities other than the first activity.

[0067] Preferably, the related user determination unit 21 includes:

[0068] The second related user determination subunit is used to select users who participated in the second activity within a second time period adjacent to the current time as second users, calculate the similarity between the user behavior of each second user and the user behavior of each second user to be recommended in the feature library, obtain the feature similarity corresponding to each second user to be recommended, and select the second users to be recommended whose feature similarity is higher than a second threshold and who have not participated in the first activity, and the second users who have not participated in the first activity as second related users. The second users to be recommended are users who participated in activities other than the second activity.

[0069] Preferably, the priority adjustment unit is specifically configured to:

[0070] When the first relevant user is determined, determining a first priority accumulated value corresponding to the first relevant user; adding the first priority accumulated value to the initial priority value corresponding to the first activity to obtain a priority value of the first activity corresponding to the first relevant user; wherein, before the first activity is issued, configuring an initial priority and a corresponding initial priority value for the first activity;

[0071] When the second relevant user is determined, determining a second priority accumulated value corresponding to the second relevant user; adding the second priority accumulated value to the initial priority value to obtain a priority value of the first activity corresponding to the second relevant user; wherein the first priority accumulated value and the second priority accumulated value are regularly updated;

[0072] When a target user is simultaneously determined as a relevant user by multiple first activities, a target first activity is configured for the target user from the multiple first activities, and the configured target first activity is displayed to the target user with the highest priority, and the configured target first activity does not participate in the priority adjustment.

[0073] Preferably, the first relevant user determination subunit is specifically configured to:

[0074] Mapping the user behavior of each of the first users in participating in each activity into a point in an N-dimensional Euclidean space;

[0075] Mapping the user behavior of each first to-be-recommended user in participating in each activity into a point in an N-dimensional Euclidean space;

[0076] Calculate the Euclidean distance between each of the first users in the N-dimensional Euclidean space and each of the first users to be recommended in the N-dimensional Euclidean space, respectively;

[0077] Determining a feature similarity corresponding to each of the first to-be-recommended users based on the Euclidean distance;

[0078] The dimension of the N-dimensional Euclidean space is the same as the total number of activities, all dimensions have a one-to-one correspondence with all activities, the coordinate value on each dimension is the sum of the behavior coefficients of the corresponding users participating in the activities corresponding to the dimension, and N is greater than or equal to 2.

[0079] Preferably, the second related user determination subunit is specifically configured to:

[0080] Mapping the user behavior of each second user participating in each activity into a point in an N-dimensional Euclidean space;

[0081] Mapping the user behavior of each second to-be-recommended user in participating in each activity into a point in an N-dimensional Euclidean space;

[0082] Calculate the Euclidean distance between each second user and the second user to be recommended by calculating the Euclidean distance between each second user and the second user to be recommended;

[0083] Determine the feature similarity corresponding to each of the second to-be-recommended users based on the Euclidean distance;

[0084] The dimension of the N-dimensional Euclidean space is the same as the total number of activities, all dimensions have a one-to-one correspondence with all activities, the coordinate value on each dimension is the sum of the behavior coefficients of the corresponding users participating in the activities corresponding to the dimension, and N is greater than or equal to 2.

[0085] Preferably, the activity operation system further includes:

[0086] A user behavior acquisition unit is used to acquire each user's user behavior for all activities and store the acquired user behavior in a distributed publish-subscribe queue;

[0087] Asynchronously obtaining the user behavior from the distributed publish-subscribe queue and storing the user behavior in a feature library;

[0088] The behavior value marking unit is used to mark the user behavior with a corresponding behavior label and a corresponding behavior coefficient based on the user, and the behavior coefficient is a normalized value.

[0089] The behavior value marking unit is specifically used to process the behavior coefficient so that the processed behavior coefficient is within the preset value range of the behavior coefficient; the behavior coefficient is normalized by dividing the behavior coefficient by the maximum value of the behavior coefficient in the behavior category to obtain the user's behavior normalized value.

[0090] Preferably, the information recommendation system further includes:

[0091] A storage setting unit is used to store the data of the first activity in a database before the first activity is issued, and to set a first-level cache and a second-level cache; when a user obtains the data of the first activity, the first-level cache is first accessed; if the data of the first activity is not obtained from the first-level cache, the second-level cache is accessed; if the data of the first activity is not obtained from the second-level cache, the database is accessed.

[0092] The above technical solutions of the embodiments of the present invention are described in detail below with reference to specific application examples. For technical details not introduced during the implementation process, please refer to the relevant description above.

[0093] An embodiment of the present invention is an information recommendation method that can configure multiple recommendation information. Different activities can correspond to different priorities. The activity priority can also be dynamically adjusted according to user behavior, thereby guiding users to receive the display of corresponding activities and participate in them.

[0094] The system architecture of information recommendation is as follows Figure 3 As shown, the gateway layer of the overall system architecture includes the Gateway for authentication, flow control, circuit breaking and degradation; the application layer calls the server API interface to return various information corresponding to the recommended activities; user data support is configured through the background and returned according to the agreed data standards.

[0095] By tracking user behavior at the application layer, the client calls the server through the API to report the data. After receiving the client request, the server stores the user behavior data in the Kafka queue. The data in the partition of the Broker is stored in HBase for consumption by consumers, and then cleans and calculates the data through Flink.

[0096] The application is located in the application layer and is a service used to run the activity recommendation system. When the client requests data from the server, the server code program needs to be matched with the operating environment to execute. The application is deployed in a Pod in K8s using Docker containerization technology. When the number of user requests surges, K8s can dynamically scale to ensure normal service availability and prevent service unavailability caused by excessive load. When a user request for a recommended activity comes from the application layer, permission verification and flow control are performed through the gateway layer. Flow control means that if there is an increase in user access during peak hours, the pressure of user requests will increase. If the server program cannot meet all requests and return them normally, in order to ensure that at least some users are normal, some requests will be directly returned through circuit breaking and degradation strategies, without participating in the program's logical calculations, thereby reducing service pressure.

[0097] For example, when a user requests the list page for the first activity, the client calls the server interface and first establishes network communication, including DNS resolution and the HTTP three-way handshake. After receiving the user request, the server needs to verify whether it is a genuine user request to prevent hackers from simulating users and illegally accessing real data. After authentication is successful, the server program begins processing the user request. If it is a genuine user request, it forwards the request to the real backend server through Nginx. After the server obtains the data from the first-level cache, it outputs it in accordance with the data standard. At the same time, to facilitate subsequent troubleshooting, the entire link process of a request is marked through trace, and the standard output of the POD is sent to ClickHouse for viewing through Granfana.

[0098] In the information recommendation method of the embodiment of the present invention, multiple activities are displayed on each user interface, and the activities displayed for different users may be different; and when multiple activities are displayed on a certain user interface, each activity has a different priority, and the priority can be dynamically adjusted, that is, which activity is displayed first. The details are as follows:

[0099] 1. Obtaining User Behavior Data

[0100] 1. Obtain each user's behavior across all activities and store it in a distributed publish-subscribe queue. For example, this can be achieved by embedding data at the application layer. When a user accesses an information page on a website or mobile device, the user's clicked topic ID and browsing time are recorded. When a user searches for a topic of interest, this behavior is also recorded and reported via front-end click events (i.e., client, H5, and mini-programs). User behavior supports subsequent actions. The front-end calls the server's reporting interface through an API. After receiving the front-end request, the server stores the user behavior in a Kafka queue.

[0101] 2. Asynchronously obtain user behaviors from the distributed publish-subscribe queue and store them in the feature library. That is, obtain user behaviors from the Kafka queue and store a large amount of user behaviors in the feature library.

[0102] 3. Data processing: Taking users as units, mark user behaviors with corresponding behavior labels and corresponding behavior coefficients. The behavior coefficients are normalized values.

[0103] By defining behavioral tags in the backend, we assign relevant behavior tags to each user, providing clues for subsequent activities. By unifying the data for each behavior within the same value range, we perform a weighted summation of the values for each behavior for a specific activity that a user is interested in, thereby making the overall preference derived from this weighted sum more accurate. Weighting can be understood as which behavior best represents the user's intentions.

[0104] The behavior coefficient is processed so that the processed behavior coefficient is within the preset value range of the behavior coefficient. The behavior coefficient is normalized by dividing it by the maximum value of the behavior coefficient in the behavior category to obtain the user's behavior normalized value, so as to ensure that the normalized data value is in the range of [0,1]. The normalized data value will be used when configuring the first activity to be recommended later, and will be calculated based on this behavior coefficient and the association between behaviors.

[0105] 3. Determine the similarity between the target user and other users based on the target user’s behavior

[0106] (1) Before the first activity is issued, the data of the first activity is stored in the database, and the first-level cache and the second-level cache are set at the same time. That is, when configuring a recommendation information, the data of the recommendation information is stored in the Mysql database, and the first-level cache (i.e., memory) and the second-level cache are also set. The second-level cache is also updated when data is added or modified. When the user obtains the data of the first activity, the first-level cache is accessed; if the data of the first activity is not obtained from the first-level cache, the second-level cache is accessed; if the data of the first activity is not obtained from the second-level cache, the database is accessed. This prevents the Mysql database from being overwhelmed by pressure due to cache failure after the first-level cache fails, and even paralysis.

[0107] (2) Determining relevant users of the first activity

[0108] 1. Based on a first user who participates in a first activity to be recommended, determining a first related user whose feature similarity with the first user is higher than a first threshold; specifically comprising: taking users who participate in the first activity within a first time period adjacent to the current time as first users, performing similarity calculations on the user behavior of each first user with the user behavior of each first user to be recommended in the feature library to obtain a feature similarity corresponding to each first user to be recommended, and taking the first users to be recommended whose feature similarity is higher than the first threshold as first related users, where the first users to be recommended are users who participate in activities other than the first activity.

[0109] The calculating the similarity between the user behavior of each first user and the user behavior of each first user to be recommended in the feature library to obtain the feature similarity corresponding to each first user to be recommended specifically includes:

[0110] Map the user behavior of each of the first users in participating in each activity into a point in the N-dimensional Euclidean space; map the user behavior of each of the first users to be recommended in participating in each activity into a point in the N-dimensional Euclidean space; calculate the point of each of the first users in the N-dimensional Euclidean space with the point of each of the first users to be recommended in the N-dimensional Euclidean space to obtain the Euclidean distance between the first user and the first users to be recommended; determine the feature similarity corresponding to each of the first users to be recommended based on the Euclidean distance; wherein the dimension of the N-dimensional Euclidean space is the same as the total number of activities, all dimensions have a one-to-one correspondence with all activities, the coordinate value on each dimension is the sum of the behavior coefficients of the corresponding users participating in the activities corresponding to the dimension, and N is greater than or equal to 2.

[0111] In the embodiment of the present invention, feature similarity calculation is performed based on the vector using the Euclidean distance algorithm as follows: Assume that x and y are two points in an n-dimensional space, and the Euclidean distance between them is: When n=2, the Euclidean distance is the distance between two points on the plane. Feature similarity is expressed as: When expressing feature similarity based on Euclidean distance, the smaller the distance, the greater the similarity. The Euclidean distance in two-dimensional space is the actual distance between two points. Mapping the user behavior dataset into a coordinate system can clearly compare the feature similarity between users. Among them, when determining the relevant users for the i-th activity to be recommended, x i and y i They represent the coordinates of user x and user y respectively. The distance between the two coordinates is the Euclidean distance. The feature similarity is the reciprocal of the sum of 1 and the Euclidean distance.

[0112] An example of feature similarity calculation is as follows: For activity A and activity B, the corresponding behavior weights are pre-set: a click coefficient of 0.5, a search coefficient of 0.8, a comment coefficient of 0.3, a like coefficient of 0.3, a forwarding coefficient of 0.5, a participation coefficient of 1, a browsing time of less than 10 minutes of 0.2, a browsing time of more than 10 minutes but less than 20 minutes of 0.4, and a browsing time of more than 20 minutes of 0.6. These coefficients and weights can be customized according to specific circumstances.

[0113] Suppose user A clicks on Topic A and browses a blog post under that topic for 5 minutes, likes and comments on the post, but does not browse Topic B. User B searches for Topic B and browses a blog post under Topic B for 1 minute, and also searches for Topic A and comments on one of the posts.

[0114] Let topic A be the vertical coordinate of the Euclidean space, and topic B be the horizontal coordinate of the Euclidean space. The coordinate of user a is that the participation in topic B is 0, then the horizontal coordinate is 0. The participation in topic A includes click behavior 0.5, browsing behavior within 10 minutes 0.2, like behavior 0.3, and comment behavior 0.3. Then the coefficient for topic A is 0.5+0.2+0.3+0.3=1.3, so the coordinate for user a is (0, 1.3). Similarly, user b's ordinate is 0.8 + 0.2 + 0.2 = 1.2, and its abscissa is 0.8 + 0.3 = 1.1. B's coordinates are (1.1, 1.2). Using the Euclidean distance formula, we calculate the distance between points a and b to be 1.10, which is the behavior coefficient. The behavior coefficient yields feature similarity, which is the inverse of 1 + 1.10, 0.47. Based on similar activities, we conclude that when feature similarity is greater than 0.5, users are considered highly similar. Table 1 lists some feature similarities between users.

[0115] Table 1 Example of Euclidean distance between users

[0116] User Relationship Behavior coefficient Similarity User a&f 4.61 0.18 User a&c 0.38 0.72 User a&d 0.69 0.59 User a&e 3.67 0.21 User b&c 4.41 0.18

[0117] From Table 1, we can see that the distance between users a and c is relatively close, and the distance between users a and d is relatively close.

[0118] The specific operation of the recommendation method is as follows: Because the user's behavior data is obtained in the early stage, when configuring a new activity, the corresponding behavior coefficient and weight are set for the new activity, and the activity is recommended to neighbors with the same preferences. Figure 4 As shown in , it is the calculation of similar neighbors. Neighbors are divided into two categories: 1. Get a fixed number of neighbors, the distance is not agreed upon, and only take the nearest K, such as Figure 42. Get the neighbors of similarity. We take the current point as the center and all points in the area with a distance of K as neighbors of the current point, as shown in Figure 4 As shown in Part B.

[0119] The embodiment of the present invention does not adopt Figure 4 Method A, because when there are not enough similar points nearby, it is forced to take some less similar neighbors, which affects the similarity. Figure 4 In method B, although the number of neighbors calculated is uncertain, the feature similarity will not have a large error.

[0120] For a cash activity (hereinafter referred to as A) in which the first activity is a participation topic, the characteristics of users who have participated in the first activity by finding the same neighbors (first related users) among users who have participated in the activity are as follows: users who have participated in the first activity within a first time period adjacent to the current time, such as the last 7 days, are regarded as the first users. The first user is taken as the user range, and the Euclidean distance algorithm is used to calculate the feature similarity between the first user to be recommended in the feature library and the first user. The first user to be recommended with a similarity higher than a first threshold, such as 0.5, is determined as the first related user, and at the same time, the display priority of activity A for the first related user is set to be enhanced.

[0121] 2. Determine a second activity similar to the first activity, and based on a second user who participates in the second activity, determine a second related user whose feature similarity with the second user is higher than a second threshold and who has not participated in the first activity; wherein the feature includes the user's behavioral characteristics with respect to the first activity and / or the second activity; specifically comprising: treating users who participate in the second activity within a second time period adjacent to the current time as second users, performing similarity calculations on the user behavior of each second user with the user behavior of each second user to be recommended in the feature library to obtain the feature similarity corresponding to each second user to be recommended, treating the second user to be recommended whose feature similarity is higher than the second threshold and who has not participated in the first activity, and the second user who has not participated in the first activity as second related users, and the second user to be recommended is a user who participates in activities other than the second activity.

[0122] Calculate the similarity between the user behavior of each second user and the user behavior of each second user to be recommended in the feature library to obtain the feature similarity corresponding to each second user to be recommended, specifically including: mapping the user behavior of each second user participating in each activity to a point in the N-dimensional Euclidean space; mapping the user behavior of each second user to be recommended in each activity to a point in the N-dimensional Euclidean space; calculating the point of each second user in the N-dimensional Euclidean space with the point of each second user to be recommended in the N-dimensional Euclidean space to obtain the Euclidean distance between the second user and the second user to be recommended; determine the feature similarity corresponding to each second user to be recommended based on the Euclidean distance; wherein the dimension of the N-dimensional Euclidean space is the same as the total number of activities, all dimensions have a one-to-one correspondence with all activities, the coordinate value on each dimension is the sum of the behavior coefficients of the corresponding user participating in the activity corresponding to the dimension, and N is greater than or equal to 2.

[0123] For a cash activity A in which the first activity is a topic of participation, users who have participated in a similar second activity, such as a prize-winning activity for answering questions (hereinafter referred to as B), find the same neighbors (second related users).

[0124] The characteristics of users who participated in the second activity are within a second time period adjacent to the current time, such as a 15-day time range. User C, who participated in Activity B, is selected as the second user. Based on this user C, the Euclidean distance algorithm is used to calculate the feature similarity of the second recommended users in the feature library. For user C whose feature similarity exceeds a second threshold, such as 0.7, and who did not participate in the first activity A, Activity A is prioritized when displayed. The second user is also selected as the second related user. If a second user did not participate in the first activity, the priority of this second user is set to the initial priority.

[0125] 3. For the first related user and the second related user, increase the priority of the first activity corresponding to each related user; the priority indicates the level at which the current activity is displayed to the corresponding user; specifically, including:

[0126] When the first relevant user is determined, determining a first priority accumulated value corresponding to the first relevant user; adding the first priority accumulated value to the initial priority value corresponding to the first activity to obtain a priority value of the first activity corresponding to the first relevant user; wherein, before the first activity is issued, configuring an initial priority and a corresponding initial priority value for the first activity;

[0127] When the second relevant user is determined, determining a second priority accumulated value corresponding to the second relevant user; adding the second priority accumulated value to the initial priority value to obtain a priority value of the first activity corresponding to the second relevant user;

[0128] When a target user is simultaneously determined as a relevant user by multiple first activities, a target first activity is configured for the target user from the multiple first activities, and the configured target first activity is displayed to the target user with the highest priority, and the configured target first activity does not participate in the priority adjustment.

[0129] 4. Recommend information about the at least one activity corresponding to each of the first related user and the second related user to each related user according to the priority of the at least one activity; wherein the at least one activity includes the first activity.

[0130] 5. The first priority accumulated value and the second priority accumulated value are updated regularly: When user behavior changes with time and behavior changes, each user behavior is reported by the client to the server and stored in the feature library. The algorithm calculates the feature similarity between the user and other users, so that the first priority accumulated value and the second priority accumulated value are updated regularly. Then, the first priority accumulated value and the second priority accumulated value corresponding to the feature similarity are updated regularly, thereby changing the display priority of the activity and giving priority to displaying the preferred activities to the user. This process does not require manual intervention.

[0131] The beneficial technical effects achieved by the embodiments of the present invention are as follows:

[0132] Determine the relevant users based on the specific circumstances of the specific first activity, and be able to dynamically display the information of the first activity to necessary users; expand the scope of recommended users and improve operational capabilities; avoid the limitations of recommending the first activity to be recommended by guiding information about activities in prominent positions on the website or software by setting time, defining user groups in advance, etc. And directly determine the corresponding display priority when the user is determined to be a relevant user, and display the corresponding activities on the user interface according to the priority level of each activity, avoiding the time-consuming and labor-intensive problem of manually calculating and adjusting the display priority of each activity. Therefore, the method for recommending and issuing activity information in the embodiment of the present invention can be applied to a variety of business scenarios and is easy to apply. The display priority can support dynamic sorting of activities according to the level of user interest, and meet the personalized needs of different users during display.

[0133] It should be understood that the specific order or hierarchy of steps in the disclosed processes is an example of an exemplary method. Based on design preferences, it should be understood that the specific order or hierarchy of steps in the process can be rearranged without departing from the scope of the present disclosure. The accompanying method claims present elements of the various steps in an exemplary order and are not intended to be limited to the specific order or hierarchy described.

[0134] In the foregoing detailed description, various features are grouped together in a single embodiment to simplify the disclosure. This method of disclosure should not be interpreted as reflecting an intention that embodiments of the claimed subject matter require more features than are expressly recited in each claim. On the contrary, as reflected in the appended claims, the invention comprises less than all the features of any individual disclosed embodiment. The appended claims are hereby expressly incorporated into the detailed description, with each claim standing on its own as a separate preferred embodiment of the invention.

[0135] The above description of the disclosed embodiments is intended to enable any person skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other embodiments without departing from the spirit and scope of the present disclosure. Therefore, the present disclosure is not limited to the embodiments presented herein but is intended to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0136] The foregoing description includes examples of one or more embodiments. Of course, it is not possible to describe all possible combinations of components or methods for the purposes of describing the above embodiments, but one of ordinary skill in the art will recognize that the various embodiments may be further combined and arranged. Therefore, the embodiments described herein are intended to encompass all such changes, modifications and variations that fall within the scope of the appended claims. Furthermore, to the extent the term "comprising" is used in the specification or claims, the term is intended to be encompassed in a manner similar to the term "including," as explained in terms of "including," used as a transitional word in the claims. Furthermore, any use of the term "or" in the specification of the claims is intended to mean a "non-exclusive or."

[0137] Those skilled in the art will also appreciate that the various illustrative logical blocks, units, and steps listed in the embodiments of the present invention can be implemented by electronic hardware, computer software, or a combination of the two. To clearly demonstrate the interchangeability of hardware and software, the various illustrative components, units, and steps described above have generally described their functions. Whether such functions are implemented by hardware or software depends on the specific application and the design requirements of the entire system. Those skilled in the art may use various methods to implement the described functions for each specific application, but such implementation should not be understood as exceeding the scope of protection of the embodiments of the present invention.

[0138] The various illustrative logic blocks or units described in the embodiments of the present invention can be implemented or operated by a general-purpose processor, a digital signal processor, an application-specific integrated circuit (ASIC), a field programmable gate array or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof. The general-purpose processor can be a microprocessor, and optionally, the general-purpose processor can also be any conventional processor, controller, microcontroller or state machine. The processor can also be implemented by a combination of computing devices, such as a digital signal processor and a microprocessor, a plurality of microprocessors, one or more microprocessors combined with a digital signal processor core, or any other similar configuration.

[0139] The steps of the methods or algorithms described in the embodiments of the present invention may be directly embedded in hardware, a software module executed by a processor, or a combination of the two. The software module may be stored in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. For example, the storage medium may be connected to the processor so that the processor can read information from the storage medium and write information to the storage medium. Alternatively, the storage medium may also be integrated into the processor. The processor and storage medium may be provided in an ASIC, which may be provided in a user terminal. Alternatively, the processor and storage medium may also be provided in different components in the user terminal.

[0140] In one or more exemplary designs, the above-mentioned functions described in the embodiments of the present invention can be implemented in hardware, software, firmware, or any combination of the three. If implemented in software, these functions can be stored on a computer-readable medium or transmitted in the form of one or more instructions or codes on a computer-readable medium. Computer-readable media include computer storage media and communication media that facilitate the transfer of computer programs from one location to another. Storage media can be any available medium that can be accessed by a general or special computer. For example, such computer-readable media can include but are not limited to RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to carry or store program code in the form of instructions or data structures and other forms that can be read by a general or special computer, or a general or special processor. In addition, any connection can be appropriately defined as a computer-readable medium. For example, if the software is transmitted from a website, server or other remote resource via a coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless methods such as infrared, wireless, and microwave, it is also included in the definition of computer-readable media. The disks and discs mentioned above include compact disks, laser disks, optical disks, DVDs, floppy disks, and Blu-ray discs. Disks typically reproduce data magnetically, while discs typically reproduce data optically with lasers. Combinations of the above may also be included in computer-readable media.

[0141] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An information recommendation method, characterized in that: include: Determining, based on a first user who has participated in a first activity to be recommended, a first related user whose feature similarity with the first user is greater than a first threshold; Determining a second activity similar to the first activity, and determining, based on a second user who participated in the second activity, a second related user whose feature similarity with the second user exceeds a second threshold and who has not participated in the first activity; wherein the features include user behavioral features with respect to the first activity and / or the second activity; For the first related user and the second related user, raising the priority of the first activity corresponding to each related user; the priority indicates the level at which the current activity is displayed to the corresponding user; According to the priority of at least one activity corresponding to each of the first related user and the second related user, information of the at least one activity corresponding to each of the first related user and the second related user is recommended to each related user; wherein the at least one activity includes the first activity.

2. The information recommendation method according to claim 1, characterized in that: The determining, based on a first user who has participated in a first activity to be recommended, a first related user whose feature similarity with the first user is higher than a first threshold, specifically includes: Users who participated in the first activity within a first time period adjacent to the current time are regarded as first users, and similarity is calculated between the user behavior of each first user and the user behavior of each first user to be recommended in the feature library to obtain the feature similarity corresponding to each first user to be recommended. The first users to be recommended whose feature similarity is higher than a first threshold are regarded as first related users. The first users to be recommended are users who participated in activities other than the first activity.

3. The information recommendation method according to claim 2, characterized in that: The calculating the similarity between the user behavior of each first user and the user behavior of each first user to be recommended in the feature library to obtain the feature similarity corresponding to each first user to be recommended specifically includes: Mapping the user behavior of each of the first users in participating in each activity into a point in an N-dimensional Euclidean space; Mapping the user behavior of each first to-be-recommended user in participating in each activity into a point in an N-dimensional Euclidean space; Calculate the Euclidean distance between each of the first users in the N-dimensional Euclidean space and each of the first users to be recommended in the N-dimensional Euclidean space, respectively; Determining a feature similarity corresponding to each of the first to-be-recommended users based on the Euclidean distance; The dimension of the N-dimensional Euclidean space is the same as the total number of activities, all dimensions have a one-to-one correspondence with all activities, the coordinate value on each dimension is the sum of the behavior coefficients of the corresponding users participating in the activities corresponding to the dimension, and N is greater than or equal to 2.

4. The information recommendation method according to claim 1, wherein: The determining, based on the second user who participated in the second activity, a second related user whose feature similarity with the second user is higher than a second threshold and who has not participated in the first activity specifically includes: Users who participated in the second activity within a second time period adjacent to the current time are regarded as second users. Similarity is calculated between the user behavior of each second user and the user behavior of each second user to be recommended in the feature library to obtain a feature similarity corresponding to each second user to be recommended. The second users to be recommended whose feature similarity is higher than a second threshold and who have not participated in the first activity, as well as second users who have not participated in the first activity, are regarded as second related users. The second users to be recommended are users who participated in activities other than the second activity.

5. The information recommendation method according to claim 4, characterized in that: The calculating the similarity between the user behavior of each second user and the user behavior of each second user to be recommended in the feature library to obtain the feature similarity corresponding to each second user to be recommended specifically includes: Mapping the user behavior of each second user participating in each activity into a point in an N-dimensional Euclidean space; Mapping the user behavior of each second to-be-recommended user in participating in each activity into a point in an N-dimensional Euclidean space; Calculate the Euclidean distance between each second user and the second user to be recommended by calculating the Euclidean distance between each second user and the second user to be recommended; Determine the feature similarity corresponding to each of the second to-be-recommended users based on the Euclidean distance; The dimension of the N-dimensional Euclidean space is the same as the total number of activities, all dimensions have a one-to-one correspondence with all activities, the coordinate value on each dimension is the sum of the behavior coefficients of the corresponding users participating in the activities corresponding to the dimension, and N is greater than or equal to 2.

6. The information recommendation method according to claim 1, characterized in that: The step of increasing the priority of the first activity corresponding to each of the first related user and the second related user specifically includes: When the first relevant user is determined, determining a first priority accumulated value corresponding to the first relevant user; adding the first priority accumulated value to the initial priority value corresponding to the first activity to obtain a priority value of the first activity corresponding to the first relevant user; wherein, before the first activity is issued, configuring an initial priority and a corresponding initial priority value for the first activity; When the second relevant user is determined, determining a second priority accumulated value corresponding to the second relevant user; adding the second priority accumulated value to the initial priority value to obtain a priority value of the first activity corresponding to the second relevant user; wherein the first priority accumulated value and the second priority accumulated value are regularly updated; When a target user is simultaneously determined as a relevant user by multiple first activities, a target first activity is configured for the target user from the multiple first activities, and the configured target first activity is displayed to the target user with the highest priority, and the configured target first activity does not participate in the priority adjustment.

7. The information recommendation method according to claim 1, characterized in that: Also includes: Obtain each user's user behavior for all activities and store the obtained user behavior in a distributed publish-subscribe queue; Asynchronously obtaining the user behavior from the distributed publish-subscribe queue and storing the user behavior in a feature library; Taking the user as a unit, the user behavior is marked with a corresponding behavior label and a corresponding behavior coefficient, where the behavior coefficient is a normalized value.

8. The information recommendation method according to claim 1, characterized in that: Also includes: Before the first activity is issued, the data of the first activity is stored in a database, and a first-level cache and a second-level cache are set; When the user obtains data of the first activity, the user first accesses the first-level cache; if the data of the first activity is not obtained from the first-level cache, the user accesses the second-level cache; if the data of the first activity is not obtained from the second-level cache, the user accesses the database.

9. An information recommendation system, characterized in that: include: a related user determining unit, configured to determine, based on a first user who has participated in a first activity to be recommended, a first related user whose feature similarity with the first user is higher than a first threshold; Determining a second activity similar to the first activity, and determining, based on a second user who participated in the second activity, a second related user whose feature similarity with the second user exceeds a second threshold and who has not participated in the first activity; wherein the features include user behavioral features with respect to the first activity and / or the second activity; a priority adjustment unit, configured to increase the priority of the first activity corresponding to each of the first related user and the second related user; the priority indicates a level at which the current activity is displayed to the corresponding user; A recommendation unit is configured to recommend information about the at least one activity corresponding to each of the first related user and the second related user according to a priority of the at least one activity corresponding to each of the first related user and the second related user; wherein the at least one activity includes the first activity.

10. The information recommendation system according to claim 9, characterized in that The relevant user determination unit includes: The first related user determination subunit is used to select users who participated in the first activity within a first time period adjacent to the current time as first users, calculate the similarity between the user behavior of each first user and the user behavior of each first user to be recommended in the feature library, obtain the feature similarity corresponding to each first user to be recommended, and select the first users to be recommended whose feature similarity is higher than a first threshold as first related users. The first users to be recommended are users who participated in activities other than the first activity.

11. The information recommendation system according to claim 9, characterized in that The relevant user determination unit includes: The second related user determination subunit is used to select users who participated in the second activity within a second time period adjacent to the current time as second users, calculate the similarity between the user behavior of each second user and the user behavior of each second user to be recommended in the feature library, obtain the feature similarity corresponding to each second user to be recommended, and select the second users to be recommended whose feature similarity is higher than a second threshold and who have not participated in the first activity, and the second users who have not participated in the first activity as second related users. The second users to be recommended are users who participated in activities other than the second activity.

12. The information recommendation system according to claim 9, characterized in that The priority adjustment unit is specifically configured to: When the first relevant user is determined, determining a first priority accumulated value corresponding to the first relevant user; adding the first priority accumulated value to the initial priority value corresponding to the first activity to obtain a priority value of the first activity corresponding to the first relevant user; wherein, before the first activity is issued, configuring an initial priority and a corresponding initial priority value for the first activity; When the second relevant user is determined, determining a second priority accumulated value corresponding to the second relevant user; adding the second priority accumulated value to the initial priority value to obtain a priority value of the first activity corresponding to the second relevant user; wherein the first priority accumulated value and the second priority accumulated value are regularly updated; When a target user is simultaneously determined as a relevant user by multiple first activities, a target first activity is configured for the target user from the multiple first activities, and the configured target first activity is displayed to the target user with the highest priority, and the configured target first activity does not participate in the priority adjustment.

Images