Activity recommendation method, device, and storage medium
By constructing user profiles and adjusting the weights of feature dimensions, personalized activity recommendations are made based on user data and historical recommendation data, solving the problem of low recommendation accuracy in existing technologies and achieving more efficient activity recommendations.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN CARKU TECH CO LTD
- Filing Date
- 2026-05-15
- Publication Date
- 2026-07-10
AI Technical Summary
Existing activity recommendation methods cannot provide personalized recommendations based on users' actual situations, resulting in low recommendation accuracy.
By constructing user profiles, we can determine whether a user is a cold start user or a non-cold start user. Based on user data and historical recommendation data, we can adjust the weights of feature dimensions and determine personalized recommendation activities based on the matching degree.
This improved the accuracy of activity recommendations, reduced invalid recommendations, and enhanced user participation in activities.
Smart Images
Figure CN122364531A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of data processing technology, and more particularly to an activity recommendation method, apparatus, and storage medium. Background Technology
[0002] With increasingly fierce market competition, event recommendations, as a key link connecting users and brands, have become a core means of enhancing user engagement and strengthening brand loyalty. Current industry methods for event recommendations do not differentiate between user characteristics and needs, uniformly pushing all events to all users. While this approach has low development costs, the accuracy of event recommendations is low. Therefore, improving the accuracy of event recommendations has become a critical issue that urgently needs to be addressed. Summary of the Invention
[0003] The main objective of this application is to provide an activity recommendation method, device, and storage medium, which aims to improve the accuracy of activity recommendations.
[0004] In a first aspect, this application provides an activity recommendation method, apparatus, and storage medium, the activity recommendation method comprising: The user profile of the user to be recommended is constructed based on the user data of the user to be recommended, wherein the user profile contains data corresponding to multiple feature dimensions; Based on the data corresponding to each feature dimension of the user profile, determine the data completeness corresponding to each feature dimension; Based on the data completeness corresponding to each of the aforementioned feature dimensions, the user type of the user to be recommended is determined to be either a cold start user or a non-cold start user. If the user type of the user to be recommended is a non-cold start user, obtain the historical recommendation data of the user to be recommended. Based on the data completeness corresponding to each feature dimension and the historical recommendation data, the weights corresponding to each feature dimension are adjusted to obtain the target weights corresponding to each feature dimension; and Based on the multiple feature dimensions of the user profile, determine the matching degree between each feature dimension and the recommendation activity; Based on the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity, at least one target recommendation activity is determined from the multiple recommendation activities and recommended to the user to be recommended.
[0005] Secondly, this application also provides a computer device, the computer device including a processor, a memory, and a computer program stored in the memory and executable by the processor, wherein when the computer program is executed by the processor, it implements the steps of the activity recommendation method as described above.
[0006] Thirdly, this application also provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, it implements the steps of the activity recommendation method described above.
[0007] This application provides an activity recommendation method, device, and storage medium. The activity recommendation method first constructs a user profile of the user to be recommended based on the user data of the user to be recommended. The user profile includes data corresponding to multiple feature dimensions. Next, based on the data corresponding to each feature dimension of the user profile, the data completeness corresponding to each feature dimension is determined, and the user to be recommended is classified as a cold-start user or a non-cold-start user based on the data completeness corresponding to each feature dimension. In the case where the user type of the user to be recommended is a non-cold-start user, historical recommendation data of the user to be recommended is obtained, and based on the data completeness corresponding to each feature dimension and the historical recommendation data, the completeness of each feature dimension is determined. The system determines the target weights and the matching degree between each feature dimension and the recommendation activity based on multiple feature dimensions of the user profile. Because the user data of different users to be recommended varies, the user profiles and data corresponding to multiple feature dimensions for different users to be recommended are also different. Consequently, the target weights for each feature dimension and the matching degree between each feature dimension and the recommendation activity have individual differences. Based on the target weights for each feature dimension and the matching degree between each feature dimension and the recommendation activity, at least one recommendation activity that fits the user's actual needs is selected from multiple recommendation activities as the target recommendation activity and recommended to the user to be recommended. This enables personalized recommendations and significantly improves the accuracy of activity recommendations. Attached Figure Description
[0008] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0009] Figure 1 This application provides an illustration of an activity recommendation method. Figure 2 This application provides a flowchart illustrating the steps of an activity recommendation method. Figure 3 A schematic diagram of the user profile of the user to be recommended; Figure 4 for Figure 2 A flowchart illustrating a sub-step of the activity recommendation method in the document; Figure 5 for Figure 2 A flowchart illustrating another sub-step of the activity recommendation method in the process; Figure 6 This is a schematic block diagram of a computer device provided in an embodiment of this application.
[0010] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation
[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0012] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.
[0013] With increasingly fierce market competition, companies often use promotional activities to enhance their market competitiveness. Promotional activities have become a key link connecting users and brands and are a common method of customer acquisition.
[0014] Current activity recommendation methods typically employ either full-scale push notifications or simple rule-based matching (such as by region or user type) to recommend activities to users. While these methods have low development costs, they suffer from low accuracy and fail to provide personalized recommendations based on individual user circumstances. For example, pushing full-scale notifications based on age, such as mass-produced business exhibitions or health and wellness activities for middle-aged and elderly users favoring arts and leisure activities, fails to consider user interests and consumption habits. This results in content that is significantly mismatched with users' actual needs, leading to low accuracy in activity recommendations.
[0015] Based on this, embodiments of this application provide an activity recommendation method, apparatus, and storage medium. The activity recommendation method can be applied to computer equipment, which may include terminal devices and servers, for example, it can be used for... Figure 1 The application software 20 in the terminal device 10 shown. The terminal device can be an electronic device such as a mobile phone, tablet computer, laptop computer, desktop computer, personal digital assistant and wearable device; the server can be a single server or a server cluster composed of multiple servers.
[0016] The activity recommendation method provided in this embodiment of the invention can be applied to, for example... Figure 1In the application environment. For example, such as Figure 1 As shown in Figure a, the user to be recommended enters their user ID and login password on the login interface of the application software 20 on the terminal device 10. The application software 20 on the terminal device obtains the user data of the user to be recommended through the user ID, and constructs a user profile of the user to be recommended based on the user data. The user profile contains data corresponding to multiple feature dimensions. Based on the data corresponding to each feature dimension of the user profile, the data completeness corresponding to each feature dimension is determined. Based on the data completeness corresponding to each feature dimension, the user type of the user to be recommended is determined to be either a cold start user or a non-cold start user. In the case where the user type of the user to be recommended is a non-cold start user, the historical recommendation data of the user to be recommended is obtained. Based on the data completeness corresponding to each feature dimension and the historical recommendation data, the weight corresponding to each feature dimension is adjusted to obtain the target weight corresponding to each feature dimension. Based on the multiple feature dimensions of the user profile, the matching degree between each feature dimension and the recommendation activity is determined. Based on the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity, at least one target recommendation activity is determined from multiple recommendation activities and recommended to the user to be recommended. Figure 1 As shown in b, this reduces invalid recommendations and improves the accuracy of activity recommendations.
[0017] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.
[0018] Please refer to Figure 2 , Figure 2 This is a flowchart illustrating the steps of an activity recommendation method provided in an embodiment of this application.
[0019] like Figure 2 As shown, the recommended method for this activity includes steps S101 to S107.
[0020] Step S101: Construct a user profile of the user to be recommended based on the user data of the user to be recommended. The user profile contains data corresponding to multiple feature dimensions.
[0021] The user data for the users to be recommended may include data corresponding to device feature dimensions, geographical location feature dimensions, usage habit feature dimensions, historical behavior feature dimensions, device status feature dimensions, and may also include basic user attribute data, etc., without limitation.
[0022] Among them, the device feature dimension is used to characterize the inherent attributes of the user's terminal device / server itself. The data corresponding to the device feature dimension may include: device model, device brand, terminal device type (mobile phone / tablet / computer), operating system type (such as Android / iOS / HarmonyOS / Windows), device resolution, device unique identifier, device carrier information, hardware configuration parameters, etc., which are not limited here.
[0023] The geolocation feature dimension is used to characterize the features of a user's real-time or permanent geolocation. The data corresponding to the geolocation feature dimension may include: real-time latitude and longitude, current province and city, urban area, business district location, permanent residence, frequently visited locations, IP address, etc., without limitation.
[0024] The usage habit feature dimension is used to characterize the long-term behavioral habits of users when using application software. The data corresponding to the usage habit feature dimension may include: daily login time of application, weekly / monthly usage frequency, single usage duration, access records of frequently used function modules, application uninstallation / reinstallation records, page jump preferences, fixed active time periods, etc., without limitation.
[0025] The historical behavior feature dimension is used to characterize the features of users' historical interactions and activity-related behaviors. The data corresponding to the historical behavior feature dimension may include: the types of activities viewed in the past, activity click records, the time spent on the activity details page, activity registration / check-in records, activity collection, sharing, liking behavior, the time and type preference of past participation in activities, etc., without limitation.
[0026] The device status feature dimension is used to characterize the current real-time operating status of the terminal device / server. The data corresponding to the device status feature dimension may include: remaining battery power, network connection type (WiFi / mobile data), network signal strength, screen on / off status, device silent status, number of background applications running, whether the device is in standby mode, etc., without limitation.
[0027] User basic attribute data is used to represent the inherent basic attribute characteristics of a user, which may include user ID, age, gender, occupation, education, membership level, registration time, user tags, consumption level, interest tags, industry attributes, etc., without limitation.
[0028] For example, such as Figure 3 As shown, Figure 3 This is a schematic diagram of a user profile for users to be recommended, constructed based on user data. The user profile includes data corresponding to the geographic location dimension: {Permanent Residence - Tianhe District, Guangzhou City, Frequently Visited Locations - XX Street}, and data corresponding to the historical behavior dimension: {Categories of Activities Viewed - Outdoor Hiking Activities, Interactive Behavior - Liking Outdoor Hiking Videos}.
[0029] It should be noted that building user profiles for users to be recommended based on their user data can better match their actual needs, avoid ineffective recommendations, and ensure the accuracy of activity recommendations.
[0030] Step S102: Based on the data corresponding to each feature dimension of the user profile, determine the data completeness corresponding to each feature dimension.
[0031] It should be noted that determining the data completeness for each feature dimension of the user profile can be achieved by pre-defining the standard data items that should be included under each feature dimension (such as device feature dimension, geographic location feature dimension, historical behavior feature dimension, etc.). For example, the geographic location feature dimension can be pre-defined to include the following four data items: city of residence, district of residence, business district frequented, and IP address. Next, the actual number of data items included in each feature dimension of the user profile is counted. For instance, if the user profile includes a geographic location feature dimension with the data items {city of residence - Guangzhou, district of residence - Tianhe District}, totaling 2 data items, then the data completeness for the geographic location feature dimension is 0.5 = (2 / 4).
[0032] It should be noted that by pre-defining the standard data items that should be included under each feature dimension, we can not only determine the evaluation benchmark and ensure the objectivity of the evaluation results, but also determine which data is still missing in the user profile. This can help in subsequent user data collection and make the user profile more and more complete.
[0033] Step S103: Based on the data completeness corresponding to each feature dimension, determine whether the user type of the user to be recommended is a cold start user or a non-cold start user.
[0034] In one embodiment, determining whether a user to be recommended is a cold-start user or a non-cold-start user based on the data completeness corresponding to each feature dimension includes: determining the data completeness of the user profile based on the data completeness corresponding to each feature dimension; if the data completeness of the user profile is less than a preset data completeness threshold, determining that the user to be recommended is a cold-start user; if the data completeness of the user profile is greater than or equal to the preset data completeness threshold, determining that the user to be recommended is a non-cold-start user.
[0035] It should be noted that determining the data completeness of a user profile based on the data completeness of each feature dimension can include summing the data completeness scores for each feature dimension and then dividing by the sum of the number of feature dimensions. For example, a user profile may include five feature dimensions: geographic location, usage habits, device, historical behavior, and device status. If the data completeness score for geographic location is 0.5, for usage habits it's 0.6, for device it's 0.4, for historical behavior it's 0.2, and for device status it's 0.5, then the data completeness score for the user profile is (0.5 + 0.6 + 0.4 + 0.2 + 0.5) / 5 = 0.44, which is less than the preset data completeness threshold of 0.5. Therefore, the user type for recommendation is determined to be a cold-start user.
[0036] It should be noted that by ensuring the completeness of user profile data, we can gain a comprehensive understanding of users, guarantee the reliability of user profiles, lay the foundation for accurate recommendations of activities to users in the future, and improve users' enthusiasm for participating in activities.
[0037] In one embodiment, after determining whether the user type of the user to be recommended is a cold-start user or a non-cold-start user based on the data completeness corresponding to each feature dimension, the method further includes: if the user type of the user to be recommended is a cold-start user, determining the activity participation location of the recommendation activity; determining whether the distance between the activity participation location of the recommendation activity and the geographical location of the user to be recommended is less than a preset distance threshold; if it is less than the preset distance threshold, then the recommendation activity is taken as the target recommendation activity; and determining at least one target recommendation activity from multiple recommendation activities to recommend to the user to be recommended.
[0038] For example, multiple recommendation activities include: Recommendation Activity A, Recommendation Activity B, and Recommendation Activity C; wherein, the participation location of Recommendation Activity A is XX Park, the participation location of Recommendation Activity B is XX School, and the participation location of Recommendation Activity C is XX Community; the distance between the participation location of Recommendation Activity A and the geographical location of the user to be recommended is 5 kilometers, which is less than the preset distance threshold of 10 kilometers; the distance between the participation location of Recommendation Activity B and the geographical location of the user to be recommended is 20 kilometers, which is greater than the preset distance threshold of 10 kilometers; and the distance between the participation location of Recommendation Activity C and the geographical location of the user to be recommended is 8 kilometers, which is less than the preset distance threshold of 10 kilometers. In this case, Recommendation Activity A and Recommendation Activity C are selected as target recommendation activities and recommended to the user to be recommended.
[0039] It should be noted that by using the distance between the activity location and the user's geographical location as the filtering criterion, recommended activities that are close to the user are prioritized. The judgment logic is simple and the recommendation efficiency is high.
[0040] It should be noted that when the user type to be recommended is a cold-start user, the above methods are not the only limitations for recommending activities to cold-start users. If at least one feature dimension of the user profile is related to the content of the recommended activity, that recommended activity can also be used as the target recommended activity to be recommended to the user.
[0041] For example, a user profile includes a device feature dimension and a geographic location feature dimension. The device feature dimension includes {device model: a**c, terminal device type: mobile phone}, and the geographic location feature dimension includes {residential location: Tianhe District, Guangzhou City, frequented locations: XX Street}. The content of the recommended activity is: "Mobile phones with device model a**c can enjoy a 50% discount on XX products." Since this is related to the device feature dimension of the user profile, this recommended activity is recommended to the users to be recommended.
[0042] For example, if the user type to be recommended is a cold start user, and the target audience of the recommendation activity is also cold start users, then the recommendation activity will be recommended to the user to be recommended as the target recommendation activity.
[0043] Step S104: If the user type of the user to be recommended is a non-cold start user, obtain the historical recommendation data of the user to be recommended.
[0044] Historical recommendation data is used to represent the data generated when recommending activities to users based on target feature dimensions, such as user activity participation rates. Target feature dimensions may include device feature dimensions, geographic location feature dimensions, usage habit feature dimensions, historical behavior feature dimensions, device status feature dimensions, etc., and are not limited here.
[0045] For example, if the user to be recommended frequently appears in Tianhe District (corresponding to the geographical location feature dimension), then the statistics are the activities A1, A2, and A3 recommended by the user in Tianhe District, which are a total of 3. If the user participated in 2 activities, then the user's activity participation rate is 0.67≈(2 / 3).
[0046] Step S105: Adjust the weights of each feature dimension based on the data completeness and historical recommendation data corresponding to each feature dimension to obtain the target weights for each feature dimension.
[0047] In one embodiment, such as Figure 4As shown, based on the data completeness and historical recommendation data corresponding to each feature dimension, the weights corresponding to each feature dimension are adjusted to obtain the target weights for each feature dimension, including steps S1051 to S1053: Step S1051: Adjust the initial weights of each feature dimension according to the data completeness of each feature dimension to obtain the first weights of each feature dimension.
[0048] In one embodiment, adjusting the initial weight of each feature dimension based on the data completeness of each feature dimension to obtain the first weight of each feature dimension may include: determining the increment of the initial weight of each feature dimension based on the data completeness of each feature dimension; and obtaining the first weight of each feature dimension based on the increment of the initial weight of each feature dimension and the initial weight of each feature dimension.
[0049] It should be noted that the greater the data completeness corresponding to a feature dimension, the more complete the data for that feature dimension, and the clearer the understanding of user needs and preferences, making it easier to recommend suitable activities to users. The initial weight increment for each feature dimension can be determined using a pre-defined first data table. For example, the pre-defined first data table is shown in Table 1.
[0050]
[0051] For example, the initial weight corresponding to the geographic location feature dimension is 0.2, the initial weight corresponding to the historical behavior feature dimension is 0.2; the data completeness corresponding to the geographic location feature dimension is 0.5, and the increment of its corresponding initial weight is 0.06 according to the preset first data table; the data completeness corresponding to the historical behavior feature dimension is 0.2, and the increment of its corresponding initial weight is 0.02 according to the preset first data table; then the first weight corresponding to the geographic location feature dimension is 0.26 (0.2+0.06), and the first weight corresponding to the historical behavior feature dimension is 0.22 (0.2+0.02).
[0052] Since feature dimensions with higher data completeness have higher data credibility, adjusting the initial weight of each feature dimension based on its data completeness to obtain its first weight allows feature dimensions with high data completeness to play a greater role in the activity recommendation process, reduces interference from feature dimensions with low data completeness, and improves the accuracy of activity recommendations.
[0053] Step S1052: Based on historical recommendation data, adjust the first weight corresponding to each feature dimension to obtain the second weight corresponding to each feature dimension.
[0054] In one embodiment, adjusting the first weight corresponding to each feature dimension based on historical recommendation data to obtain the second weight corresponding to each feature dimension includes: determining the increment of the first weight corresponding to each feature dimension based on historical recommendation data; and obtaining the second weight corresponding to each feature dimension based on the increment of the first weight corresponding to each feature dimension and the first weight corresponding to each feature dimension.
[0055] It's important to note that recommending activities to users solely based on the data completeness corresponding to each feature dimension cannot adapt to changes in user preferences and will lead to decreased recommendation accuracy. Therefore, to further ensure the accuracy of activity recommendations, the first weight corresponding to each feature dimension can be adjusted based on historical recommendation data to obtain a second weight for each feature dimension. This allows recommended activities to keep pace with changes in user preferences and effectively improves the accuracy of activity recommendations.
[0056] Historical recommendation data can include user activity participation rates. The level of historical recommendation effectiveness can be determined based on user activity participation rates. The higher the level of historical recommendation effectiveness corresponding to the target feature dimension, the more willing the users to be recommended are to participate in recommendation activities related to the target feature dimension.
[0057] For example, the increment of the first weight corresponding to each feature dimension can be determined through a preset second data table. The preset second data table is shown in Table 2.
[0058] If the target feature dimension includes the geographic location feature dimension, the activity is recommended to the user to be recommended based on the geographic location feature dimension. The activity participation rate of the user to be recommended is 0.7. According to the preset second data table, the increment of the first weight corresponding to the geographic location feature dimension is 0.05. When the first weight corresponding to the geographic location feature dimension is 0.26, the second weight corresponding to the geographic location feature dimension is 0.31 (0.26+0.05).
[0059]
[0060] Step S1053: Normalize the second weights corresponding to multiple feature dimensions to obtain the target weights corresponding to each feature dimension of the user profile.
[0061] For example, if the second weight corresponding to the geographic location feature dimension is 0.31, the second weight corresponding to the historical behavior feature dimension is 0.25, the second weight corresponding to the device feature dimension is 0.22, the second weight corresponding to the usage habit feature dimension is 0.4, and the second weight corresponding to the device status feature dimension is 0.36, then the target weight corresponding to the geographic location feature dimension is 0.2≈(0.31 / 1.54), the target weight corresponding to the historical behavior feature dimension is 0.16≈(0.25 / 1.54), the target weight corresponding to the device feature dimension is 0.14≈(0.22 / 1.54), the target weight corresponding to the usage habit feature dimension is 0.26≈(0.4 / 1.54), and the target weight corresponding to the device status feature dimension is 0.23≈(0.36 / 1.54).
[0062] It should be noted that normalizing the second weights corresponding to multiple feature dimensions can reduce the difference in dimensionality between different feature dimensions, thus providing support for improving the accuracy of subsequent activity recommendations.
[0063] Step S106: Based on multiple feature dimensions of the user profile, determine the matching degree between each feature dimension and the recommendation activity.
[0064] It should be noted that the matching degree between each feature dimension and the recommendation activity can be determined based on the matching function corresponding to each feature dimension, and the matching function corresponding to each feature dimension is different. Those skilled in the art can set the matching function corresponding to each feature dimension according to the actual situation, and there are no restrictions here.
[0065] For example, the geographic location feature dimension corresponds to a geographic location matching function. Determining the matching degree between the geographic location feature dimension and the recommended activity based on the geographic location matching function can include: if the user's location is the location of the recommended activity, then the matching degree between the geographic location feature dimension and the recommended activity is 1; if the user's location is unknown, then the matching degree between the geographic location feature dimension and the recommended activity is 0; if the user's location is known, then the matching degree between the geographic location feature dimension and the recommended activity is calculated based on (1 - distance between the user's location and the location of the recommended activity / preset distance threshold). For example, if the user's location is 5 km away from the location of the recommended activity, and the preset distance threshold is 8 km, then the matching degree between the geographic location feature dimension and the recommended activity = (1 - 5 / 8) = 0.375.
[0066] For example, the historical behavior feature dimension corresponds to the historical behavior matching function. The historical behavior matching function can be used to check whether the user to be recommended has participated in similar activities, determine the number of similar activities the user to be recommended has participated in, and determine whether the user to be recommended has purchased related products and the number of related products purchased. If the number of similar activities the user to be recommended has participated in and the number of related products purchased are not 0, then the matching degree between the historical behavior feature dimension and the recommended activity = minimum value (1.0, (number of similar activities * 0.3 + number of related purchases * 0.2)).
[0067] It should be noted that by determining the matching degree between each feature dimension and the recommendation activity, the degree of matching between the feature dimension and the recommendation activity can be determined, effectively improving the accuracy of the recommendation activity.
[0068] Step S107: Based on the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity, determine at least one target recommendation activity from multiple recommendation activities and recommend it to the user to be recommended.
[0069] In one embodiment, such as Figure 5 As shown, based on the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity, at least one target recommendation activity is determined from multiple recommendation activities and recommended to the user to be recommended, including steps S1071 to S1073: Step S1071: Determine the comprehensive matching score between the user to be recommended and the recommendation activity based on the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity.
[0070] In one embodiment, the comprehensive matching score between the user to be recommended and the recommendation activity is determined based on the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity. This includes: determining a first matching score between each feature dimension and the recommendation activity based on the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity; and accumulating multiple first matching scores to obtain the comprehensive matching score between the user to be recommended and the recommendation activity.
[0071] For example, the target weight for the geographic location feature dimension is 0.2, the matching degree between the geographic location feature dimension and the recommendation activity is 0.3, and the first matching score between the geographic location feature dimension and the recommendation activity is 0.2*0.3=0.06; the target weight for the historical behavior feature dimension is 0.16, the matching degree between the historical behavior feature dimension and the recommendation activity is 0.2, and the first matching score between the historical behavior feature dimension and the recommendation activity is 0.16*0.2=0.032; the target weight for the device feature dimension is 0.14, the matching degree between the device feature dimension and the recommendation activity is 0.4, and the first matching score between the device feature dimension and the recommendation activity is 0.14*0.4=0.056; the comprehensive matching score between the user to be recommended and the recommendation activity is 0.06+0.032+0.056=0.148.
[0072] It should be noted that by determining the comprehensive matching score between the user to be recommended and the recommended activity through the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity, the recommendation of activities can be made more reasonable and accurate, avoiding the need to make activity recommendations based on a single feature dimension.
[0073] Step S1072: Sort the multiple recommendation activities according to the comprehensive matching score between the user to be recommended and the multiple recommendation activities.
[0074] For example, multiple recommendation activities include: recommendation activity A, recommendation activity B, and recommendation activity C; the overall matching score of the user to be recommended with recommendation activity A is 0.148, the overall matching score with recommendation activity B is 0.18, and the overall matching score with recommendation activity C is 0.4. Then, the users are sorted according to the overall matching scores of each recommendation activity in ascending or descending order.
[0075] Step S1073: From the sorted multiple recommendation activities, determine at least one target recommendation activity with a comprehensive matching score greater than or equal to a preset comprehensive matching score threshold and recommend it to the user to be recommended.
[0076] For example, if the overall matching score between the user to be recommended and recommendation activity A is 0.5, the overall matching score with recommendation activity B is 0.8, and the overall matching score with recommendation activity C is 0.9, the order of recommendation activities A, B, and C is recommendation activity C > recommendation activity B > recommendation activity A. If the preset overall matching score threshold is 0.8, then recommendation activities C and B will be recommended to the user to be recommended as target recommendation activities.
[0077] It should be noted that when the user type to be recommended is a non-cold-start user, the above methods are not the only approach for recommending activities to non-cold-start users. If the user type to be recommended is a non-cold-start user, and the target audience of the recommendation activity is also non-cold-start users, then that recommendation activity will be recommended to the user to be recommended.
[0078] The activity recommendation method provided in the above embodiments first constructs a user profile of the user to be recommended based on the user data of the user to be recommended. The user profile includes data corresponding to multiple feature dimensions. Next, based on the data corresponding to each feature dimension of the user profile, the data completeness corresponding to each feature dimension is determined, and the user to be recommended is divided into cold-start users or non-cold-start users based on the data completeness corresponding to each feature dimension. In the case where the user type of the user to be recommended is a non-cold-start user, historical recommendation data of the user to be recommended is obtained, and the target weight corresponding to each feature dimension is determined based on the data completeness corresponding to each feature dimension and the historical recommendation data. Finally, based on... The user profile is analyzed using multiple feature dimensions to determine the matching degree between each feature dimension and the recommendation activity. Because user data varies among different users to be recommended, the corresponding user profiles and data for each feature dimension also differ. Consequently, the target weight for each feature dimension and the matching degree between each feature dimension and the recommendation activity exhibit individual differences. Based on the target weight for each feature dimension and the matching degree between each feature dimension and the recommendation activity, at least one recommendation activity that aligns with the user's actual needs is selected from multiple recommendation activities as the target recommendation activity and recommended to the user to be recommended. This enables personalized recommendations and significantly improves the accuracy of activity recommendations.
[0079] The apparatus provided in the above embodiments can be implemented as a computer program, which can be used in, for example... Figure 6 It runs on the computer device shown.
[0080] Please see Figure 6 , Figure 6 This is a schematic block diagram of the structure of a computer device provided in an embodiment of this application.
[0081] like Figure 6 As shown, the computer device includes a processor, memory, and network interface connected via a system bus. The memory may include a storage medium and internal memory, and the storage medium may be non-volatile or volatile.
[0082] The storage medium may store the operating system and computer programs. These computer programs include program instructions that, when executed, cause the processor to perform any recommended method of activity.
[0083] The processor provides computing and control capabilities, supporting the operation of the entire computer device.
[0084] Internal memory provides an environment for the execution of computer programs stored in storage media. When these computer programs are executed by a processor, the processor can perform any recommended method of activity.
[0085] This network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0086] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.
[0087] In one embodiment, the processor is configured to run a computer program stored in memory to perform the following steps: The user profile of the user to be recommended is constructed based on the user data of the user to be recommended, wherein the user profile contains data corresponding to multiple feature dimensions; Based on the data corresponding to each feature dimension of the user profile, determine the data completeness corresponding to each feature dimension; Based on the data completeness corresponding to each of the aforementioned feature dimensions, the user type of the user to be recommended is determined to be either a cold start user or a non-cold start user. If the user type of the user to be recommended is a non-cold start user, obtain the historical recommendation data of the user to be recommended. Based on the data completeness corresponding to each feature dimension and the historical recommendation data, the weights corresponding to each feature dimension are adjusted to obtain the target weights corresponding to each feature dimension; and Based on the multiple feature dimensions of the user profile, determine the matching degree between each feature dimension and the recommendation activity; Based on the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity, at least one target recommendation activity is determined from the multiple recommendation activities and recommended to the user to be recommended.
[0088] In one embodiment, when the processor adjusts the weights corresponding to each feature dimension based on the data completeness and historical recommendation data corresponding to each feature dimension to obtain the target weights corresponding to each feature dimension, it is configured to: Based on the data completeness corresponding to each feature dimension, the initial weight corresponding to each feature dimension is adjusted to obtain the first weight corresponding to each feature dimension; Based on the historical recommendation data, the first weight corresponding to each feature dimension is adjusted to obtain the second weight corresponding to each feature dimension; The second weights corresponding to the multiple feature dimensions are normalized to obtain the target weights corresponding to each feature dimension of the user profile.
[0089] In one embodiment, when the processor adjusts the initial weights corresponding to each feature dimension based on the data completeness corresponding to each feature dimension to obtain the first weights corresponding to each feature dimension, it is configured to: Based on the data completeness corresponding to each feature dimension, determine the increment of the initial weight corresponding to each feature dimension; The first weight corresponding to each feature dimension is obtained based on the increment of the initial weight corresponding to each feature dimension and the initial weight corresponding to each feature dimension.
[0090] In one embodiment, when the processor adjusts the first weight corresponding to each feature dimension based on the historical recommendation data to obtain the second weight corresponding to each feature dimension, it is configured to: Based on the historical recommendation data, determine the increment of the first weight corresponding to each feature dimension; The second weight corresponding to each feature dimension is obtained based on the increment of the first weight corresponding to each feature dimension and the first weight corresponding to each feature dimension.
[0091] In one embodiment, when the processor determines at least one target recommendation activity from among the multiple recommendation activities and recommends it to the user to be recommended based on the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity, it is configured to: The comprehensive matching score between the user to be recommended and the recommendation activity is determined based on the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity. The multiple recommendation activities are ranked according to the comprehensive matching score between the user to be recommended and the multiple recommendation activities; From the sorted list of recommended activities, at least one target recommended activity with a comprehensive matching score greater than or equal to a preset comprehensive matching score threshold is selected and recommended to the user to be recommended.
[0092] In one embodiment, when the processor implements the step of determining the comprehensive matching score between the user to be recommended and the recommendation activity based on the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity, it is configured to: Based on the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity, a first matching score between each feature dimension and the recommendation activity is determined; By summing up multiple first matching scores, a comprehensive matching score between the user to be recommended and the recommendation activity is obtained.
[0093] In one embodiment, after determining whether the user type of the user to be recommended is a cold-start user or a non-cold-start user based on the data completeness corresponding to each feature dimension, the processor is further configured to implement: In the case where the user type of the user to be recommended is a cold start user, the activity participation location of the recommendation activity is determined; Determine whether the distance between the location of the recommended activity and the geographical location of the user to be recommended is less than a preset distance threshold; If the distance is less than a preset distance threshold, then the recommended activity will be taken as the target recommended activity; From the multiple recommendation activities, at least one target recommendation activity is determined and recommended to the user to be recommended.
[0094] In one embodiment, when the processor determines whether the user type of the user to be recommended is a cold-start user or a non-cold-start user based on the data completeness corresponding to each feature dimension, it is configured to: The data completeness of the user profile is determined based on the data completeness corresponding to each of the aforementioned feature dimensions; If the data completeness of the user profile is less than a preset data completeness threshold, the user type of the user to be recommended is determined to be a cold start user. If the data completeness of the user profile is greater than or equal to a preset data completeness threshold, the user type of the user to be recommended is determined to be a non-cold start user.
[0095] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the computer device described above can be referred to the corresponding process in the aforementioned activity recommendation method embodiments, and will not be repeated here.
[0096] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can 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. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.
[0097] This application also provides a computer-readable storage medium storing a computer program, the computer program including program instructions, and the method implemented when the program instructions are executed can be referred to various embodiments of the recommended method in this application.
[0098] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiments, such as the hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, SmartMedia Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the computer device.
[0099] Furthermore, the computer's usable storage medium may primarily include a stored program area and a stored data area. The stored program area may store the operating system, applications required for at least one function, etc.; the stored data area may store data created based on the use of blockchain nodes, etc. The blockchain referred to in this application is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. A blockchain is essentially a decentralized database, a chain of data blocks linked using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain may include a blockchain underlying platform, a platform product service layer, and an application service layer, etc.
[0100] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.
[0101] It should also be understood that the term "and / or" as used in this specification and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations. It should be noted that, herein, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.
[0102] It should be noted that any AI models, software tools, or components not belonging to this company appearing in the embodiments of this application are merely illustrative examples and do not represent actual use. All user personal information involved in the embodiments of this application has been authorized (with the knowledge and consent) by the relevant parties or has been fully authorized by all parties, and the executing entity may obtain it through various legal and compliant means. The collection, storage, use, processing, transmission, provision, and disclosure of the information, data, and signals involved all comply with relevant laws and regulations and do not violate public order and good morals.
[0103] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments. The above descriptions are merely specific implementations of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. An activity recommendation method, characterized in that, include: The user profile of the user to be recommended is constructed based on the user data of the user to be recommended, wherein the user profile contains data corresponding to multiple feature dimensions; Based on the data corresponding to each feature dimension of the user profile, determine the data completeness corresponding to each feature dimension; Based on the data completeness corresponding to each of the aforementioned feature dimensions, the user type of the user to be recommended is determined to be either a cold start user or a non-cold start user. If the user type of the user to be recommended is a non-cold start user, obtain the historical recommendation data of the user to be recommended. Based on the data completeness corresponding to each feature dimension and the historical recommendation data, the weights corresponding to each feature dimension are adjusted to obtain the target weights corresponding to each feature dimension; and Based on the multiple feature dimensions of the user profile, determine the matching degree between each feature dimension and the recommendation activity; Based on the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity, at least one target recommendation activity is determined from the multiple recommendation activities and recommended to the user to be recommended.
2. The activity recommendation method as described in claim 1, characterized in that, The step of adjusting the weight of each feature dimension based on the data completeness and the historical recommendation data to obtain the target weight of each feature dimension includes: Based on the data completeness corresponding to each feature dimension, the initial weight corresponding to each feature dimension is adjusted to obtain the first weight corresponding to each feature dimension; Based on the historical recommendation data, the first weight corresponding to each feature dimension is adjusted to obtain the second weight corresponding to each feature dimension; The second weights corresponding to the multiple feature dimensions are normalized to obtain the target weights corresponding to each feature dimension of the user profile.
3. The activity recommendation method as described in claim 2, characterized in that, The step of adjusting the initial weights corresponding to each feature dimension based on the data completeness corresponding to each feature dimension to obtain the first weights corresponding to each feature dimension includes: Based on the data completeness corresponding to each feature dimension, determine the increment of the initial weight corresponding to each feature dimension; The first weight corresponding to each feature dimension is obtained based on the increment of the initial weight corresponding to each feature dimension and the initial weight corresponding to each feature dimension.
4. The activity recommendation method as described in claim 2, characterized in that, The step of adjusting the first weight corresponding to each feature dimension based on the historical recommendation data to obtain the second weight corresponding to each feature dimension includes: Based on the historical recommendation data, determine the increment of the first weight corresponding to each feature dimension; The second weight corresponding to each feature dimension is obtained based on the increment of the first weight corresponding to each feature dimension and the first weight corresponding to each feature dimension.
5. The activity recommendation method as described in any one of claims 1-4, characterized in that, The step of determining at least one target recommendation activity from among the multiple recommendation activities and recommending it to the user to be recommended, based on the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity, includes: The comprehensive matching score between the user to be recommended and the recommendation activity is determined based on the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity. The multiple recommendation activities are ranked according to the comprehensive matching score between the user to be recommended and the multiple recommendation activities; From the sorted list of recommended activities, at least one target recommended activity with a comprehensive matching score greater than or equal to a preset comprehensive matching score threshold is selected and recommended to the user to be recommended.
6. The activity recommendation method as described in claim 5, characterized in that, The step of determining the comprehensive matching score between the user to be recommended and the recommendation activity based on the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity includes: Based on the target weight corresponding to each feature dimension and the matching degree between each feature dimension and the recommendation activity, a first matching score between each feature dimension and the recommendation activity is determined; By summing up multiple first matching scores, a comprehensive matching score between the user to be recommended and the recommendation activity is obtained.
7. The activity recommendation method as described in any one of claims 1-4, characterized in that, After determining whether the user type of the user to be recommended is a cold-start user or a non-cold-start user based on the data completeness corresponding to each feature dimension, the method further includes: In the case where the user type of the user to be recommended is a cold start user, the activity participation location of the recommendation activity is determined; Determine whether the distance between the location of the recommended activity and the geographical location of the user to be recommended is less than a preset distance threshold; If the distance is less than a preset distance threshold, then the recommended activity will be taken as the target recommended activity; From the multiple recommendation activities, at least one target recommendation activity is determined and recommended to the user to be recommended.
8. The activity recommendation method as described in any one of claims 1-4, characterized in that, The step of determining whether the user type of the user to be recommended is a cold-start user or a non-cold-start user based on the data completeness corresponding to each feature dimension includes: The data completeness of the user profile is determined based on the data completeness corresponding to each of the aforementioned feature dimensions; If the data completeness of the user profile is less than a preset data completeness threshold, the user type of the user to be recommended is determined to be a cold start user. If the data completeness of the user profile is greater than or equal to a preset data completeness threshold, the user type of the user to be recommended is determined to be a non-cold start user.
9. A computer device, characterized in that, The computer device includes a processor, a memory, and a computer program stored in the memory and executable by the processor, wherein when the computer program is executed by the processor, it implements the activity recommendation method as described in any one of claims 1 to 8.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the activity recommendation method as described in any one of claims 1 to 8.