Intelligent recommendation method and system based on big data mining and storage medium

Abstract

The invention discloses an intelligent recommendation method and system based on big data mining and a storage medium, and the method comprises the steps: carrying out the attribute division of obtained user source data, and obtaining a plurality of data sub-tables corresponding to a user basic attribute portrait, a business rule attribute portrait and a behavior attribute portrait; obtaining a portrait wide table according to each data sub-table, and obtaining a portrait label value of each user according to the portrait wide table and the crowd result table; obtaining a user number according to the portrait label value, and performing distribution analysis according to the user number to obtain a crowd portrait analysis result table; calculating according to the behavior attribute portrait to obtain a current behavior score and a historical behavior score, and further obtaining a user portrait score and writing time according to the current behavior score and the historical behavior score; and performing intelligent recommendation through the user portrait large model according to the crowd portrait analysis result table, the user portrait score and the writing time. The intelligent recommendation accuracy of the user portrait large model can be improved, and the method can be widely applied to the technical field of intelligent recommendation.

Description

Technical Field

[0001] The present application relates to the field of intelligent recommendation technology, and in particular to an intelligent recommendation method, system and storage medium based on big data mining. Background Art

[0002] In order to meet the needs of new network services and lightweight and dynamic computing in the future, the integration of network and computing has become a new development trend. The industry has proposed the concept of intelligent perception network (or simply intelligent perception network) to achieve intelligent perception and recommend products to users. The input of the model mainly includes model user portrait features, context features, target item features, and auxiliary network. However, the product recommendation of the model may be affected by external factors, resulting in non-global optimal solutions and inaccurate recommendations. Summary of the invention

[0003] In order to solve the above technical problems, the purpose of the present invention is to provide an intelligent recommendation method, system and storage medium based on big data mining with high accuracy of recommendation results.

[0004] To achieve the above objectives, an embodiment of the present application proposes an intelligent recommendation method based on big data mining, comprising the following steps:

[0005] Acquire user source data, divide the user source data by attributes, and obtain multiple data sub-tables corresponding to user basic attribute portraits, business rule attribute portraits, and behavior attribute portraits;

[0006] Obtaining a portrait wide table according to each of the data sub-tables, and then obtaining a portrait tag value of each user according to the portrait wide table and a preset population result table;

[0007] Obtain the number of users according to the portrait tag value, and then perform distribution analysis based on the number of users to obtain a crowd portrait analysis result table;

[0008] Calculate a current behavior score and a historical behavior score according to the behavior attribute portrait, and then obtain a user portrait score and a writing time according to the current behavior score and the historical behavior score;

[0009] Intelligent recommendations are made through a pre-built user portrait model based on the crowd portrait analysis result table, the user portrait score and the writing time.

[0010] In some embodiments, the image width table is obtained according to each of the data sub-tables, specifically:

[0011] The data sub-tables are summarized to obtain the portrait wide table.

[0012] In some embodiments, the crowd result table includes user identification numbers and crowd identification numbers, and obtaining the portrait tag value of each user according to the portrait width table and the preset crowd result table specifically includes:

[0013] Determine that the connection field is at least one of the user identification number or the group identification number;

[0014] Connecting the portrait wide table and the crowd result table through a JOIN statement according to the connection field to obtain a crowd information table;

[0015] The portrait label value corresponding to each of the users is obtained according to the crowd information table.

[0016] In some embodiments, obtaining the number of users according to the portrait tag value, and then performing distribution analysis according to the number of users to obtain a crowd portrait analysis result table specifically includes:

[0017] Performing aggregation calculation on each of the portrait tag values ​​to obtain the number of users corresponding to each of the portrait tag values;

[0018] Crowd portrait distribution analysis and index analysis are performed based on the number of users to obtain the crowd portrait analysis result table.

[0019] In some embodiments, the behavior attribute portrait includes a user identification number and a user behavior entity, and the current behavior score and the historical behavior score are calculated based on the behavior attribute portrait, specifically including:

[0020] Calculate a behavior score and a behavior attenuation factor according to the user behavior entity;

[0021] Obtaining the current behavior score according to the behavior score, the preset behavior weight and the behavior attenuation factor;

[0022] A short-term behavior profile calculation and a long-term behavior profile calculation are performed on the behavior attribute portrait according to the user identification number to obtain the historical behavior score.

[0023] In some embodiments, the short-term behavior profile calculation and the long-term behavior profile calculation are performed on the behavior attribute profile according to the user identification number to obtain the historical behavior score, specifically including:

[0024] Obtaining a first user historical behavior portrait and a corresponding first historical behavior score within a first preset time period according to the user identification number;

[0025] Acquire a second user historical behavior portrait and a corresponding second historical behavior score within a second preset time period according to the user identification number;

[0026] Attenuating the first historical behavior score according to the update time of the first user historical behavior portrait by using a preset attenuation function to obtain a first attenuated historical score;

[0027] The first historical behavior score is decayed according to the update time of the second user historical behavior portrait through the decay function to obtain a second decayed historical score.

[0028] In some embodiments, obtaining the user portrait score and the writing time according to the current behavior score and the historical behavior score is specifically as follows:

[0029] The current behavior score, the first decay history score, and the second decay history score are accumulated to obtain the user portrait score and the writing time.

[0030] To achieve the above purpose, another aspect of the embodiment of the present application proposes an intelligent recommendation system based on big data mining, including:

[0031] An attribute division module is used to obtain user source data, divide the user source data by attributes, and obtain multiple data sub-tables corresponding to user basic attribute portraits, business rule attribute portraits, and behavior attribute portraits;

[0032] A portrait tag value acquisition module, used to obtain a portrait wide table according to each of the data sub-tables, and then obtain a portrait tag value of each user according to the portrait wide table and a preset population result table;

[0033] A distribution analysis module is used to obtain the number of users according to the portrait tag value, and then perform distribution analysis according to the number of users to obtain a crowd portrait analysis result table;

[0034] A portrait score calculation module, used to calculate the current behavior score and the historical behavior score according to the behavior attribute portrait, and then obtain the user portrait score and writing time according to the current behavior score and the historical behavior score;

[0035] The intelligent recommendation module is used to make intelligent recommendations based on the crowd portrait analysis result table, the user portrait score and the writing time through a pre-built user portrait model.

[0036] To achieve the above-mentioned purpose, another aspect of an embodiment of the present application proposes an electronic device, which includes a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory. When the program is executed by the processor, the intelligent recommendation method based on big data mining as described above is realized.

[0037] To achieve the above-mentioned purpose, another aspect of an embodiment of the present application proposes a storage medium, which is a computer-readable storage medium used for computer-readable storage, and the storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the intelligent recommendation method based on big data mining as described above.

[0038] The beneficial effects of the present invention are as follows: the intelligent recommendation method, system and storage medium based on big data mining of the present invention first obtain user source data, perform attribute division on the user source data, and obtain multiple data sub-tables corresponding to user basic attribute portraits, business rule attribute portraits and behavior attribute portraits; secondly, obtain a portrait wide table according to each data sub-table, and then obtain the portrait label value of each user according to the portrait wide table and a preset crowd result table, obtain the number of users according to the portrait label value, and then perform distribution analysis according to the number of users to obtain a crowd portrait analysis result table; then, calculate the current behavior score and the historical behavior score according to the behavior attribute portrait, and then obtain the user portrait score and writing time according to the current behavior score and the historical behavior score; finally, perform intelligent recommendation according to the crowd portrait analysis result table, the user portrait score and the writing time through a pre-built user portrait large model. On the one hand, the present invention divides the attributes of user source data according to the portrait dimension, and then aggregates multiple data sub-tables corresponding to the user basic attribute portrait, business rule attribute portrait and behavior attribute portrait into a portrait wide table, which can integrate user source data of different dimensions to form a more comprehensive user portrait; on the other hand, the current behavior score and the historical behavior score are combined to update the new user portrait score and writing time, which can take into account the timeliness of user behavior and more accurately reflect the user's interests and preferences, thereby improving the accuracy of intelligent recommendations of the user portrait large model. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solution in the embodiments of the present invention, the following introduction is made to the drawings required for use in the embodiments of the present invention. It should be understood that the drawings introduced below are only for the convenience of clearly describing some embodiments of the technical solution of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0040] Figure 1 A flowchart of a step of an intelligent recommendation method based on big data mining provided by an embodiment of the present invention;

[0041] Figure 2 A flowchart of steps for obtaining a portrait tag value provided by an embodiment of the present invention;

[0042] Figure 3 A schematic diagram of a basic flow chart of distribution analysis provided by an embodiment of the present invention;

[0043] Figure 4 A step flow chart of step S103 provided in an embodiment of the present invention;

[0044] Figure 5 A calculation flow chart of crowd portrait distribution analysis and index analysis using gender and online time as examples provided in an embodiment of the present invention;

[0045] Figure 6 A flowchart of the steps of calculating the behavior score provided by an embodiment of the present invention;

[0046] Figure 7 A step flow chart of step S1043 provided in an embodiment of the present invention;

[0047] Figure 8 The overall framework diagram of the user portrait model provided by the embodiment of the present invention;

[0048] Fig. 9 A schematic diagram of a process flow of an intelligent recommendation method based on big data mining provided by an embodiment of the present invention;

[0049] Fig.10 A schematic diagram of the structure of an intelligent recommendation system based on big data mining provided by an embodiment of the present invention;

[0050] Fig.11 A schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is further described in detail below in conjunction with the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of the present application. They are only examples of devices and methods consistent with some aspects of the embodiments of the present application as detailed in the attached claims.

[0052] It is understood that the terms "first", "second", etc. used in this application can be used to describe various concepts in this article, but unless otherwise specified, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another concept. For example, without departing from the scope of the embodiment of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the words "if" and "if" as used herein can be interpreted as "at the time of" or "when" or "in response to determination".

[0053] The terms "at least one", "multiple", "each", "any", etc. used in this application, at least one includes one, two or more, multiple includes two or more, each refers to each of the corresponding multiple, and any refers to any one of the multiple.

[0054] Before describing the intelligent recommendation method based on big data mining of the present invention, it is necessary to clearly state that the method follows the privacy protection principle when processing user data. In particular, all user source data involved in the method, including but not limited to user data filled in when the user registers or collected through questionnaires, such as user age, education level, marital status, video viewing time, interactive operations, etc., are strictly obtained and used with the explicit authorization and consent of the user.

[0055] Before describing the embodiments of the present application in detail, some nouns and terms involved in the embodiments of the present application are first described. The nouns and terms involved in the embodiments of the present application are subject to the following explanations.

[0056] Intelligent recommendation: The essence of the intelligent recommendation system is an information publishing tool that has the characteristics of personalized distribution. It can automatically connect users with products, explore users' personalized needs, and help users find products that interest them.

[0057] Data wide table: Data wide table is a database table design characterized by many fields, which usually associates indicators, dimensions and attributes related to business topics. Because it aggregates different contents in one table, it can simplify the data analysis process, improve query efficiency, and reduce the complexity of table association. Data wide table does not conform to the database design specification of the third normal form because it may cause data redundancy, but this redundancy is beneficial in some cases because it simplifies iterative calculations in the data mining model training process.

[0058] In order to meet the needs of new network services and lightweight and dynamic computing in the future, the integration of network and computing has become a new development trend. The industry has proposed the concept of intelligent perception network (or simply intelligent perception network) to achieve intelligent perception and recommend products to users. The input of the model mainly includes model user portrait features, context features, target item features, and auxiliary network. However, the product recommendation of the model may be affected by external factors, resulting in non-global optimal solutions and inaccurate recommendations.

[0059] To this end, an embodiment of the present invention proposes an intelligent recommendation method based on big data mining. First, user source data is obtained, and attributes of the user source data are divided to obtain multiple data sub-tables corresponding to user basic attribute portraits, business rule attribute portraits, and behavior attribute portraits. Secondly, a portrait wide table is obtained according to each data sub-table, and then the portrait label value of each user is obtained according to the portrait wide table and a preset crowd result table, and the number of users is obtained according to the portrait label value, and then a distribution analysis is performed based on the number of users to obtain a crowd portrait analysis result table, and then the current behavior score and the historical behavior score are calculated according to the behavior attribute portrait, and then the user portrait score and writing time are obtained according to the current behavior score and the historical behavior score. Finally, a pre-built user portrait big model is used to perform intelligent recommendation based on the crowd portrait analysis result table, the user portrait score, and the writing time. On the one hand, the present invention divides the user source data into attributes according to the portrait dimension, and then aggregates multiple data sub-tables corresponding to the user basic attribute portrait, business rule attribute portrait, and behavior attribute portrait into a portrait wide table, which can integrate user source data of different dimensions to form a more comprehensive user portrait; on the other hand, the current behavior score and the historical behavior score are combined to update the new user portrait score and writing time, which can consider the timeliness of user behavior and more accurately reflect the user's interests and preferences, thereby improving the accuracy of intelligent recommendation of the user portrait large model. The intelligent recommendation method can be applied to e-commerce, social media, online education and other fields, but is not limited to this.

[0060] Reference Figure 1 , Figure 1 A flowchart of a method for intelligent recommendation based on big data mining provided by an embodiment of the present invention. An embodiment of the present invention proposes a method for intelligent recommendation based on big data mining, which includes steps S101 to S105:

[0061] S101, obtaining user source data, dividing the user source data by attributes, and obtaining multiple data sub-tables corresponding to user basic attribute portraits, business rule attribute portraits, and behavior attribute portraits;

[0062] Specifically, the user source data is divided into user basic attribute portraits (i.e. user basic static attribute portraits), business rule attribute portraits, and behavior attribute portraits according to the data content, and data sub-tables are generated and marked respectively. Among them, the behavior attribute portrait can be divided into short-term behavior attribute portraits and long-term behavior attribute portraits according to the real-time nature of the user behavior data.

[0063] In some optional embodiments, the user basic static attribute portrait includes basic information of the user, which needs to be filled in when the user registers or collected and recorded through questionnaires, etc. For example, the user's gender, age, education level, resident city, marital status, whether there are children, etc., as shown in Table 1 below. Because there are few and difficult ways to obtain this information, this attribute information is often missing, so it can be further predicted and supplemented through machine learning modeling.

[0064] Table 1

[0065]

[0066] Business rule attribute profiles can include user activity stratification in recommendation scenarios, and users are divided into new users, low-activity users, moderately-activity users, and heavily-activity users according to the number of short video contents effectively consumed by users on the platform. The specific classification rule examples are shown in Table 2 below. The definition of effective consumption rule refers to whether the user has effectively watched the video, which is determined by the viewing time. For example, if the user watches a video for more than 20 seconds, it is determined to be effectively watched.

[0067] Table 2

[0068]

[0069] User behavior attribute portrait is based on various behavior operation information generated by users in the process of using the product, including but not limited to user clicks, views (including duration), likes, searches, shares, comments, favorites, and follows, as well as some negative feedback user behaviors such as dislikes, reports, and quick swipes. These behavior information can explicitly or implicitly express user preferences for content. User behavior attribute portraits will also use these user behavior information to calculate the user's preference score for content entities. The specific classification rules are shown in Table 3 below.

[0070] It should be noted that a common practice in the current industry recommendation system is to retain the user's original behavior sequence information, such as the video viewing sequence, the video liking sequence, etc., and then the intelligent recommendation system processes each link as needed according to the model or strategy used. The embodiment of the present invention uniformly calculates the user's video viewing (duration), liking, collection, and sharing behaviors into a score for video preference, which can make the preferences of users with different behaviors comparable to a certain extent.

[0071] Table 3

[0072]

[0073] S102, obtaining a portrait wide table according to each data sub-table, and then obtaining a portrait tag value of each user according to the portrait wide table and a preset population result table;

[0074] In some optional embodiments, the three data sub-tables of the marked user basic static attribute portrait, business rule attribute portrait and behavior attribute portrait are first aggregated into a portrait wide table (i.e., data wide table), and then the portrait wide table is connected to the crowd result table to obtain the portrait tag value of each user. The embodiment of the present invention aggregates multiple data sub-tables corresponding to the user basic static attribute portrait, business rule attribute portrait and behavior attribute portrait into a portrait wide table, which can integrate user source data of different dimensions to form a more comprehensive user portrait;

[0075] As an optional implementation, the step of obtaining the portrait wide table according to each data sub-table can be further divided into the following steps S1021:

[0076] S1021. Summarize each data sub-table to obtain a portrait wide table.

[0077] For example, the UNION ALL function of SQL can be used to vertically merge the data of the three data sub-tables into a portrait wide table. Furthermore, during the merging process, the DISTINCT keyword can be used to remove duplicate rows.

[0078] Reference Figure 2 , Figure 2 A flowchart of steps for obtaining a portrait tag value provided in an embodiment of the present invention, further as an optional implementation manner, the crowd result table includes a user identification number and a crowd identification number, and the step of obtaining the portrait tag value of each user according to the portrait wide table and the preset crowd result table can be further divided into the following steps S1022 to S1024:

[0079] S1022, determining that the connection field is at least one of a user identification number or a group identification number;

[0080] S1023, connect the portrait wide table and the crowd result table through a JOIN statement according to the connection field to obtain a crowd information table;

[0081] In some optional embodiments, the crowd result table includes a user identification number and a crowd identification number, and the following table 4 is an example of a crowd result table. First, the connection field is determined to be the user identification number or the crowd identification number, and then the portrait wide table and the crowd result table are connected according to the determined connection field through the JOIN statement of SOL to obtain the crowd information table.

[0082] Table 4

[0083] Crowd identification number User identification number 100 1000 100 1001 101 1002 101 1000 101 1010

[0084] S1024. Obtain the portrait label value corresponding to each user according to the crowd information table.

[0085] Specifically, the portrait wide table contains all the portrait attributes of the user. By connecting the crowd result table and the portrait wide table, we can obtain the crowd to which each user belongs and all the portrait attributes corresponding to the user. Therefore, we can further assign corresponding portrait label values ​​to each user.

[0086] S103, obtaining the number of users according to the portrait tag value, and then performing distribution analysis according to the number of users to obtain a crowd portrait analysis result table;

[0087] Specifically, the embodiment of the present invention implements distribution analysis based on the number of users, which enables the user portrait model to understand the characteristics and distribution of the user group, thereby more accurately locating the target user. Figure 3 The figure shows the basic flow chart of distribution analysis. The crowd information table is stored in the data table demo_userprofile_crowd, where the calculate_labels field records the list of portrait labels that need to be analyzed in the form of a JSON array; the crowd portrait analysis result table is stored in the data table demo_userprofile_crowd_overview, and the overview_result field stores the portrait analysis results of the specified label, and the userprofile_date field records the data date used for the portrait analysis.

[0088] Reference Figure 4 , Figure 4 A step flow chart of step S103 provided in an embodiment of the present invention is further provided as an optional implementation mode. The step of obtaining the number of users according to the portrait tag value, and then performing distribution analysis according to the number of users to obtain a crowd portrait analysis result table can be further divided into the following steps S1031 and S1032:

[0089] S1031, performing aggregation calculation on each portrait tag value to obtain the number of users corresponding to each portrait tag value;

[0090] S1032. Perform crowd portrait distribution analysis and index analysis based on the number of users to obtain a crowd portrait analysis result table.

[0091] Specifically, the crowd distribution analysis and indicator analysis can be realized through the join operation between the crowd result table and the portrait wide table. After the crowd result table is connected to the portrait wide table, the portrait label value of each user can be obtained. According to different portrait label values, the corresponding number of users can be aggregated and calculated, and the distribution analysis can be realized based on the number of users. For example, a certain crowd has 1 million people, of which the number of male and female users is 600,000 and 400,000 respectively. Then the male and female distribution of the gender label of the crowd is 60% (60 / 100 and 40% (40 / 100). Indicator analysis can be calculated using aggregation functions. Aggregation functions can use average function AVG, sum function SUM, maximum (minimum) value function MAX (MIN), etc., but are not limited to this. Figure 5 Taking gender and online time as examples, the calculation process of population portrait distribution analysis and indicator analysis is demonstrated.

[0092] S104, calculating the current behavior score and the historical behavior score according to the behavior attribute portrait, and then obtaining the user portrait score and the writing time according to the current behavior score and the historical behavior score;

[0093] Specifically, the short-term behavior profile and the mid- to long-term behavior profile of the behavior attribute profile corresponding to the portrait wide table are calculated and the current behavior score is calculated. Finally, the current behavior score and the attenuated historical behavior score are accumulated and updated to the new user portrait score and writing time, thereby improving the accuracy of intelligent recommendation of the user portrait large model.

[0094] Reference Figure 6 , Figure 6 A flowchart of a behavior score calculation step provided in an embodiment of the present invention, further as an optional implementation mode, the behavior attribute profile includes a user identification number and a user behavior entity, and the step of calculating the current behavior score and the historical behavior score according to the behavior attribute profile can be further divided into the following steps S1041 to S1043:

[0095] S1041. Calculate a behavior score and a behavior attenuation factor according to the user behavior entity;

[0096] S1042, obtaining a current behavior score according to the behavior score, a preset behavior weight, and a behavior attenuation factor;

[0097] In some optional embodiments, the input of the behavior attribute portrait includes the user identification number and the user behavior entity corresponding to the user behavior. The user behavior entity includes the video identification number corresponding to the behavior, the behavior type, and the behavior operation time. If it is a viewing behavior, the viewing time information is also required. The current behavior score is calculated by the following formula:

[0098] score=actionScore×actionWeight×decayWeigh t

[0099] Among them, actionScore represents the behavior score, which is generally set to 1. Mainly for playback behavior, the actionScore score varies depending on the viewing time. For example, the calculation of the viewing behavior score can refer to actionScore = ln (viewing time / (video duration + 1)). decayWeight represents the behavior decay factor, which is calculated based on the time when the behavior occurs in the user behavior entity. actionWeight represents the behavior weight of different behaviors. The behavior weight can be set as a whole in combination with business data analysis, or it can be set individually according to the user's behavior preferences. For example, the behavior weight of display can be set to -2, the behavior weight of dislike can be set to -5, the behavior weight of like can be set to 5, the behavior weight of download can be set to 5, and so on.

[0100] S1043. Perform short-term behavior profile calculation and long-term behavior profile calculation on the behavior attribute profile according to the user identification number to obtain a historical behavior score.

[0101] Specifically, user basic attribute portraits and business rule attribute portraits can be directly obtained or extracted according to specific rule definitions. In the embodiment of the present invention, the calculation of behavior attribute portrait tags is emphasized. The behavior attribute portrait includes short-term behavior attribute portraits and long-term behavior attribute portraits, and the calculation logic of the two is roughly the same.

[0102] Reference Figure 7 , Figure 7 A step flow chart of step S1043 provided in an embodiment of the present invention is further provided as an optional implementation mode. The step of performing short-term behavior profile calculation and long-term behavior profile calculation on the behavior attribute profile according to the user identification number to obtain the historical behavior score can be further divided into the following steps S10431 to S10434:

[0103] S10431. Obtain a first user historical behavior portrait and a corresponding first historical behavior score within a first preset time period according to the user identification number;

[0104] S10432. Obtain a second user historical behavior portrait and a corresponding second historical behavior score within a second preset time period according to the user identification number;

[0105] It should be noted that the first preset time period and the second preset time period can be selected according to actual needs, and the first preset time period is greater than the second preset time period.

[0106] Specifically, the behavior attribute portrait includes the user identification number, so the user's long-term historical behavior portrait (i.e., the first user historical behavior portrait) and the corresponding first historical behavior score within the first preset time period are obtained according to the user identification number; the user's short-term historical behavior portrait (i.e., the second user historical behavior portrait) and the corresponding second historical behavior score within the second preset time period are obtained according to the user identification number. The embodiment of the present invention reflects the user's recent behavior and interests through short-term behavior portrait calculation, and understands the user's long-term behavior pattern and preference through long-term behavior portrait calculation, so that the user portrait model can more comprehensively understand the user's behavior characteristics.

[0107] The first user historical behavior portrait and the second user historical behavior portrait include a user identification number, a user behavior entity, a preference score, and an update time, etc. If it is a video preference, the user behavior entity corresponds to the video identification number; if it is a video tag preference, the user behavior entity corresponds to the tag content or the tag identification number.

[0108] S10433. Attenuate the first historical behavior score according to the update time of the first user's historical behavior portrait using a preset attenuation function to obtain a first attenuated historical score;

[0109] S10434. Attenuate the first historical behavior score according to the update time of the second user's historical behavior portrait through an attenuation function to obtain a second attenuated historical score.

[0110] Specifically, the first historical behavior score and the second historical behavior score are decayed according to the update time of the first user historical behavior portrait and the second user historical behavior portrait, respectively. The decay function can be constructed using the Newton cooling law mathematical model or the Ebbinghaus forgetting curve. Exemplarily, the decay function is shown in the following formula:

[0111] F(t)=F(t0)×e -λ×t

[0112] Wherein, F(t) represents the first decay history score or the second decay history score, F(t0) represents the initial first historical behavior score or the second historical behavior score, λ represents the self-set decay coefficient, and t represents the update time.

[0113] It should be noted that, considering the timeliness of user behavior, by attenuating the historical portrait score, it can be ensured that the user portrait can more accurately reflect the user's current behavior and interests, rather than relying too much on past data.

[0114] As an optional implementation, the step of obtaining the user portrait score and the writing time according to the current behavior score and the historical behavior score can be further divided into the following steps S1044:

[0115] S1044. Accumulate the current behavior score, the first decay history score, and the second decay history score to obtain a user portrait score and a writing time.

[0116] Specifically, the current behavior score, the first decay history score, and the second decay history score are finally accumulated and updated to a new user portrait score and writing time, thereby improving the accuracy of intelligent recommendations of the user portrait large model.

[0117] S105. Make intelligent recommendations based on the crowd portrait analysis result table, user portrait score and writing time using the pre-built user portrait model.

[0118] In some optional embodiments, such as Figure 8 The figure shows the overall framework of the user portrait model, which includes three levels: source data, portrait calculation, and portrait data storage. The source data layer accesses various user source data required for subsequent portrait calculations, such as user behavior data, user basic static data, third-party data, and short video content data; the portrait calculation layer calculates and outputs user portrait results based on the defined user portrait system; the portrait data storage layer eventually updates the calculation results to storage systems such as Redis and HBase for use by recommendation services.

[0119] Among them, user portraits and their scores can be used in all aspects of the user portrait model. In the recall stage, there are recalls based on user portraits, such as recalls based on static portraits such as user age, gender, and region, and recalls based on behavioral portraits. User portraits and their scores can also be used as features of models at each stage. In the re-ranking stage, different break-up strategies and diversity strategies can also be set based on the interest distribution of users in the portraits. User portraits and their scores are also important bases for analyzing user recommendation results and generating recommendation reasons.

[0120] In summary, the processing flow of the intelligent recommendation method based on big data mining in the embodiment of the present invention is as follows: Fig. 9 As shown:

[0121] The first step is to divide the user basic attribute portrait, business rule attribute portrait, and behavior attribute portrait according to the portrait dimension, and generate data sub-tables for marking respectively;

[0122] The second step is to aggregate the three data tables of the marked user basic attribute portrait, business rule attribute portrait, and behavior attribute portrait into a portrait wide table;

[0123] The third step is to connect the portrait wide table and the crowd result table to obtain the portrait label value of each user;

[0124] The fourth step is to aggregate and calculate the corresponding number of users based on different portrait tag values, implement distribution analysis based on the number of users, and obtain a crowd portrait analysis result table;

[0125] Step 5: Calculate the short-term behavior profile and the mid- to long-term behavior profile for the behavior attribute profile corresponding to the portrait wide table, obtain the historical behavior score, and calculate the current behavior score;

[0126] Step 6: Accumulate the current behavior score and the decayed historical behavior score to update the new user profile score and writing time;

[0127] The seventh step is to make intelligent recommendations based on the pre-built user portrait model according to the crowd portrait analysis result table, user portrait score and writing time.

[0128] The above describes the intelligent recommendation method based on big data mining in the embodiment of the present invention. It can be appreciated that compared with the intelligent recommendation method in the prior art, the embodiment of the present invention has the following advantages:

[0129] 1. Divide the user source data by attributes according to the portrait dimension, and then aggregate multiple data sub-tables corresponding to the user basic attribute portrait, business rule attribute portrait, and behavior attribute portrait into a portrait wide table, which can integrate user source data of different dimensions to form a more comprehensive user portrait;

[0130] Second, the corresponding number of users is calculated based on the aggregation of different portrait tag values, and then distribution analysis is performed based on the number of users. This enables the user portrait model to understand the characteristics and distribution of the user group, thereby more accurately locating the target users;

[0131] 3. Combining the current behavior score and the historical behavior score to update the new user portrait score and writing time can take into account the timeliness of user behavior and more accurately reflect the user's interests and preferences, thereby improving the accuracy of intelligent recommendations of the user portrait model.

[0132] Reference Fig.10 The embodiment of the present invention further provides an intelligent recommendation system based on big data mining, comprising:

[0133] The attribute division module is used to obtain user source data, divide the user source data by attributes, and obtain multiple data sub-tables corresponding to user basic attribute portraits, business rule attribute portraits, and behavior attribute portraits;

[0134] The portrait tag value acquisition module is used to obtain the portrait wide table according to each data sub-table, and then obtain the portrait tag value of each user according to the portrait wide table and the preset population result table;

[0135] The distribution analysis module is used to obtain the number of users based on the portrait tag value, and then perform distribution analysis based on the number of users to obtain a crowd portrait analysis result table;

[0136] A portrait score calculation module is used to calculate the current behavior score and the historical behavior score based on the behavior attribute portrait, and then obtain the user portrait score and writing time based on the current behavior score and the historical behavior score;

[0137] The intelligent recommendation module is used to make intelligent recommendations based on the crowd portrait analysis result table, user portrait score and writing time through the pre-built user portrait model.

[0138] The contents of the above-mentioned intelligent recommendation method embodiment based on big data mining are all applicable to the present intelligent recommendation system embodiment based on big data mining. The functions specifically implemented by the present intelligent recommendation system embodiment based on big data mining are the same as those of the above-mentioned intelligent recommendation method embodiment based on big data mining, and the beneficial effects achieved are also the same as those achieved by the above-mentioned intelligent recommendation method embodiment based on big data mining.

[0139] The embodiment of the present invention also provides an electronic device, the electronic device comprising: a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory, and the program is executed by the processor to realize the above-mentioned intelligent recommendation method based on big data mining. The electronic device can be any intelligent terminal including a tablet computer, a car computer, etc.

[0140] like Fig.11 FIG. 1 is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present invention, referring to FIG. Fig.11 , an embodiment of the present invention provides an electronic device, including:

[0141] The processor 1001 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present invention.

[0142] The memory 1002 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 1002 can store an operating system and other application programs. When the technical solution provided in the embodiments of this specification is implemented by software or firmware, the relevant program code is stored in the memory 1002, and the processor 1001 calls and executes the intelligent recommendation method based on big data mining in the embodiments of the present invention;

[0143] Input / output interface 1003, used to implement information input and output;

[0144] The communication interface 1004 is used to realize the communication interaction between the device and other devices. The communication can be realized through a wired manner (such as USB, network cable, etc.) or a wireless manner (such as mobile network, WIFI, Bluetooth, etc.);

[0145] A bus 1005 , which transmits information between various components of the device (e.g., the processor 1001 , the memory 1002 , the input / output interface 1003 , and the communication interface 1004 );

[0146] The processor 1001 , the memory 1002 , the input / output interface 1003 and the communication interface 1004 are connected to each other in communication within the device via the bus 1005 .

[0147] An embodiment of the present invention also provides a storage medium, which is a computer-readable storage medium used for computer-readable storage. The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the above-mentioned intelligent recommendation method based on big data mining.

[0148] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0149] The embodiment of the present invention also discloses a computer program product or a computer program, wherein the computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes Figure 1 The method shown.

[0150] In some optional embodiments, the function / operation mentioned in the block diagram may not occur in the order mentioned in the operation diagram. For example, depending on the function / operation involved, the two boxes shown in succession can actually be executed substantially simultaneously or the above-mentioned boxes can sometimes be executed in reverse order. In addition, the embodiment presented and described in the flow chart of the present invention is provided by way of example, for the purpose of providing a more comprehensive understanding of technology. The disclosed method is not limited to the operation and logic flow presented herein. The optional embodiment is expected, wherein the order of various operations is changed and the sub-operation described as a part of a larger operation is performed independently.

[0151] In addition, although the present invention is described in the context of functional modules, it should be understood that, unless otherwise specified to the contrary, one or more of the above-mentioned functions and / or features can be integrated into a single physical device and / or software module, or one or more functions and / or features can be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the present invention. More specifically, in view of the properties, functions and internal relationships of the various functional modules in the device disclosed herein, the actual implementation of the module will be understood within the conventional skills of the engineer. Therefore, those skilled in the art can implement the present invention set forth in the claims without excessive experimentation using ordinary techniques. It is also understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the present invention, which is determined by the full scope of the appended claims and their equivalents.

[0152] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the above methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0153] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0154] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and editable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the above-mentioned program is printed, since the above-mentioned program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or processing in other suitable ways as necessary, and then stored in a computer memory.

[0155] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0156] In the above description of this specification, the description with reference to the terms "one embodiment / example", "another embodiment / example" or "certain embodiments / examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0157] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

[0158] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the embodiments. Those skilled in the art may make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.

Claims

1. An intelligent recommendation method based on big data mining, characterized in that: The following steps are involved: Acquire user source data, divide the user source data by attributes, and obtain multiple data sub-tables corresponding to user basic attribute portraits, business rule attribute portraits, and behavior attribute portraits; Obtaining a portrait wide table according to each of the data sub-tables, and then obtaining a portrait tag value of each user according to the portrait wide table and a preset population result table; Obtain the number of users according to the portrait tag value, and then perform distribution analysis based on the number of users to obtain a crowd portrait analysis result table; Calculate a current behavior score and a historical behavior score according to the behavior attribute portrait, and then obtain a user portrait score and a writing time according to the current behavior score and the historical behavior score; Intelligent recommendations are made through a pre-built user portrait model based on the crowd portrait analysis result table, the user portrait score and the writing time.

2. The intelligent recommendation method based on big data mining according to claim 1, characterized in that: The image width table is obtained according to each of the data sub-tables, specifically: The data sub-tables are summarized to obtain the portrait wide table.

3. The intelligent recommendation method based on big data mining according to claim 1, characterized in that: The crowd result table includes user identification numbers and crowd identification numbers. The portrait tag value of each user is obtained according to the portrait wide table and the preset crowd result table, specifically including: Determine that the connection field is at least one of the user identification number or the group identification number; Connecting the portrait wide table and the crowd result table through a JOIN statement according to the connection field to obtain a crowd information table; The portrait label value corresponding to each of the users is obtained according to the crowd information table.

4. The intelligent recommendation method based on big data mining according to claim 1, characterized in that: The number of users is obtained according to the portrait tag value, and then distribution analysis is performed according to the number of users to obtain a crowd portrait analysis result table, which specifically includes: Performing aggregation calculation on each of the portrait tag values ​​to obtain the number of users corresponding to each of the portrait tag values; Crowd portrait distribution analysis and index analysis are performed based on the number of users to obtain the crowd portrait analysis result table.

5. The intelligent recommendation method based on big data mining according to claim 1, characterized in that: The behavior attribute portrait includes a user identification number and a user behavior entity, and the current behavior score and the historical behavior score are calculated based on the behavior attribute portrait, specifically including: Calculate a behavior score and a behavior attenuation factor according to the user behavior entity; Obtaining the current behavior score according to the behavior score, the preset behavior weight and the behavior attenuation factor; A short-term behavior profile calculation and a long-term behavior profile calculation are performed on the behavior attribute portrait according to the user identification number to obtain the historical behavior score.

6. The intelligent recommendation method based on big data mining according to claim 5, characterized in that: The performing of short-term behavior portrait calculation and long-term behavior portrait calculation on the behavior attribute portrait according to the user identification number to obtain the historical behavior score specifically includes: Acquire a first user historical behavior portrait and a corresponding first historical behavior score within a first preset time period according to the user identification number; Acquire a second user historical behavior portrait and a corresponding second historical behavior score within a second preset time period according to the user identification number; Attenuating the first historical behavior score according to the update time of the first user historical behavior portrait by using a preset attenuation function to obtain a first attenuated historical score; The first historical behavior score is decayed according to the update time of the second user historical behavior portrait through the decay function to obtain a second decayed historical score.

7. The intelligent recommendation method based on big data mining according to claim 6, characterized in that: The user portrait score and writing time are obtained according to the current behavior score and the historical behavior score, specifically: The current behavior score, the first decay history score, and the second decay history score are accumulated to obtain the user portrait score and the writing time.

8. An intelligent recommendation system based on big data mining, characterized in that: include: An attribute division module is used to obtain user source data, divide the user source data by attributes, and obtain multiple data sub-tables corresponding to user basic attribute portraits, business rule attribute portraits, and behavior attribute portraits; A portrait tag value acquisition module, used to obtain a portrait wide table according to each of the data sub-tables, and then obtain a portrait tag value of each user according to the portrait wide table and a preset population result table; A distribution analysis module is used to obtain the number of users according to the portrait tag value, and then perform distribution analysis according to the number of users to obtain a crowd portrait analysis result table; A portrait score calculation module, used to calculate the current behavior score and the historical behavior score according to the behavior attribute portrait, and then obtain the user portrait score and writing time according to the current behavior score and the historical behavior score; The intelligent recommendation module is used to make intelligent recommendations based on the crowd portrait analysis result table, the user portrait score and the writing time through a pre-built user portrait model.

9. An electronic device, characterized in that: The electronic device includes a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory. When the program is executed by the processor, the steps of the intelligent recommendation method based on big data mining as described in any one of claims 1 to 7 are realized.

10. A storage medium, the storage medium being a computer-readable storage medium, used for computer-readable storage, characterized in that: The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of the intelligent recommendation method based on big data mining as described in any one of claims 1 to 7.

Images