CN116501957A

CN116501957A - User tag portrait processing method, user portrait system, apparatus and storage medium

Info

Publication number: CN116501957A
Application number: CN202310352131.5A
Authority: CN
Inventors: 吕利峰
Original assignee: Shenzhen Rabbit Exhibition Intelligent Technology Co ltd
Current assignee: Shenzhen Rabbit Exhibition Intelligent Technology Co ltd
Priority date: 2023-03-28
Filing date: 2023-03-28
Publication date: 2023-07-28

Abstract

The application relates to the technical field of artificial intelligence and discloses a user tag portrait processing method, a user portrait system, equipment and a storage medium so as to construct a more comprehensive user portrait. The method comprises the following steps: integrating the collected first user data and second user data to obtain target user data, wherein the target user data comprises user attribute data, user association relationship data, user behavior data, user value data, user risk evaluation data and user marketing data; carrying out classification statistics on the user attribute data, and constructing a first type tag set of the user by using a classification statistics result; performing feature analysis on the user behavior data, and constructing a second type tag set of the user according to a feature analysis result; performing mining analysis according to the target user data to construct a third type tag set of the user; and carrying out fusion processing on the first type tag set, the second type tag set and the third type tag set to obtain a fused tag portrait result of the user.

Description

User tag portrait processing method, user portrait system, apparatus and storage medium

Technical Field

The application relates to the technical field of artificial intelligence, in particular to a user tag portrait processing method, a user portrait system, equipment and a storage medium.

Background

User tag portraits are user information collected in various ways, and the user is depicted and drawn in a data and tag manner. On one hand, the enterprise is helped to know and know the clients served by the enterprise more deeply, and the clients are served better through refined operation; on the other hand, the product is optimized and upgraded by analyzing the user portrait.

In the traditional label portrait processing, the user data is built only according to the basic attribute data of the user, so that the label is not comprehensive and perfect enough, the label types are not abundant enough, the built label portrait is not comprehensive enough, and the label portrait built according to the mode cannot be effectively processed.

Disclosure of Invention

The application provides a user tag portrait processing method, a user portrait system, equipment and a storage medium, which are used for solving the technical problems that a constructed tag portrait is not comprehensive enough and cannot be effectively processed.

In order to solve the technical problems, the technical proposal is as follows:

A user tag portrait processing method is provided, the method includes:

collecting first user data from a target service system and second user data from a third party platform system;

integrating the first user data and the second user data to obtain target user data, wherein the target user data comprises user attribute data, user association relationship data, user behavior data, user value data, user risk evaluation data and user marketing data;

carrying out classification statistics on the user attribute data, and constructing a first type tag set of the user by using a classification statistics result;

performing feature analysis on the user behavior data, and constructing a second type tag set of the user according to a feature analysis result;

performing mining analysis according to the target user data to construct a third type tag set of the user;

and carrying out fusion processing on the first type tag set, the second type tag set and the third type tag set to obtain a fusion tag portrait result of the user.

Optionally, the user attribute data includes population attribute information, life information, location information and custom tag information of various custom attributes, where the custom tag information is reported by guiding the user through a guiding page;

The user behavior data includes financial product preference information, non-financial product preference information, internal channel preference information of the user at the target business system, and external channel preference information at the third party platform system.

Optionally, the user association relationship data includes life association relationship information, financial association relationship information and social network association relationship information;

the user value data comprises user own value information and contribution value information of the user;

the risk evaluation data comprises blacklist information and risk evaluation information of a plurality of different dimensions for the user;

the user marketing data comprises financial demand information and non-financial demand information of the user in a preset period and marketing activity information of the user aiming at activity pages provided by the target business system.

Optionally, the contribution value information includes closeness and support of the user to the business page activities provided by the target business system, and the marketing campaign information includes loyalty, satisfaction, acceptance and liveness of the user to the activity page product services provided by the target business system.

Optionally, the performing feature analysis on the user behavior data, and constructing a second type tag set of the user according to a feature analysis result, including:

performing feature statistical analysis on the user behavior data to count various behavior features of the user;

respectively matching the various behavior characteristics with predefined labels to obtain behavior labels with the matched various behavior characteristics;

and constructing a second type tag set of the user according to the behavior tags matched with the behavior features.

Optionally, the data analysis is performed according to the target user data to construct a third type tag set of the user, so as to construct the third type tag set of the user, including:

mining the relevance among all the classified data in the target user data through a preset algorithm so as to mine out the potential relevance information of the user;

correspondingly generating an analysis tag of the user according to the relevance information;

and constructing a third type tag set of the user according to different analysis tags of the user.

Optionally, the mining, by a preset algorithm, the relevance between the classified data in the target user data to mine the potential relevance information of the user includes:

Classifying and marking the target user data by using a preset classification algorithm to obtain a plurality of data classification results, wherein the classification algorithm comprises a decision tree algorithm, a clustering algorithm, a naive Bayes algorithm and a support vector machine algorithm;

combining the data classification results to obtain a combined data classification result;

and mining each frequent item set and association rule in the merged data classification result by using an association rule algorithm to obtain the association information.

A user tag portrait system is provided, which includes a target service system, a third party platform system and a user tag portrait processing system, where the user tag portrait processing system is used to implement a user tag portrait processing method as described in any one of the foregoing.

There is provided a computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of any one of the user tag portrait processing methods described above when the computer program is executed.

There is provided a computer readable storage medium storing a computer program which when executed by a processor performs the steps of a user tag portrait processing method as described in any one of the preceding claims.

In the technical scheme provided by the method, besides collecting user data from a self service system, the user data can be collected from any third party platform and integrated, the data is wider in related range, more in sources and finer in data granularity, the data can be guaranteed to be more comprehensive, perfected and more effective, the defects of single traditional data collection mode, deficient data types and the like are overcome, multiple different types of labels of a user are built based on the searched data, multiple different labels are covered by each type of label, the types of labels are enriched, the types of the labels are rich and varied, the problem that the labels are single in type due to the fact that the labels are built only by means of attributes is solved, the application range is wider, finally, fusion processing is carried out on the labels of different label sets to form final fusion label portrait results, the constructed fusion label portrait results are more comprehensive and accurate, more accurate processing can be effectively realized according to the fusion label portrait results, and thousands of people are really implemented effectively.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the description of the embodiments of the present application will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a system diagram of a user portrayal system in an embodiment of the present application;

FIG. 2 is a flow chart of a user tag portrait processing method according to an embodiment of the present application;

FIG. 3 is a schematic flow chart of a user tag image processing method according to an embodiment of the present application;

FIG. 4 is a schematic diagram of a user tag image processing device according to an embodiment of the present application;

fig. 5 is a schematic structural diagram of a computer device according to an embodiment of the present application.

Detailed Description

The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all, of the embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.

The embodiment of the application provides a user tag portrait processing scheme, which is used for realizing comprehensive user tag portrait processing so as to improve the accuracy of subsequent processing including user marketing and the like, and the contents of the aspects are respectively described through various embodiments.

First, a description will be given of a user portrait system provided in an embodiment of the present application. Referring to fig. 1, fig. 1 is a schematic system diagram of a user portrait system provided in an embodiment of the present application, where the user portrait system includes a user tag portrait processing system, a target service system and a third party platform system, where the target service system refers to a service system provided by a service company, and by way of example, the target service system may refer to an H5 page service system or other service systems that provide active pages, and the third party platform system includes any one or more third party systems that provide different service services, and the user tag portrait processing system may collect required data from the target service system and the third party platform system to implement the user tag portrait processing method provided in the embodiment of the present application. The terminal device is operated with a user tag portrait processing system, and the terminal device comprises, but is not limited to, various personal computers, notebook computers, smart phones and tablet computers. The following describes in detail the various embodiments.

As shown in fig. 2 and 3, there is provided a user tag portrait processing method, including:

s10: first user data is collected from the target business system and second user data is collected from the third party platform system.

S20: and integrating the first user data and the second user data to obtain target user data, wherein the target user data comprises user attribute data, user association relationship data, user behavior data, user value data, user risk evaluation data and user marketing data.

In the embodiment, in order to enable the user tag portrait obtained later to be more complete and detailed, more and more perfect data are needed to support, and target user data are needed to be obtained from a plurality of channels and integrated, and the target user data comprise user attribute data, user association relationship data, user behavior data, user value data, user risk evaluation data and user marketing data.

In order to comprehensively and perfectly acquire the target user data, besides collecting the user data from the target service system, the user data are collected from the third party platform system and respectively recorded as first user data and second user data, as shown in fig. 2, the first user data can be obtained by analyzing service data provided by the target service system, the second user data can be obtained by docking the third platform system through an SDK interface, and it is to be noted that the first user data and the second user data are not necessarily identical and have wider coverage range, so that the above-mentioned collection mode can ensure the comprehensiveness of the data.

The user attribute data is related data representing user attributes, the user association relationship data is related data representing user association relationship, the user behavior data is related data representing user behavior, the user value data is related data representing user value, the user risk evaluation data is related data representing user analysis, and the user marketing data is data representing user marketing conditions.

The target service system may refer to an H5 page service system in the company or other service systems for providing active pages, for example, accessing form data such as related public numbers and applets authorized by a certain work, participating in a certain activity, drawing a lottery, submitting a mobile phone number, and the like.

The third party platform system comprises any one or more other third party systems which are different in business service relative to the target business system of the company. As an example, the target service system may dock other third party platform systems through the SDK, open the target service system and the third party platform system, and periodically push the uploaded target service system of the third party platform or the third party platform system to a data bin of the target service system to obtain freshness and perfection degree of the guaranteed data, where the data of the target service system and the third party platform may be directly collected from the data bin of the target service system; as another example, the target service system and the third party platform system may be collected separately, and the method is not limited in particular.

After the first user data and the second user data are collected, the first user data and the second user data are integrated, and target user data are obtained.

S30: and carrying out classification statistics on the user attribute data, and constructing a first type tag set of the user by using the classification statistics result.

S40: and carrying out feature analysis on the user behavior data, and constructing a second type tag set of the user according to a feature analysis result.

S50: and carrying out data analysis according to the target user data to construct a third type tag set of the user.

S60: and carrying out fusion processing on the first type tag set, the second type tag set and the third type tag set to obtain a fused tag portrait result of the user.

The first type tag set comprises attribute tags corresponding to various attributes of a user. The second type tag set comprises behavior tags respectively corresponding to various user behavior features. The third type tag set comprises a plurality of analysis tags corresponding to the data analysis results respectively.

In this embodiment, after the target user data is obtained, marking is performed by combining the target user data in different manners, so as to obtain different types of tag data sets. And for the user attribute data, directly carrying out classification statistics on the user attribute data, wherein the classification statistics result comprises attribute types of various users, and then constructing a first type tag set of the users by utilizing the classification statistics result, wherein the first type tag set covers attribute tags corresponding to various attributes of the user attribute data. For example, the statistical index tag includes attributes such as gender, age, occupation, etc. of the user, and the statistical index tag includes attribute tag classification of the user, and the first final tag includes event types of the first or last completion of the user, that is, the attribute tag is classified based on the user attribute, and as an example, attribute tag classification may be performed using manually defined rules, for example, classification may be performed according to the attributes such as the age, gender, etc. of the user, and the user may be labeled with the corresponding attribute tag.

And for the user behavior data, carrying out feature analysis on the user behavior data, and constructing a second type tag set of the user according to the feature analysis result, wherein the second type tag set covers behavior tags corresponding to various user behavior features. Illustratively, the user is behavior-tagged with a classification of behavior, e.g., by clicking, purchasing, searching, etc., of the user. The behavior label can be obtained by recording behavior data of the user, counting behavior characteristics of the user and matching the behavior characteristics with a predefined label. For example, purchase preference tags may be tagged based on user purchase records, browse preference tags may be tagged based on user browse records, and so on.

It should be noted that, in the embodiment of the present application, data analysis is further performed according to the target user data to construct a third type tag set of the user, that is, data mining analysis is performed on the user attribute data, the user association relationship data, the user behavior data, the user value data, the user risk evaluation data and the user marketing data, another type tag set is constructed according to the data analysis result, and is recorded as the third type tag set, where the analysis tag reflects a tag corresponding to an analysis result characteristic obtained based on the analysis result of the multiple data, and the analysis tag includes a tag class such as a personalized recommendation result tag, a crowd analysis result tag, a user subdivision tag, and a content preference tag, and may also include a tag such as a calculated loyalty of the user, which is not particularly limited.

In addition, in a specific implementation, the attribute tag, the behavior tag and the analysis tag may be preset tag templates, and the user may directly use the preset tag templates through simple configuration, for example, configure a tag corresponding to a specific action (browse or participate in or draw a lottery, or browse times are greater than a fixed value), where the configured tag templates may include custom tags, and the custom tags may enable the user to define specific rules of the tags, and the custom tags may include attribute tags or behavior tags.

In an embodiment, a corresponding tag value may be configured for each tag in the tag set, where the tag value is a specific result of a tag rule, for example, when a tag is customized, a different tag value may be set for a certain condition of the rule when the tag rule is set. The tag value is a specific ranking of the corresponding tag that is finer than the granularity of the tag type. For example, the label "purchased a merchandise" is marked with the rule that the user purchased a merchandise. The user tag value with a number of purchased a commodity of more than 5 in one year is defined as "a commodity-light user"; the user tag value with a commodity purchased for a year of more than 20 is defined as "a commodity-heavy user"; for another example, if there is a member tag, there may be member tag values of primary members, advanced members, etc. to distinguish the member levels by different tag values, so that the pushing or touching accuracy is higher, such as issuing coupons with different amounts. At least one tag value may be set when setting a specific behavior tag, an attribute tag, and an analysis tag.

In the embodiment, by generating the label value with finer granularity for each label in the label set, the concept of the label value is newly added, and the data under different labels are graded with finer granularity, so that the comprehensiveness and richness of the subsequent label portrait are further improved.

And finally, carrying out fusion processing on the first type tag set, the second type tag set and the third type tag set to obtain a fused tag portrait result of the user.

In this embodiment, it may be seen that, in addition to collecting user data from a self service system, user data is collected from any third party platform and integrated, the data is wider in range, more in source and finer in data granularity, so that the data can be guaranteed to be more comprehensive, more perfect and more effective, the defects of single data collection mode, deficient data types and the like in the traditional data collection mode are overcome, then, a plurality of different types of labels of users are constructed based on the searched and taken data, and each type of label of the same type covers a plurality of different labels, so that the types of labels are enriched and diversified, the problem that the types of labels are single due to construction of the labels only by means of attributes is solved, the application range is wider, and finally, the fusion processing is performed on the labels of different label sets to form a final fusion label portrait result, so that the constructed fusion label portrait result is more comprehensive and accurate, the more accurate processing can be effectively realized according to the fusion label portrait result, and the effective implementation of thousands of people is really achieved.

In an embodiment, in step S60, that is, the multiple tag sets are fused and integrated, various algorithms and models may be used, such as weight-based fusion, cluster-based integration, etc., which are exemplified herein by the fusion of the given weights. Assume that one user a has the following types of tags:

first type tag set: age, sex; second type tag set: commodity purchase records, commodity browsing records and commodity browsing classifications; third type tag set: purchasing preferences, consumption levels; each tag is configured with a corresponding tag value as follows:

first type tag set:

age: 18-20 (0.2)

Gender: man (0.1)

Second type tag set:

commodity purchase record: shoe (0.1)

And (5) commodity browsing records: shoe (0.2)

Third type tag set:

purchase preferences: certain brand (0.3)

Consumption grade: medium (0.1)

In the above example, the relative importance of different types of labels in the user portrait is reflected by weighting them, in some schemes, similar labels can be combined, for example, the "commodity browsing classification" and the "commodity browsing record" are combined into one label, so as to better reflect the preference and interest of the user, and through the weighting and integration processing, a comprehensive user portrait, that is, a fused label portrait result, is finally obtained, and can be used for personalized recommendation, advertisement delivery and other business scenes. It should be noted that different business scenarios may require different labels and weight assignments, so the construction of the fused label portrait result may also be adjusted and optimized according to specific business requirements.

In an embodiment, as shown in fig. 3, in addition to collecting data from the target service system and the third party platform system, the imported data imported by the user's active data may be used as third user data, and then the first user data, the second user data and the third user data may be integrated to obtain the target user data, which is not limited specifically. In the embodiment, the data channel is further widened, and the data comprehensiveness is improved.

It should be noted that, in an embodiment, in step S20, the first user data and the second user data are integrated to obtain the target user data, and the integration process includes data preprocessing, noise removal enhancement, cleaning, duplication removal, standardization processing, and data identifier mapping processing after forming the data. Wherein, each treatment process is as follows:

pretreatment: a Flink is a framework and distributed processing engine that processes streaming data. In this embodiment of the present application, the flow executionenvironment and the DataStream API are provided by using the link to read and process the stream data of the target service system and the third party platform system, that is, the first user data and the second user data, and the DataStream.

Noise removal and enhancement: the stream data read out from the target service system and the third party platform system may also be subjected to noise reduction and signal enhancement processing using a filter function (filter function) provided by the link. For example, assuming that a data stream is read and contains noise points (such as abnormal values of certain parameters), in order to effectively process the data, the embodiment needs to filter the noise points, and in order to accurately and quickly extract the noise points of the data, the embodiment uses a filter function provided by the Flink to filter the data stream, remove abnormal points which do not conform to the data characteristics or do not conform to the normal rules, and keep the data points which conform to the data characteristics and rules, thereby improving the value and reliability of the stream data provided by the target service system and the third party platform system.

Cleaning: the DataStream API provided by Apache link can be used to flush and remove redundant information from the stream data, such as removing useless fields, deleting duplicate records, unifying different identifiers, etc. For example, a filter function may be used to remove useless log data.

And (5) de-duplication: the data stream API provided by Apache Flink can be used for carrying out de-duplication processing on stream data, so that the influence on analysis results caused by repeated data is avoided, the workload of subsequent data analysis is reduced, and the efficiency is improved. For example, the stream data may be grouped according to a specified key value (such as uuid for identifying one piece of data) using a key by () method in Apache link, and then de-duplicated using a discrete () method in Apache link.

And (3) standardization treatment: the data stream API provided by Apache link may be used to normalize the stream data so that the data is comparable and consistent, for example, by normalizing the values to within a certain range. For example, mapfield in Apache link may be used to map values into a specified range, some raw data values are too large to fit for subsequent reading comparisons, etc., and the embodiment may map values into the appropriate range by the normalization process.

Mapping: after the stream data provided by the target service system and the third party platform system are processed through the steps, the first user data and the second user data can be subjected to uid matching in an ID-Mapping mode. That is, the target service system identifies a user by using a user identifier uid, where the uid is unique, but the data collected or imported from the third party platform system carries user data of the third party platform system, such as a mobile phone number, and in this embodiment, ID-Mapping is used to make an association with the user data of the different systems, and the user is uniformly mapped to a unique user identifier uid to identify the user, where operations such as labeling, displaying a label, querying details and the like are all performed on the data for the uid. Thus, the target user data identification is formed with a unique user identification uid.

In this embodiment, in order to improve the data processing efficiency and reduce the influence of the data processing result on the generation and analysis of the subsequent label, the Apache link tool may be used to integrate the first user data and the second user data to obtain the target user data with the user identifier uid identified.

In an embodiment, the user attribute data includes population attribute information, life information, location information and custom tag information of different custom attributes, where the custom tag information may be reported by a target service system or a target service system guiding a user through a guiding page; exemplary, such as demographic attribute information, including but not limited to gender, age, region, and contact information, may be obtained by a third party interfacing with the user, even in communication software; such as life information, including but not limited to water, electricity, and natural gas usage information, etc., may be obtained by interfacing with a public service website; such as location information, including but not limited to home, unit address, radius of half life, daily taxi path and airline flight records, may be obtained by interfacing with third party travel software; such as custom tag information, including but not limited to custom tags such as white collar and high income groups, may guide the user to actively report through the SDK interface of the h5 work channel provided by the target business system. In the embodiment, the user attribute data ensures the pertinence of the basic attribute, so that the marketing capability can be enhanced, the preliminary and accurate circle selection of the user is facilitated, and the user-defined label information is introduced, so that the accuracy can be further enhanced due to the fact that the label information is user-defined.

In one embodiment, the user behavior data includes financial product preference information, non-financial product preference information, internal channel preference information of the user at the target business system, and external channel preference information. Illustratively, the internal channel preference information, for example, the frequency and habit of use, the type of service, etc. of a certain service including the target service system of the company, may be obtained through the SDK interface of the h5 work channel of the target service system; external channel preference information, such as full-network channel preference, including surfing time, surfing habit, etc.; available through interfacing with a third party platform system. Financial product preference information, such as insurance and fund product preferences, including financial product specific types and financial product holding amounts, etc., is available to the third party insurance platform; non-financial product information preferences, such as user interests, including sports and reading interests, such as like golf and regular financial news, may be available to other non-financial property related corporate platforms. In the embodiment, channel optimization can be realized for the user by defining various user behavior data types, and the pertinence and the comprehensiveness of the channel are optimized.

In an embodiment, the user association data includes life association information, financial association information, social network association information; for example, life association relationship information including but not limited to family relationships, colleague relationships, friend relationships and community life circles can be led to users for active reporting through an SDK interface of the h5 work channel; such as financial association information, including business associations, fund associations, employment relationships, marketing relationships, and vouch-for relationships of the user at a row, may be obtained by interfacing a data interface; compared with the social network association relationship, the social network association relationship comprises the number of vermicelli, whether v authentication is passed through some social platforms, even if the communication applies to the circle of friends and the influence capability of the social network (which can be embodied as the number of praise or comment of the user and the forwarding quantity), the social network association relationship can be obtained by interfacing with a third party development data interface, such as interfacing with a short video platform and a video website. In this embodiment, various user association relationship data are defined, and through the various association relationship data, the user relationship network can be insight, so that whether the marketing range has further expansion capability and trend, and whether the marketing audience surface can expand and accurately analyze the user requirements can be obtained.

In one embodiment, the user value data includes user own value information and user contribution value information in a target business system; exemplary, for example, the user's own value information includes whether there is a car and its brand model, whether there is a room and its size and position, annual income interval, whether the enterprise is high-level, whether it is a VIP user served by a business, where the above contents can be obtained by guiding the user to actively report and submit to the real estate transaction center, the related vehicle management platform and other third party platforms, respectively; such as contribution value information including how close and support the user is to business page activities provided by the target business system. In the embodiment, various user value data are defined, and the user value and the actual contribution value can be used for judging whether to perform product innovation to match users meeting different grades.

In one embodiment, the risk assessment data includes risk assessment information and blacklist information for the user in a plurality of different dimensions; exemplary, for example, risk assessment information, including bank credit rating, third party credit rating, credit risk level, money laundering risk level, comprehensive credit line, credit breach record, delinquent payment record, repayment capability, breach probability, etc., specific data may be obtained by interfacing the bank; such as blacklist information including credit card overdue blacklist, credit overdue blacklist, owed subscriber list, insurance fraud insurance subscriber list, fraud list, specific data may be obtained by interfacing with specific third parties such as authorities such as the relevant banks. In this embodiment, the risk evaluation data is added with the risk evaluation information and the blacklist information, and further enlarges the data range, so that the risk prevention capability can be improved, for example, the risk evaluation data can be used for judging whether to put in a related product or limit purchasing the related product to the user so as to perform a certain risk prevention process.

The user marketing data includes financial demand information and non-financial demand information of the user during a preset period, and marketing campaign information of a campaign page provided by the user for the target service system. For example, non-financial demand information and financial demand information such as whether a wedding is prepared recently, whether a child is born recently and works recently, etc., financial demand information including whether a financial is desired to be purchased and whether a stock is desired to be purchased, etc., whether a user has a relevant demand recently can be determined by interfacing with a third party such as wedding, hospital, recruitment software, financial management software; such as marketing campaign information, may determine user loyalty, user satisfaction, user churn probability, marketing campaign acceptance, marketing campaign liveness, etc. by determining user relationships and behavior dynamics for marketing campaign product services through a campaign page, such as an H5 campaign page, provided by the user to the target business system. In the embodiment, the situation of attitude of a user to a certain marketing product is obtained more comprehensively, the situation can be used for improving the operation capability, related products tend to be pushed, and certain drainage is achieved.

It should be noted that the foregoing examples are merely exemplary, and in other embodiments, other data may be included, which is not limited herein. In addition, the applicant discovers that when the target service system is an H5 active page service system, because the H5 page is widely applied to various third party platform systems, such as government service systems, banking systems, common application systems and other third party platforms, the target service system and the third party platform are opened to guide the user to authorize and report data, and more comprehensive data can be obtained therefrom, so that the method is easier to realize, and the required data can be ensured to be comprehensively obtained.

In one embodiment, in step S40, that is, performing feature analysis on the user behavior data, and constructing a second type tag set of the user according to the feature analysis result, the method includes the following steps:

s41: performing feature statistical analysis on the user behavior data to count various behavior features of the user;

s42: respectively matching various behavior characteristics with predefined labels to obtain behavior labels with various behavior characteristics matched;

s43: and constructing a second type tag set of the user according to the behavior tags matched with the behavior features.

For user behavior data, the embodiment provides a way to generate a second type of tag set, and the user is tagged based on the user behavior data, such as clicking, purchasing, searching, etc., of the user. The behavior tags may be obtained by recording user behavior data and then parsing the user behavior data to count various behavior features of the user and match the various behavior features with predefined tags, respectively. For example, purchase preference tags may be tagged based on user purchase records, browse preference tags may be tagged based on user browse records, and so forth. The analytical features may be implemented based on a trained neural network model.

In this embodiment, a manner of constructing a second type tag set is provided, and various behavior tags are directly marked for a user in a manner of matching with features, so that the second type tag set based on user behavior data is constructed more comprehensively and accurately, and it is worth to say that, since the user behavior data acquired in the embodiment of the present application includes multiple aspects, the behavior tag can be acquired more quickly and simply by using the manner of matching the behavior features.

It should be noted that in other embodiments, there may be other ways of performing feature analysis on the user behavior data, and constructing a second type tag set of the user according to the feature analysis result, for example, performing feature analysis on the user behavior data by using a multi-mode algorithm to obtain behavior features of different modes, fusing the behavior features of different modes to obtain fused features, and finally inputting the fused features into a classifier to classify the fused features to obtain the behavior tag. In the analysis process of the user behavior data, the user behavior data can be better analyzed through a multi-mode algorithm, and a more appropriate behavior label is constructed.

In one embodiment, in step S50, that is, performing data analysis according to the target user data to construct a third type tag set of the user, to construct the third type tag set of the user, the method includes the following steps:

S51: mining the relevance among all classified data in the target user data through a preset algorithm so as to mine out potential relevance information of the user;

it should be understood that the target user data includes user attribute data, user association relationship data, user behavior data, user value data, user risk assessment data, and user marketing data, and in this embodiment, the classification data refers to user attribute data, user association relationship data, user behavior data, user value data, user risk assessment data, and user marketing data, and in this embodiment, relevance and regularity between the classification data in the target user data are mined to mine potential relevance information of the user.

S52: correspondingly generating an analysis tag of the user according to the relevance information;

s53: and constructing a third type tag set of the user according to different analysis tags of the user.

The analysis tag shows a tag corresponding to an analysis result characteristic obtained based on the analysis results of the plurality of data, and the analysis tag includes, for example, a personalized recommendation result tag, a crowd analysis result tag, a tag class such as user subdivision and a tag class such as content preference, and may also include a deduced analysis tag such as user loyalty, which is not particularly limited. According to the embodiment, through a preset algorithm, data analysis is carried out on the obtained target user data comprising the user attribute data, the user association relationship data, the user behavior data, the user value data, the user risk evaluation data and the user marketing data, the user data with multiple dimensions are considered, merging and mining processing is carried out based on the data, deep association relationship can be mined, and the analysis result and the corresponding analysis tag type of the close user can be analyzed.

In one implementation, the data analysis may be performed on the target user data by using algorithms including, but not limited to, a decision tree algorithm, a clustering algorithm, a naive bayes algorithm, a support vector machine algorithm, an association rule algorithm, and the like, to obtain a corresponding analysis tag type, which is not limited in detail.

In an embodiment, in order to perform data analysis more accurately to mine out relevance information, the embodiment of the present application further optimizes an algorithm, that is, in step S51, the relevance among all the classified data in the target user data is mined by a preset algorithm to mine out potential relevance information of the user, which includes the following steps:

s511: classifying and marking target user data by using a preset classification algorithm to obtain a plurality of data classification results, wherein the classification algorithm comprises a decision tree algorithm, a clustering algorithm, a naive Bayesian algorithm and a support vector machine algorithm;

s512: combining the data classification results to obtain a combined data classification result;

s513: and mining each frequent item set and association rule in the merged data classification result by using an association rule algorithm to obtain association information between data in the merged data classification result.

In the embodiment, a preset classification algorithm is used for classifying and marking target user data to obtain a plurality of data classification results, wherein the classification algorithm comprises a decision tree algorithm, a clustering algorithm, a naive Bayesian algorithm and a support vector machine algorithm. In the specific implementation, different algorithms are selected to combine and process, so that a more accurate and effective label result can be obtained, and in step S511, any 2 algorithms, 3 algorithms or all algorithms of a decision tree algorithm, a clustering algorithm, a naive bayes algorithm and a support vector machine algorithm can be selected, which is not particularly limited. For example, if 4 algorithms are selected, classifying and marking the target user data by adopting the 4 classification algorithms respectively to obtain 4 data classification results, and combining a plurality of data classification results to obtain a combined data classification result; and finally, mining all frequent item sets and association rules in the merged data classification result by using an association rule algorithm to obtain association information between data in the merged data classification result, wherein:

decision tree algorithm: and classifying and marking each classified data according to different characteristics and attributes in the target user data by constructing a decision tree model to obtain a data classification result. The decision tree algorithm can process various data types and has the characteristics of interpretability and easy understanding.

Clustering algorithm: by dividing the target user data into different groups and dividing the data of the same category into the same group, another data classification result can be obtained, and exemplary common clustering algorithms include K-means, hierarchical clustering and the like.

Naive bayes algorithm: based on Bayes formula and assumed condition independence, classifying and marking the target user data to obtain another data classification result, and the naive Bayes algorithm has the advantages of simplicity, rapidness and easiness in implementation.

Support vector machine algorithm: by constructing a classification hyperplane in a high-dimensional space, classifying and marking the target user data to obtain another data classification result, and the support vector machine algorithm has higher classification precision and generalization capability.

Association rule algorithm: by mining frequent item sets and association rules in the merged data classification result, association and correlation among different attributes in the merged data classification result are obtained, so that association information is mined, and finally, analysis tags are marked according to the association information, for example, 10% of users who buy shoes can buy socks, 60% of customers who buy bread can buy milk, which is not only the association information, but also the association information.

It should be noted that, in addition to the above-listed processing manner, there are many other analysis tag calculation algorithms, for example, classification and labeling using neural networks, random forests, logistic regression, etc., so as to obtain different data classification results, and when specific implementation is performed, a suitable algorithm may be selected according to specific service requirements and data characteristics, so as to obtain better and more suitable analysis tag results.

In the embodiment, the classified data results obtained through various algorithms are combined, so that on one hand, the data can be more comprehensive and different characteristics of the data are reflected, and on the other hand, the deep association relationship can be mined through combination and mining processing, so that the analysis tag type of a user is analyzed.

It should be noted that after fusion processing is performed on the first type tag set, the second type tag set and the third type tag set to obtain a fused tag portrait result of a user, a labeling process may be performed, where the labeling process is a process of performing certain data processing and labeling based on the method provided in the embodiment of the present application, and the data processing process is mainly divided into two processes, namely, offline processing, to perform a labeling task of the label periodically or regularly, and determine whether a customer meets the labeling rule to label. The method has the advantages that the requirement on server resources is low; secondly, real-time processing is performed, and whether the current client meets the tag rule is judged in near real time through the Flink stream processing, and the disadvantage of the mode is that the resource requirement of the server is high. The method has the advantages of near real-time labeling and high timeliness. According to the embodiment of the application, the off-line or real-time mode can be selected to realize the marking process, different marking requirements are met, and scenes with higher real-time requirements, such as accurate searching recommendation and other scenes with lower real-time requirements, can be met. The situation that the traditional scheme is single in processing scene is made up, and the practicality of the scheme is improved.

In an embodiment, after step S60, that is, fusion processing is performed on the first type tag set, the second type tag set and the third type tag set to obtain a fused tag image result of the user, including:

s70: binding and storing the fusion tag portrait result and the user identification of the user;

s80: receiving a visualization request carrying a user identifier, responding to the visualization request, and acquiring a fusion tag image result corresponding to the user identifier;

s90: and displaying the panoramic view of the fusion tag image result according to a preset panoramic image display mode.

After the fusion tag portrait result of the user is obtained, binding and storing the fusion tag portrait result and the user identification of the user, so that the final portrait construction of the user is completed, then when the user portrait analysis requirement or the knowledge requirement exists, the user can trigger a visualization request carrying the user identification, then respond to the visualization request to obtain the fusion tag portrait result corresponding to the user identification, and the fusion tag portrait result is displayed in a panoramic view according to a preset panoramic portrait display mode. The preset panoramic portrait display modes comprise display modes such as a chart, a report, a data instrument panel and the like, wherein the chart, the report and the data instrument panel contain tag information.

In this embodiment, the user data may be classified and labeled according to different business requirements and label classification rules. And calculating and matching the user data according to different algorithms and models to obtain a specific label result. And (3) fusing and integrating different label results to obtain a comprehensive label portrait of the user, and finally displaying the label portrait of the user in a visual mode, wherein the visual mode comprises charts, reports, data dashboards and the like, so that the display effect of the fused label portrait result is improved, and related personnel (such as business personnel) can conveniently analyze and apply the integrated label portrait.

It should be appreciated that after the fused tag portrait result of the user is obtained, various personalized activities can be provided for the user based on the fused tag portrait result, so that enterprises can better understand the user, user satisfaction and service benefit are improved, accurate marketing is achieved, and the thousand-person and thousand-face effect is achieved.

Personalized recommendation: by analyzing and fusing the label portrait results, interest preference and demand of the user are known, accurate personalized recommendation is realized, and satisfaction and loyalty of the user are improved. For example, a user likes an automobile, b user likes a watch, related commodities are respectively recommended through the automobile and watch type labels, and commodities in the consumption range capacity can be pushed more accurately through the consumption capacity labels.

Marketing and popularization: through analyzing and fusing the label portrait results, the purchasing intent and the consumption habit of the user are known, accurate marketing promotion is realized, and the advertising effect and the conversion rate are improved. For example, a users can push related commodity advertisements according to the tag data by browsing certain types of commodities frequently.

User subdivision: and the user is divided into different sub-divided groups by analyzing and fusing the label portrait results, and personalized marketing strategies are carried out aiming at the different groups, so that the user retention and conversion are improved. For example, if the members are classified into three classes a, b and c through the member class labels, the suitable marketing strategies under the label data of the three classes a, b and c can be personalized respectively.

And (3) product optimization: and the user habit and feedback of the user are known through analyzing and fusing the label portrait result, the product is optimized and improved according to the user demand and feedback, and the user experience and satisfaction of the product are improved. If a large number of clients have tag data which are unsatisfactory for certain products, the product optimization can be performed preferentially for the problem fed back by the tag data, so that the user satisfaction can be rapidly and accurately improved, and the client retention rate is maximized. Further improving the product quality and the income conversion.

Business decision: by analyzing and fusing the label portrait results, the behavior characteristics and trends of the user are known, and references and bases are provided for business decisions of enterprises, such as product strategies, market strategies and the like. Operators can further analyze the whole label portrait data of the user in a visual chart mode and the like, and a certain data basis is provided for subsequent strategic positioning and expansion of enterprises.

It can be seen that the embodiment enriches the number and types of the labels, and the label type construction basis data is richer and channels are more, so that a more perfect, comprehensive and accurate label system can be constructed, and the label system can be directly applied to various applications and has a wider application range.

It should be understood that the sequence number of each step in the foregoing embodiment does not mean that the execution sequence of each process should be determined by the function and the internal logic of each process, and should not limit the implementation process of the embodiment of the present application in any way.

In one embodiment, a user tag portrait processing device is provided, which corresponds to the user tag portrait processing method in the above embodiment one by one. As shown in fig. 4, the user tag portrait processing device includes a collection module 101, an integration module 102, a first construction module 103, a second construction module 104, a third construction module 105, and a fusion module 106. The functional modules are described in detail as follows:

A collection module 101, configured to collect first user data from a target service system and collect second user data from a third party platform system;

the integration module 102 is configured to integrate the first user data and the second user data to obtain target user data, where the target user data includes user attribute data, user association relationship data, user behavior data, user value data, user risk evaluation data, and user marketing data;

a first construction module 103, configured to perform classification statistics on the user attribute data, and construct a first type tag set of the user according to a classification statistics result;

a second construction module 104, configured to perform feature analysis on the user behavior data, and construct a second type tag set of the user according to a feature analysis result;

a third construction module 105, configured to perform mining analysis according to the target user data, so as to construct a third type tag set of the user;

and the fusion module 106 is configured to perform fusion processing on the first type tag set, the second type tag set and the third type tag set, so as to obtain a fused tag portrait result of the user.

In this embodiment, it may be seen that, the user tag portrait processing device is provided, except that the user data is collected from the own service system, user data is collected from any third party platform and integrated, the data is wider in range, more sources and finer in data granularity, it can be guaranteed that data is more comprehensive, perfected and more effective, the defects of single traditional data collection mode, deficient data types and the like are overcome, then a plurality of different types of tags of users are built based on the data obtained through searching, and each type of same tag covers a plurality of different tags, tag types are enriched and diversified, the problem that the tag types are single due to the fact that the tag is built only by means of attributes is overcome, the application range is wider, finally, the final fusion tag portrait result is formed by fusion processing of the tags of different tag sets, the constructed fusion tag portrait result is more comprehensive and accurate, the accurate processing can be achieved effectively according to the fusion tag portrait result, and the effective implementation of thousands of people is truly achieved.

In combination with the embodiment shown in fig. 4, in an embodiment, the user attribute data includes demographic attribute information, life information, location information, and custom tag information of various custom attributes, where the custom tag information is reported by guiding the user through a guiding page; the user behavior data includes financial product preference information, non-financial product preference information, internal channel preference information of the user at the target business system, and external channel preference information at the third party platform system.

In the embodiment, the user attribute data ensures the pertinence of the basic attribute, so that the marketing capability can be enhanced, the preliminary and accurate circle selection of the user is facilitated, and the user-defined label information is introduced, so that the accuracy can be further enhanced due to the fact that the label information is user-defined.

In combination with the embodiment shown in fig. 4, in an embodiment, the user association relationship data includes life association relationship information, financial association relationship information and social network association relationship information; the user value data comprises user own value information and contribution value information of the user; the risk evaluation data comprises blacklist information and risk evaluation information of a plurality of different dimensions for the user; the user marketing data comprises financial demand information and non-financial demand information of the user in a preset period and marketing activity information of the user aiming at activity pages provided by the target business system.

Further, the contribution value information comprises the closeness and the support degree of the user to the business page activities provided by the target business system, and the marketing activity information comprises the loyalty, the satisfaction, the acceptance and the liveness of the user to the activity page product services provided by the target business system.

In the embodiment, channel optimization can be realized for the user by defining various user behavior data types, and the pertinence and the comprehensiveness of the channel are optimized.

In one embodiment, the second building block 104 is configured to:

In an embodiment, the third building block 105 is configured to:

In the embodiment, the classified data results and the merging are obtained through various algorithms, so that on one hand, the data can be more comprehensive and different characteristics of the data are reflected, and on the other hand, the merging and the mining processing are carried out, so that deep association relations can be mined, and the analysis tag types of users are analyzed.

The specific limitation of the user tag portrait processing device can be referred to the limitation of the user tag portrait processing method hereinabove, and will not be described herein. The modules in the user tag portrait processing device may be all or partially implemented by software, hardware, or a combination thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.

In one embodiment, a computer device is provided, the internal structure of which may be as shown in FIG. 5. The computer device includes a processor, a memory, a network interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface of the computer device is used for communicating with an external service end through network connection. The computer program, when executed by the processor, performs the steps of the user tag portrait processing method, such as S10-S60 shown in fig. 3, or performs the functions of a user tag portrait processing device, which are not described herein again for the sake of avoiding repetition.

In an embodiment, a computer readable storage medium is provided, and a computer program is stored on the computer readable storage medium, where the computer program when executed by a processor implements each step of the user tag image processing method in the above embodiment, for example, S10-S30 shown in fig. 1, and is not repeated herein. Alternatively, the computer program when executed by the processor implements the functions of the modules/units in the embodiment of the user tag portrait processing device described above, and in order to avoid repetition, a description thereof will be omitted. The computer readable storage medium may be nonvolatile or may be volatile.

Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the various embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.

It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of the functional units and modules is illustrated, and in practical application, the above-described functional distribution may be performed by different functional units and modules according to needs, i.e. the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-described functions.

The above embodiments are only for illustrating the technical solution of the present application, and are not limiting; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present application, and are intended to be included in the scope of the present application.

Claims

1. A user tag portrait processing method, the method comprising:

2. The user tag portrait processing method of claim 1, wherein:

the user attribute data comprises population attribute information, life information, position information and custom tag information of various custom attributes, wherein the custom tag information is reported by the user guided by a guide page;

3. The user tag portrait processing method of claim 2, wherein:

the user association relationship data comprises life association relationship information, financial association relationship information and social network association relationship information;

4. The user tag portrait processing method of claim 3 wherein said contribution value information includes closeness and support of said user to business page activities provided by said target business system, and said marketing campaign information includes loyalty, satisfaction, acceptance and liveness of said user to active page product services provided by said target business system.

5. The method for processing a user tag portrait of any one of claims 1 to 4, wherein the performing feature analysis on the user behavior data and constructing a second type tag set of the user according to a feature analysis result includes:

6. The user tag portrait processing method according to any one of claims 1 to 4, wherein said performing data analysis according to said target user data to construct a third type tag set of said user, includes:

7. The method for processing a user tag portrait of claim 6, wherein mining the association between the classified data in the target user data by a preset algorithm to mine the potential association information of the user includes:

8. A consumer representation system, said consumer tag representation processing system comprising

A target business system, a third party platform system and a user tag portrait processing system for implementing the user tag portrait processing method according to any one of claims 1 to 7.

9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the user tag portrait processing method according to any one of claims 1 to 7 when the computer program is executed.

10. A computer readable storage medium storing a computer program, wherein the computer program when executed by a processor implements the steps of the user tag portrait processing method according to any one of claims 1 to 7.