User behavior analysis and precision marketing method based on model driving

Abstract

The invention discloses a user behavior analysis and precision marketing method based on model driving, and relates to the technical field of behavior analysis. Comprising the steps that basic information of a user is obtained, a binding portrait of a target user is set based on the basic information, a target portrait item is obtained, the target portrait item is used for representing information data matched with the target user, online information of the target user is obtained based on the target portrait item, a first information item is obtained, and an initial portrait model is built based on the first information item; and outputting initial analysis data of the target user based on the initial portrait model to obtain an initial information item. According to the method, the binding portrait is set by acquiring the basic information of the user, and the initial portrait model is built in combination with the online information, so that the initial analysis data can be output, the recommendation information can be generated, interests and requirements of the target user can be preliminarily known, recommendation contents fitting the target user can be provided for the target user, and the acceptability and click rate of the user to the recommendation information are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of behavior analysis, and specifically to a method for user behavior analysis and precision marketing based on model-driven Background Art

[0002] The CBSS service refers to its communication business core support system, which is the core information system used by China Unicom to support the operation of its entire communication business, covering key functions such as customer management, order processing, billing, network management, and financial settlement. In the marketing scenario of China Unicom's business hall, user behavior analysis and marketing mainly collect and analyze various behavior data of customers in the business hall to achieve more precise marketing and improve the customer experience.

[0003] A method and system for recommending online marketing strategies with the patent publication number CN118172091A analyzes multi-dimensional data information in the form of multi-dimensional data fusion, and uses a clustering algorithm to perform intelligent classification on data features, divides the consumer group in multiple dimensions, provides detailed data, and thus achieves a more precise effect of marketing strategy recommendation.

[0004] When analyzing user behavior in the above and similar technical solutions, since user behavior includes online and offline behaviors, and there are different correlations between different behaviors, and the salesperson cannot quickly associate and match the online and offline behaviors of users, it is impossible to quickly analyze the behavior data of users, and it is impossible to accurately complete the corresponding marketing. At the same time, due to the relatively unified salesperson's words, it is difficult to conduct personalized marketing according to different customers, and it is impossible to effectively introduce the product advantages and the matching situation with user needs, resulting in a low conversion rate and reducing the marketing success rate. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for user behavior analysis and precision marketing based on model-driven to solve the problems raised in the above background art.

[0006] To achieve the above purpose, the present invention provides the following technical solution: A method for user behavior analysis and precision marketing based on model-driven, including: Obtain the basic information of the user, set the bound portrait of the target user based on the basic information to obtain the target portrait item, and the target portrait item is used to represent the information data matching the target user; Based on the target portrait item, obtain the online information of the target user to obtain the first information item, build an initial portrait model based on the first information item, output the initial analysis data of the target user based on the initial portrait model to obtain the initial information item, and generate recommendation information for the target user based on the initial information item; Obtain language data information through big data acquisition methods, create a language library, and based on the basic information, obtain the personality information of the target user to obtain personality information items; Based on the personality information items, obtain the matching information in the language library to obtain the target language items, and the intelligent agent gives feedback and outputs based on the target language items; Obtain the personality information of the employees in the business hall to obtain a matching information set, and filter and obtain the corresponding reception employees through a screening method to obtain the target employee items; Judge the target employee items, set a judgment threshold, and when the target employee items do not meet the judgment threshold, re-screen the corresponding reception employees to obtain updated employee items; Obtain the movement heat map information of the target user to obtain the second information item, interfere with the initial portrait model based on the second information item, output the optimized analysis data of the target user to obtain the optimized information item, and generate recommendation information for the target user again based on the optimized information item. Integrate the analysis results of the target user's online and offline behaviors and fuse the model to output accurate marketing information.

[0007] Furthermore, the basic information includes mobile phone number information and identity information, and the acquisition method of the target portrait items includes: Set an intelligent agent, which interacts with the target user to obtain the mobile phone number information and identity information of the target user to obtain the basic information; Based on the basic information, set the binding information of the target user, and the intelligent agent obtains the appearance information of the target user. The appearance information is combined with the binding information to obtain the target portrait items.

[0008] Furthermore, the online information includes click trajectories, page stay durations, and service evaluation texts, and the acquisition method of the first information item includes: Obtain the bound software information to obtain the target information item, and based on the target portrait items, record the click trajectories of the target user through data embedding to obtain the click information item; Based on the click information item, obtain the page stay duration of the target user to obtain an information duration set, and the information duration set includes at least the page stay duration of one click information item; Obtain the service evaluation text of the target user to obtain the evaluation information item, and obtain the keyword information of the evaluation information item to obtain the evaluation feature item; Based on the combined result of the information duration set and the evaluation feature item, obtain the first information item.

[0009] Furthermore, the construction method of the initial portrait model includes: Use the basic information as the unique identifier, associate the first information item, parse the user access path through the first information item, and obtain the high-frequency access modules of the user to obtain the access information item; Based on the access information items, set the determination period, obtain the total access time of the access information items within the determination period to get the access time items, use the target model to divide the user value levels, use the basic information as the reference label, and use the access information items and access time items as training data for model training, and output the initial portrait model.

[0010] Furthermore, the method for creating the language library includes: Determine the coverage range, collect the designated dialect points based on the data acquisition method, set the classification area, and classify according to the target classification criteria to obtain the classified language items; Set the data dimensions, including speech, text, video, and semantic annotation, to obtain the dimension modality items; Based on the combined results of the classified language items and the dimension modality items, store them through the target storage structure to obtain the language library.

[0011] Furthermore, the personal information includes territorial information and dynamic information, and the method for obtaining the target language items includes: Based on the basic information, obtain the territorial information of the target user to get the territorial information items, and match the data information in the language library based on the territorial information items to get the territorial matching items; Obtain the dynamic information of the target user, where the dynamic information includes speech information and body information, to get the dynamic information items, and match the speech information in the territorial matching items based on the dynamic information items to get the target language items.

[0012] Furthermore, the screening method includes: Based on the comparison relationship between the target language items and the matching information set, obtain the matching relationship set, and sort the matching relationship set in the order of the degree of fit to obtain the sorted relationship set; Obtain the employees corresponding to the sorted relationship set to get the target employee set, and obtain the employee ranked first in the target employee set as the reception employee to get the target employee item.

[0013] Furthermore, the method for obtaining the updated employee item includes: According to the set judgment threshold, the judgment threshold is the queuing number threshold. When the queuing number of the target employee item exceeds the judgment threshold, based on the target employee set, select the target employee in the next order as the reception employee to get the updated employee item.

[0014] Furthermore, the moving line heat information includes the stay duration and window access information, and the method for obtaining the second information item includes: Obtain the dynamic path of the target user through the target device, obtain the window access information of the target user to get the access window item; Obtain the window stay time of the target user to get the window stay item corresponding to the access window item; Obtain the key feature information of the access window item to obtain the window feature item, and combine the window feature item with the window stay item to obtain the second information item.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This model-driven user behavior analysis and precision marketing method sets a bound portrait by obtaining user basic information, and then builds an initial portrait model in combination with online information. It can output initial analysis data and generate recommendation information, can initially understand the interests and needs of target users, provide personalized recommendation content for them, improve the acceptance rate and click-through rate of recommendation information by users, obtain user personality information and match it with the language library, enable the intelligent agent to perform feedback output based on the target language item, realize a personalized interaction experience, enhance the interactivity and stickiness between users and the system, interfere with the initial portrait model by analyzing the user movement heat information, output optimized analysis data and generate recommendation information again, and can further accurately grasp the changes in user needs, and improve the accuracy and effectiveness of recommendations.

[0016] At the same time, by obtaining the personality information of the business hall employees to form a matching information set, screening to obtain the target employee item, and setting a judgment threshold for optimization, it can ensure that the most suitable reception employee is assigned to the target user, improve the service quality and user satisfaction. If the target employee does not meet the requirements, it can be re-screened in time to ensure the efficiency and professionalism of the service. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram of the overall process of the present invention; Figure 2 It is a schematic diagram of the method for obtaining the first information item of the present invention; Figure 3 It is a schematic diagram of the relationship between the first information item and the click information item of the present invention; Figure 4 It is a schematic diagram of the recommendation information generation process of the present invention; Figure 5 It is a schematic diagram of the target language item acquisition process of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0019] Due to the complex form of the user's behavior trajectory, which presents an online-offline integration, users may browse product information, participate in community interactions, and download applications through online channels, while conducting business consultations, handling services, and participating in promotional activities through offline channels. These behaviors are not isolated but are interrelated and mutually influential. For example, a user may learn about a certain preferential package online and then go to the offline business hall for detailed consultation and handling. However, China Unicom's systems and processes have not yet been able to effectively correlate and match the user's online and offline behaviors. Salespersons often can only obtain the user's behavior data from a single channel and cannot comprehensively understand the user's preferences, needs, and potential intentions. This information asymmetry directly affects the accuracy and effectiveness of marketing decisions. It is difficult for salespersons to judge which products and services the user is more interested in based on the user's complete behavior portrait, and it is even more difficult to accurately predict the user's future consumption behavior. Furthermore, even for the user behavior data of a single channel, salespersons lack fast and convenient analysis tools. Facing a large amount of user data, manual analysis is inefficient and prone to errors. Even with preliminary user data, due to the lack of professional analysis skills and tools, salespersons often have difficulty uncovering the patterns and trends hidden behind the data. In addition to the difficulties in behavior analysis, the lack of personalization in marketing strategies is also an important reason for the low conversion rate. Traditional marketing models often use unified sales scripts and promotional materials, lacking personalized recommendations for different users. The uniform marketing methods are difficult to effectively touch the user's needs and even less able to stimulate their purchase desire. If salespersons still use unified marketing scripts and mechanically introduce the parameters and functions of products, it will be difficult to effectively attract users of different groups, fail to effectively introduce the product advantages and the matching with user needs, and ultimately lead to the user's lack of interest in the product, thereby reducing the marketing conversion rate. However, the model-driven user behavior analysis and precision marketing method provided in this application can set a bound portrait for the target user based on the obtained basic information of the user, and then build an initial portrait model in combination with online information, output initial analysis data, and generate recommendation information, which can initially understand the interests and needs of the target user, provide personalized recommended content for them, improve the acceptance and click-through rate of the user for the recommended information, and thus increase the success rate of marketing. Secondly, by obtaining the user's personality information and matching it with the language library, the intelligent agent can output feedback based on the target language item, realizing a personalized interaction experience and enhancing the interactivity and stickiness between the user and the system. Moreover, by analyzing the user's movement heat information to interfere with the initial portrait model, outputting optimized analysis data and generating recommendation information again, it can further accurately grasp the changes in user needs, improve the accuracy and effectiveness of the recommendation, and lay a solid foundation for precise marketing, as Figure 1 shown, including steps S100 - S800.

[0020] Step S100: Obtain the basic information of the user, set the bound portrait of the target user based on the basic information, and obtain the target portrait item.

[0021] It should be noted that the target portrait item is used to represent the information data matching the target user. The basic information includes mobile phone number information and identity information. The methods for obtaining the target portrait item include: setting an intelligent agent, which interacts with the target user. The intelligent agent includes an intelligent robot. Obtain the mobile phone number information and identity information of the target user to obtain the basic information; based on the basic information, set the binding information of the target user. The intelligent agent obtains the appearance information of the target user. There is a high-definition camera on the intelligent robot, which is used to capture the appearance information of the target user. The appearance information is combined with the binding information to obtain the target portrait item.

[0022] Step S200: Based on the target portrait item, obtain the online information of the target user to obtain the first information item.

[0023] It should be noted that as Figure 2 shown, the online information includes click trajectory, page stay duration, and service evaluation text. The methods for obtaining the first information item include: obtaining the bound software information to obtain the target information item, based on the target portrait item, logging the click trajectory of the target user through data embedding to obtain the click information item; based on the click information item, obtain the page stay duration of the target user to obtain the information duration set, and the information duration set includes at least the page stay duration of one click information item; obtain the service evaluation text of the target user to obtain the evaluation information item, and obtain the keyword information of the evaluation information item to obtain the evaluation feature item; based on the combined result of the information duration set and the evaluation feature item, obtain the first information item.

[0024] Embodiment 1 In the specific implementation process, as Figure 3 shown, a certain target user goes to the business hall to consult relevant services. The intelligent robot set in the business hall obtains through interaction with the target user that the mobile phone number of a certain target user is 186****5624, and the identity information is **************5874. After obtaining the appearance information of the target user at the same time, the target portrait item A is created. At the same time, the bound software information obtained is "China Unicom", and the click trajectory of the target user logged through data embedding is "traffic", "data package", "traffic query", "large data package per month", "phone bill", and then the click information item is obtained. The page stay durations of this user are obtained as 10s, 30s, 20s, 20s, 5s respectively to obtain the information duration set, which corresponds to the click information item respectively. At the same time, the service evaluation text of this user is not obtained. At this time, the first information item is obtained according to the information duration set.

[0025] Step S300: Build an initial portrait model based on the first information item, output the initial analysis data of the target user based on the initial portrait model to obtain the initial information item, and generate recommended information for the target user based on the initial information item.

[0026] It should be noted that the method for building the initial portrait model includes: using the basic information as the unique identifier, associating the first information item, parsing the user access path through the first information item, obtaining the user's high-frequency access module, and obtaining the access information item; based on the access information item, setting a determination period, the determination period is 7 days, obtaining the total access time of the access information item within the determination period, obtaining the access time item, using the target model to divide the user value level, the target model is the K-means+RFM model, using the basic information as the control label, and using the access information item and the access time item as the training data for model training, and outputting the initial portrait model.

[0027] Embodiment 2 In the specific implementation process, as Figure 4 shown, a certain target user goes to the business hall to consult relevant services. The intelligent robot set in the business hall obtains the mobile phone number, identity information of the certain target user through the interaction with the target user, and at the same time obtains the appearance information of the target user. At this time, using these information as the unique identifier, associating the online information of the user, identifying the high-frequency access module through the click trajectory, obtaining that the traffic query in the access module ratio of this user is 60%, the package change is 25%, and the points exchange is 15%, obtaining the access information item, according to the set determination period, obtaining that the total access durations of the access information item are 30 minutes, 12.5 minutes, and 7.5 minutes respectively, obtaining the access time item. At this time, using the K-means+RFM model to divide the user value level: R (recent access): the number of logins in the past 7 days, F (access frequency): the access frequency, M (consumption potential): speculated based on the historical package price and the staying page, using XGBoost to build a binary classification model to predict the user demand direction, features: the ranking of the page staying duration, the sentiment score of the service evaluation, target variable: the historical business handling type, obtaining the output initial portrait model of this user, based on the initial portrait model, outputting the initial analysis data of the target user, that is, "traffic query", obtaining the initial information item, and generating recommendation information for the target user based on the initial information item, that is, the traffic change and the recommended window for handling this item.

[0028] Step S400: Obtain the language data information through the big data acquisition method, create a language library, and obtain the personality information of the target user based on the basic information, obtaining the personality information item.

[0029] It should be noted that the method for creating a language library includes: determining the coverage range, collecting the designated dialect points based on the data acquisition method, with preference given to collecting the more than 1,000 dialect points designated by the "Project for the Protection of Chinese Language Resources", setting up classification regions, classifying according to the target classification criteria, and classifying into seven major language families such as Northern, Wu, Xiang, Gan, Hakka, Yue, and Min to obtain classified language items; setting up data dimensions, including speech, text, video, and semantic annotation, to obtain dimensional modality items, where the video includes body language; storing based on the combined results of the classified language items and the dimensional modality items through the target storage structure, and the target storage structure is the MongoDB storage structure to obtain the language library.

[0030] Step S500: Based on the personalized information items, obtain the matching information in the language library to get the target language item, and the intelligent agent makes a feedback output based on the target language item.

[0031] It should be noted that as Figure 5 shown, the personalized information includes territorial information and dynamic information. The method for obtaining the target language item includes: based on the basic information, obtaining the territorial information of the target user to get the territorial information item, and matching the data information in the language library based on the territorial information item to get the territorial matching item; obtaining the dynamic information of the target user, where the dynamic information includes speech information and body information, to get the dynamic information item, and matching the speech information in the territorial matching item based on the dynamic information item to get the target language item.

[0032] Embodiment III In the specific implementation process, a certain target user goes to the business hall to consult relevant services. The intelligent robot set in the business hall obtains the mobile phone number of the target user as 186****5655 and the identity information as 440300********3804 through interaction with the target user. After obtaining the appearance information of the target user, a target portrait item B is created. At this time, according to the basic information, the territorial information of this user is obtained as Shenzhen, Guangdong, and the data information in the language library is matched to get the territorial matching item, which is Shenzhen, Guangdong. The intelligent robot further obtains the speech information and body information of this user. The speech information is Mandarin, and the body language is conventional body language without representing special meanings. At this time, based on the dynamic information item, the speech information in the territorial matching item is matched to get the target language item, that is, the Shenzhen dialect. At this time, the intelligent robot makes a feedback output in the Shenzhen dialect.

[0033] Step S600: Obtain the personalized information of the employees in the business hall to get a set of matching information, and screen and obtain the corresponding reception employees through the screening method to get the target employee item.

[0034] It should be noted that the screening method includes: obtaining a matching relationship set based on the comparative relationship between the target language item and the matching information set, sorting the matching relationship set in order of fit, and obtaining a sorted relationship set; obtaining employees corresponding to the sorted relationship set to obtain a target employee set, and obtaining the employee ranked first in the target employee set as the reception employee to obtain a target employee item.

[0035] Example 4 During the specific implementation process, a target user goes to the business hall to consult about related business. The intelligent robot installed in the business hall obtains the user's location information as Shenzhen, Guangdong based on basic information, and matches the voice information in the location matching item based on the dynamic information item to obtain the target language item, that is, the Shenzhen dialect. At the same time, the personality information of the employees in the business hall is obtained. There are ten employees in the business hall, namely 1-10, among which employee 1 is from Shenzhen, Guangdong, employee 2 is from Guangzhou, Guangdong, employee 3 is from Xi'an, Shaanxi, employee 4 is from Wuhan, Hubei, employee 5 is from Hangzhou, Zhejiang, employee 6 is from Zhuhai, Guangdong, employee 7 is from Foshan, Guangdong, employee 8 is from Chengdu, Sichuan, employee 9 is from Nanjing, Jiangsu, and employee 10 is from Suzhou, Jiangsu. At this time, the matching relationship set is sorted in order of the degree of fit to obtain the sorted relationship set, which is employee 1, employee 2, employee 6, employee 7, employee 4, employee 5, employee 9, employee 10, employee 8, and employee 3. The employee ranked first in the target employee set is selected as the reception employee, and the target employee item, namely employee 1, is obtained.

[0036] Step S700: the target employee item is judged and a judgment threshold is set. When the target employee item does not meet the judgment threshold, the corresponding reception employee is re-screened to obtain an updated employee item.

[0037] It should be noted that the method for obtaining the updated employee item includes: according to the set judgment threshold, the judgment threshold is the queue number threshold, and the queue number threshold is 5 people. When the number of people in the queue of the target employee item exceeds the judgment threshold, based on the target employee set, the target employee with the highest priority is selected as the reception employee to obtain the updated employee item.

[0038] Step S800: Acquire the target user's movement path thermal information to obtain a second information item, interfere with the initial portrait model based on the second information item, output the target user's optimized analysis data, and obtain an optimized information item.

[0039] It should be noted that, based on the optimized information items, recommendation information is generated for the target user again. By comprehensively analyzing the online and offline behavior analysis results of the target user and integrating the model, accurate marketing information is output. The moving line heat information includes the stay duration and window access information. The method for obtaining the second information item includes: obtaining the dynamic path of the target user through the target device, where the target device is a high-definition camera set in the business hall, obtaining the window access information of the target user to obtain the access window item; obtaining the window stay time of the target user to obtain the window stay item corresponding to the access window item; obtaining the key feature information of the access window item to obtain the window feature item, and combining the window feature item with the window stay item to obtain the second information item.

[0040] Specifically, different access windows represent different business handling. Therefore, by obtaining the key feature information of the access window item, the characteristic business information of different access windows can be obtained, and then the window feature item can be obtained. According to the window feature items of different windows and the stay time of each window, the second information item is combined. Based on the second information item, the initial portrait model is interfered to output the optimized analysis data of the target user.

[0041] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended embodiments and their equivalents.

Claims

1. A model-driven user behavior analysis and precision marketing method, including: Obtain the basic information of the user, set the bound portrait of the target user based on the basic information, and obtain the target portrait items, where the target portrait items are used to represent the information data matching the target user; It is characterized in that: Based on the target portrait items, obtain the online information of the target user to get the first information items, build an initial portrait model based on the first information items, output the initial analysis data of the target user based on the initial portrait model to get the initial information items, and generate recommendation information for the target user based on the initial information items; Obtain the language data information through big data acquisition, create a language library, and obtain the personality information of the target user based on the basic information to get the personality information items; Based on the personality information items, obtain the matching information in the language library to get the target language items, and the intelligent agent makes a feedback output based on the target language items; Obtain the personality information of the employees in the business hall to get a set of matching information, and screen and obtain the corresponding reception employees through a screening method to get the target employee items; Judge the target employee items, set a judgment threshold, and when the target employee items do not meet the judgment threshold, re-screen the corresponding reception employees to get the updated employee items; Obtain the movement heat map information of the target user to get the second information items, interfere with the initial portrait model based on the second information items, output the optimized analysis data of the target user to get the optimized information items, generate recommendation information for the target user again based on the optimized information items, and comprehensively analyze the online and offline behavior analysis results of the target user and fuse the model to output precise marketing information.

2. The method for model-driven user behavior analysis and precision marketing according to claim 1, wherein: The basic information includes mobile phone number information and identity information, and the acquisition method of the target portrait items includes: Set an intelligent agent, the intelligent agent interacts with the target user to obtain the mobile phone number information and identity information of the target user to get the basic information; Based on the basic information, set the bound information of the target user, and the intelligent agent obtains the appearance information of the target user, and the appearance information is combined with the bound information to get the target portrait items.

3. The model-driven user behavior analysis and precision marketing method according to claim 1, characterized in that: The online information includes click trajectory, page stay duration, and service evaluation text, and the acquisition method of the first information items includes: Obtain the bound software information to get the target information items, and based on the target portrait items, record the click trajectory of the target user by embedding points to get the click information items; Based on the click information items, obtain the page stay duration of the target user to get the information duration set, and the information duration set includes at least the page stay duration of one click information item; Obtain the service evaluation text of the target user to get the evaluation information items, and obtain the keyword information of the evaluation information items to get the evaluation feature items; 4. The method for model-driven user behavior analysis and precision marketing according to claim 1, characterized in that: Based on the combined result of the information duration set and the evaluation feature items, get the first information items. The building method of the initial portrait model includes: Use the basic information as the unique identifier, associate the first information items, parse the user access path through the first information items, and obtain the high-frequency access modules of the user to get the access information items; Based on the access information items, set the determination period, obtain the total access time of the access information items within the determination period to get the access time items, use the target model to divide the user value levels, use the basic information as the comparison label, and use the access information items and access time items as training data for model training, and output the initial portrait model.

5. The model-driven user behavior analysis and precision marketing method according to claim 1, characterized in that: The method for creating the language library includes: Determine the coverage range, collect the designated dialect points based on the data acquisition method, set the classification area, and classify according to the target classification standard to obtain the classified language items; Set the data dimensions, including speech, text, video, and semantic annotation, to obtain the dimension modality items; Based on the combined results of the classified language items and the dimension modality items, store them through the target storage structure to obtain the language library.

6. The model-driven user behavior analysis and precision marketing method according to claim 1, characterized in that: The personal information includes territorial information and dynamic information. The method for obtaining the target language items includes: Based on the basic information, obtain the territorial information of the target user to get the territorial information items, and match the data information in the language library based on the territorial information items to get the territorial matching items; Obtain the dynamic information of the target user. The dynamic information includes speech information and body information to get the dynamic information items, and match the speech information in the territorial matching items based on the dynamic information items to get the target language items.

7. The model-driven user behavior analysis and precision marketing method according to claim 1, characterized in that: The screening method includes: Based on the comparison relationship between the target language items and the matching information set, obtain the matching relationship set, sort the matching relationship set in the order of the degree of fit to get the sorted relationship set; Obtain the employees corresponding to the sorted relationship set to get the target employee set, and obtain the employee ranked first in the target employee set as the reception employee to get the target employee item.

8. The method for model-driven user behavior analysis and precision marketing according to claim 7, characterized in that: The method for obtaining the updated employee item includes: According to the set judgment threshold, the judgment threshold is the queuing number threshold. When the queuing number of the target employee item exceeds the judgment threshold, select the target employee in the next order from the target employee set as the reception employee to get the updated employee item.

9. The method for model-driven user behavior analysis and precision marketing according to claim 1, wherein: The moving line heat information includes the stay duration and window access information. The method for obtaining the second information item includes: Obtain the dynamic path of the target user through the target device, obtain the window access information of the target user to get the access window item; Obtain the window stay time of the target user to get the window stay item corresponding to the access window item; Obtain the key feature information of the access window item to get the window feature item, and combine the window feature item with the window stay item to get the second information item.

Images