Marketing data analysis method based on big data portrait

Abstract

The invention discloses a marketing data analysis method based on a big data portrait, and relates to the technical field of electric digital data processing. The marketing data analysis method based on the big data portrait comprises the following steps of obtaining a portrait similarity coefficient, obtaining a comparison result and adopting a portrait data acquisition strategy. According to the method, the video portrait data is collected in the preset time period, the portrait similarity coefficient is obtained according to the collected video portrait data, then the portrait similarity coefficient is compared with the similarity threshold obtained from the preset database to obtain the comparison result, and finally the portrait data collection strategy is adopted according to the comparison result. The effect of improving the updating timeliness of the user portrait data in the marketing data is achieved, and the problem that the user portrait data in the marketing data is not updated in time in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the field of electronic digital data processing technology, and in particular to a marketing data analysis method based on big data portraits. Background Art

[0002] With the rapid development of big data technology, enterprises can collect massive amounts of user information, including basic attributes, behavioral data, purchase preferences, etc. By building refined user portraits, enterprises can deeply understand the target customer groups and achieve precision marketing. This method not only improves marketing efficiency, but also reduces costs and enhances market competitiveness. Big data portraits have become a key tool for enterprises to formulate marketing strategies, optimize product design, and improve user experience, bringing unprecedented business insights and growth opportunities to enterprises.

[0003] Existing marketing data analysis methods based on big data portraits mainly focus on building user portraits by collecting, processing and analyzing big data, thereby guiding precision marketing. These methods usually include using big data technology to comprehensively obtain static and dynamic information of users, generating personalized user tags through data analysis and feature extraction, and building user portrait models based on these tags. Subsequently, based on the user portrait model, enterprises can implement precision marketing strategies such as personalized recommendations and targeted advertising. These technologies not only improve marketing efficiency, but also enhance user experience, bringing significant market competitive advantages to enterprises.

[0004] For example, the patent application with the publication number CN111737338A discloses a closed-loop marketing data analysis method based on big data portrait, including: step S1: acquiring data; step S2: data processing, step S3: forming results and storing them in the warehouse; step S4: visualizing the stored data, reading and printing through the database; step S5: precise push, wherein precise push is pushed through commercial advertisements. Advantage 1: The present invention reduces the physical storage space of the computer occupied by these data through efficient analysis of sales data, can more accurately grasp the dynamic changes of sales data, determine the production direction according to sales information, maximize the development of the functions of sales data, and manage and make decisions on the entire sales process in real time. Advantage 2: The present invention uses an e-commerce marketing system based on big data analysis to put the company's marketing, sales, customer service, production scheduling and resource management in a complete closed-loop system, enabling the company to maintain rapid development with the help of the O2O model in the fierce market competition.

[0005] For example, the invention patent announcement with announcement number: CN115168740B discloses a method and system for generating marketing tasks based on big data analysis, including: constructing a label sequence for a new game and obtaining a label sequence for shipped games based on big data, obtaining games of the same category as the new game based on the similarity of the label sequence; obtaining all users of each shipped game in the same category of games and each user's activity representation value and the interaction value between users; constructing a user association graph corresponding to each shipped game; obtaining the degree of the node based on the node information of each node in the user association graph, performing modular calculations based on the degree of each node and performing graph clustering on the user association graph to obtain multiple node categories; obtaining the user recommendation priority of each user and the category recommendation priority of each node category; generating marketing tasks in combination with the user recommendation priority and the category recommendation priority; the marketing tasks are more applicable and have a wider promotion scope.

[0006] However, in the process of implementing the technical solution of the invention in the embodiments of the present application, the present application found that the above technology has at least the following technical problems:

[0007] In the existing technology, especially in short video platforms, since users' interests and needs are constantly changing with various factors such as time, environment, and social trends, user portraits constructed based on big data portraits may not be able to timely reflect changes in users' interests in watching videos, and there is a problem of untimely updating of user portrait data in marketing data. Summary of the invention

[0008] The embodiment of the present application solves the problem of untimely updating of user portrait data in marketing data in the prior art by providing a marketing data analysis method based on big data portraits, thereby improving the timeliness of updating user portrait data in marketing data.

[0009] An embodiment of the present application provides a marketing data analysis method based on big data portraits, comprising the following steps: collecting video portrait data within a preset time period, and obtaining a portrait similarity coefficient based on the collected video portrait data, wherein the video portrait data includes feedback portrait data and user portrait data, and the portrait similarity coefficient is used to measure the degree of similarity between the current user portrait data and the feedback portrait data; comparing the portrait similarity coefficient with a similarity threshold obtained from a preset database to obtain a comparison result, wherein the comparison result includes whether the portrait meets the standard and whether the portrait does not meet the standard; and adopting a portrait data collection strategy based on the comparison result, wherein the portrait data collection strategy is used to improve the timeliness of the user portrait.

[0010] Furthermore, the process of obtaining the video portrait data is as follows: collecting feedback portrait data by sending a portrait questionnaire to the users to be investigated in the short video platform to be analyzed within a preset time period, and transmitting the obtained feedback portrait data to a preset database for storage, wherein the feedback portrait data includes feedback video tags and feedback video duration; obtaining user portrait data of the users to be investigated from the short video platform to be analyzed, wherein the user portrait data includes user video tags and user video duration, and the content category of the user portrait data is consistent with that of the feedback portrait data; obtaining the video type similarity between the feedback video tags and the user video tags through a hash algorithm, and obtaining the video duration similarity between the feedback video duration and the user video duration through a hash algorithm, wherein the video type similarity indicates the similarity between the user video tags and the feedback video tags, and the video duration similarity indicates the similarity between the user video duration and the feedback video duration.

[0011] Furthermore, the process of obtaining the portrait similarity coefficient is as follows: obtaining evaluation data from a preset database, the evaluation data including a video type evaluation threshold and a video duration evaluation threshold; numbering the users to be investigated, and obtaining individual portrait similarity coefficients based on video type similarity and video duration similarity, and obtaining portrait similarity coefficients based on the individual portrait similarity coefficients, wherein the individual portrait similarity coefficient represents the portrait similarity coefficient of a single user to be investigated, and the portrait similarity coefficient is calculated using the following formula:

[0012]

[0013] Where f is the number of users to be investigated, f = 1, 2,,, F, F is the total number of users to be investigated, VTS f VDS represents the video type similarity of the fth user to be investigated. f represents the video duration similarity of the fth user to be investigated, μ represents the video type evaluation threshold, ρ represents the video duration evaluation threshold, ISC f represents the individual portrait similarity coefficient of the fth user to be investigated, and PSC represents the portrait similarity coefficient of the user to be investigated in the current preset time period in the short video platform to be analyzed.

[0014] Furthermore, the process of obtaining the comparison result is as follows: a similarity threshold is obtained from a preset database, and the obtained similarity threshold is compared with the portrait similarity coefficient: if the portrait similarity coefficient is not less than the similarity threshold, the comparison result is that the portrait meets the standard; if the portrait similarity coefficient is less than the similarity threshold, the comparison result is that the portrait does not meet the standard.

[0015] Furthermore, the specific process of adopting the portrait data collection strategy based on the comparison result is as follows: if the comparison result is that the portrait meets the standards, the current user portrait will continue to be used; if the comparison result is that the portrait does not meet the standards, the impact of the current data collection delay on the construction of the user portrait is judged based on the data collection delay index. The data collection delay includes data processing delay and data transmission delay. The data collection delay index is used to quantify the impact of data collection delay on video portrait data collection.

[0016] Furthermore, the process of obtaining the data collection delay index is as follows: obtaining data processing delay and data transmission delay through the log records of the short video platform to be analyzed, the data processing delay represents the time delay generated by the short video platform to be analyzed when processing video portrait data, and the data transmission delay represents the time delay generated by the short video platform to be analyzed in the process of transmitting video portrait data; obtaining reference data from a preset database, the reference data including average data processing delay, average data transmission delay, processing delay evaluation value and transmission delay evaluation value; obtaining an individual delay index based on the data collection delay and the reference data, and then obtaining a data collection delay index based on the individual delay index, the individual delay index represents the data collection delay index of a single user to be investigated, and the data collection delay index is calculated using the following formula:

[0017]

[0018]

[0019] In the formula, f represents the number of users to be investigated, f=1,2,,,F, F represents the total number of users to be investigated, CD f represents the data processing delay of the fth user to be investigated, represents the average data processing delay, α represents the processing delay evaluation value, TD f represents the data transmission delay of the fth user to be investigated, represents the average data transmission delay, β represents the transmission delay evaluation value, and IDI f represents the individual delay index of the fth user to be investigated, and DCI represents the data collection delay index within a preset time period.

[0020] Furthermore, the data collection delay index is obtained, and then the data collection delay index is compared with a data collection delay index threshold obtained from a preset database to obtain a judgment result, which is as follows: the judgment result includes normal data collection delay and abnormal data collection delay; the specific method for obtaining the judgment result is as follows: if the data collection delay index is not less than the data collection delay index threshold, the judgment result is that the data collection delay is normal; if the data collection delay index is less than the data collection delay index threshold, the judgment result is that the data collection delay is abnormal.

[0021] Furthermore, the judgment result is obtained, and then feedback on the judgment result is included, and the specific process is as follows: if the judgment result is that the data collection delay is normal, a portrait update judgment is performed based on the user portrait update frequency, and the portrait update judgment is used to determine whether the user portrait update frequency has an impact on video marketing; if the judgment result is that the data collection delay is abnormal, feedback is given to the preset analyst that the factor causing the portrait similarity coefficient to fail to meet the video marketing requirements is the data collection delay, and a command prompt for reducing the data collection delay is sent to the preset analyst.

[0022] Furthermore, the specific steps of judging the update of user portraits according to the update frequency of user portraits are as follows: Step 1, obtaining the current user portrait update frequency from the log records of the short video platform to be analyzed, and obtaining the average user portrait update frequency from the preset database, the user portrait update frequency represents the frequency of updating the user portraits of the short video platform to be analyzed in the current preset time period; Step 2, obtaining the update frequency evaluation value from the preset database, and obtaining the individual user portrait update frequency index based on the user portrait update frequency, the average user portrait update frequency and the update frequency evaluation value, and obtaining the user portrait update frequency index according to the individual user portrait update frequency index, the individual user portrait update frequency index represents the user portrait update frequency index of a single user to be investigated, and the user portrait update frequency index is used to quantify the influence of the user portrait update frequency on video marketing; Step 3, comparing the user portrait update frequency index with the user portrait update frequency index threshold obtained from the preset database and performing update feedback;

[0023] The user portrait update frequency index is calculated using the following formula:

[0024]

[0025] Where f is the number of users to be investigated, f = 1, 2,,, F, F is the total number of users to be investigated, UF f represents the update frequency of the user profile of the fth user to be investigated, represents the average user profile update frequency, θ represents the update frequency evaluation value, and IUI fIt represents the update frequency index of the individual user portrait of the fth user to be investigated, and UFI represents the update frequency index of the user portrait of the short video platform to be analyzed within the preset time period.

[0026] Furthermore, the specific process of comparing the user portrait update frequency index with the user portrait update frequency index threshold obtained from a preset database and performing update feedback is as follows: if the user portrait update frequency index is not less than the user portrait update frequency index threshold, feedback is given to the preset analyst that there are no problems with data collection delay and user portrait update frequency; if the user portrait update frequency index is less than the user portrait update frequency index threshold, a command prompt to increase the user portrait update frequency is sent to the preset analyst.

[0027] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0028] 1. By collecting video portrait data within a preset time period, and obtaining a portrait similarity coefficient based on the collected video portrait data, and then comparing the portrait similarity coefficient with the similarity threshold obtained from a preset database to obtain a comparison result, and finally adopting a portrait data collection strategy based on the comparison result, dynamic updating of user portraits is achieved, thereby improving the timeliness of updating user portrait data in marketing data, and effectively solving the problem of untimely updating of user portrait data in marketing data in the prior art.

[0029] 2. By obtaining the similarity threshold from the preset database and comparing the obtained similarity threshold with the portrait similarity coefficient, it is possible to determine whether the current user portrait is timely, and then adopt a portrait data collection strategy to improve the timeliness of the user portrait.

[0030] 3. By comparing the data collection delay index with the data collection delay index threshold obtained from the preset database, the judgment result is obtained, so as to determine the impact of data collection delay on user portrait construction, and then take different methods to improve the accuracy of user portrait construction according to the judgment result. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 A flowchart of a marketing data analysis method based on big data portrait provided in an embodiment of the present application;

[0032] Figure 2 A schematic diagram of the change in the individual portrait similarity coefficient provided in an embodiment of the present application. DETAILED DESCRIPTION

[0033] The embodiment of the present application solves the problem of untimely updating of user portrait data in marketing data in the prior art by providing a marketing data analysis method based on big data portraits. The method collects video portrait data within a preset time period, obtains a portrait similarity coefficient based on the collected video portrait data, obtains a similarity threshold from a preset database, and compares the obtained similarity threshold with the portrait similarity coefficient to obtain a comparison result. The data processing delay and data transmission delay are obtained through the log records of the short video platform to be analyzed, and reference data is obtained from the preset database to obtain a data acquisition delay index. The data acquisition delay index is then compared with the data acquisition delay index threshold obtained from the preset database to obtain a judgment result. Finally, the current user portrait update frequency is obtained from the log records of the short video platform to be analyzed, and the average user portrait update frequency is obtained from the preset database to obtain a user portrait update frequency index. At the same time, the user portrait update frequency index is compared with the user portrait update frequency index threshold obtained from the preset database and update feedback is performed, thereby improving the timeliness of updating user portrait data in marketing data.

[0034] The technical solution in the embodiment of the present application is to solve the problem of untimely update of user portrait data in the above marketing data. The overall idea is as follows:

[0035] By collecting video portrait data within a preset time period and obtaining a portrait similarity coefficient based on the collected video portrait data, and then comparing the portrait similarity coefficient with the similarity threshold obtained from a preset database to obtain a comparison result, and finally adopting a portrait data collection strategy based on the comparison result, the timeliness of updating user portrait data in marketing data is improved.

[0036] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0037] like Figure 1As shown, it is a flowchart of a marketing data analysis method based on big data portrait provided by an embodiment of the present application, and the method includes the following steps: collecting video portrait data within a preset time period, and obtaining a portrait similarity coefficient based on the collected video portrait data, the video portrait data comes from the short video platform to be analyzed, the video portrait data includes feedback portrait data and user portrait data, the feedback portrait data is used to reflect the actual user portrait of the video that the user is interested in, the user portrait data is used to reflect the user portrait of the video that the user is interested in constructed based on the big data portrait, and the portrait similarity coefficient is used to measure the similarity between the current user portrait data and the feedback portrait data; comparing the portrait similarity coefficient with the similarity threshold obtained from a preset database to obtain a comparison result, the similarity threshold is used to describe the minimum requirements that the current user portrait data and the feedback portrait data meet, and the comparison results include whether the portrait meets the standard or not; adopting a portrait data collection strategy based on the comparison result, and the portrait data collection strategy is used to improve the timeliness of the user portrait.

[0038] In this embodiment, the preset time period refers to the time range for data collection and analysis. Within this time range, the short video platform to be analyzed (such as Douyin, Bilibili, Kuaishou, Xiaohongshu, etc.) will collect relevant data for constructing user portraits for processing, ensuring that the user portrait is generated based on the latest data and can reflect the user's current interests and behaviors. Among them, the portrait similarity coefficient is a metric used to measure the similarity between the current user portrait data (user interest video portrait constructed based on big data) and the feedback portrait data (the portrait reflected by the video that the user is actually interested in). By calculating the portrait similarity coefficient, the short video platform to be analyzed can determine whether the current user portrait accurately reflects the user's true interests; the feedback portrait data is portrait data generated based on the user's actual behavior (such as viewing history, likes, comments, etc.), reflecting the type and characteristics of videos that the user is actually interested in. Provide a direct reflection of the user's real interests as a comparison benchmark to ensure that the construction of the user portrait is accurate; user portrait data is a predictive portrait constructed based on big data analysis, reflecting the types and features of videos that the user may be interested in, and is used to predict the user's interests and provide personalized recommendation services; through the method provided in this embodiment, the short video platform can continuously collect and update the user's portrait data within a certain period of time, monitor the accuracy of the user portrait in real time, and take corresponding measures to make adjustments based on the degree of match between the portrait and the user's actual interests, thereby improving the timeliness of the update of user portrait data in marketing data.

[0039] Furthermore, the process of obtaining video portrait data is as follows: feedback portrait data is collected through a portrait questionnaire fed back to the users to be investigated in the short video platform to be analyzed within a preset time period, and the obtained feedback portrait data is transmitted to a preset database for storage, wherein the feedback portrait data includes feedback video tags and feedback video duration, wherein the feedback video tags indicate the types of videos of interest fed back by the users through the portrait questionnaire, and the feedback video duration indicates the duration of the videos of interest fed back by the users through the portrait questionnaire, and the portrait questionnaire is used to obtain the users' actual preferences for watching videos; user portrait data of the users to be investigated are obtained from the short video platform to be analyzed, wherein the user portrait data includes user video tags and user video duration, and the user video tags indicate the types of videos of interest fed back by the short video platform to be analyzed based on the users' video preferences. The video viewing log is a label for the type of videos that the user is interested in watching added to the user portrait. The user video duration indicates the duration of the videos that the short video platform to be analyzed is interested in watching and added to the user portrait based on the user's video viewing log. The video viewing log includes viewing history, likes and comments, video sharing times and search records. The content category of the user portrait data is consistent with the feedback portrait data. The hash algorithm is used to obtain the video type similarity between the feedback video label and the user video label, and the hash algorithm is used to obtain the video duration similarity between the feedback video duration and the user video duration. The video type similarity indicates the similarity between the user video label and the feedback video label, and the video duration similarity indicates the similarity between the user video duration and the feedback video duration. The hash algorithm is used to obtain the similarity of video portrait data of the same type.

[0040] In this embodiment, the process of acquiring video portrait data through user portrait questionnaires and short video platform data is explained, wherein the feedback portrait data is collected by sending a portrait questionnaire to the users to be surveyed within a preset time period, and the data includes the user's feedback on the label (type) and duration of the video of interest. For example, if a video type that the user is interested in is sports, then sports is a label. In addition, in actual applications, video labels also include beauty, life, study, news, etc.; the user portrait data is extracted from the user video viewing log of the short video platform, and the data includes the video of interest label and video duration generated by the short video platform to be analyzed based on user behavior. For example, if the log shows that the user is interested in beauty short videos, then beauty is recorded as a label of the user; then the hash algorithm is used to calculate the video type similarity between the feedback video label and the user video label, and the video duration similarity between the feedback video duration and the user video duration. The hash algorithm compares the similarity between different data by converting the data into hash values. The method of this embodiment combines user feedback portrait data and platform user portrait data, calculates the similarity using a hash algorithm, and quantifies the degree of match between the two; this method can provide data support for marketing optimization and user portrait analysis.

[0041] Specifically, a user portrait questionnaire containing video type labels and video duration is sent to the users to be surveyed. After the users fill out the portrait questionnaire, the questionnaire is submitted to the short video platform to be analyzed. The short video platform to be analyzed then performs statistical analysis to obtain feedback video labels and feedback video duration. At the same time, the short video platform to be analyzed checks the users' viewing logs, counts the types of videos watched by users, labels each type to obtain user video labels, and counts the duration of videos watched by users to obtain feedback video duration.

[0042] Specifically, a hash algorithm is an algorithm that converts input data into a fixed-length output. Here, it is used to convert video labels and duration data into hash values, and then compare these hash values ​​to evaluate similarity. The calculated similarity reflects the degree of match between the portrait data fed back by users through surveys and the portrait data generated by the platform; for example, suppose there are two video data: Video A is labeled ["Music", "Popular", "MTV"], with a corresponding duration of 210 seconds, and Video B is labeled ["Music", "Rock", "MV"], with a corresponding duration of 205 seconds. The video labels and duration data are converted into a string, and as the input of the hash algorithm, the video A string is "Music-Popular-MTV-210", and the video B string is "Music-Rock-MV-205"; then, the string is applied to a hash algorithm, such as SHA-256, and the hash algorithm converts each string into a fixed-length hash value, and then the hash value of Video A and the hash value of Video B are obtained; finally, these hash values ​​are compared to evaluate the similarity of the two videos. In a hash algorithm, even if the input data is slightly different, the generated hash value will usually be very different. Therefore, if two hash values ​​are very similar or even the same, it means that the two videos may be very close in terms of label and duration data; in this example, the hash value difference is large, indicating that video A and video B are significantly different in label and duration.

[0043] Furthermore, the process of obtaining the portrait similarity coefficient is as follows: obtaining evaluation data from a preset database, the evaluation data including a video type evaluation threshold and a video duration evaluation threshold, the video type evaluation threshold indicating the minimum video type similarity within the evaluation range required for video marketing for video type similarity, and the video duration evaluation threshold indicating the minimum video duration similarity within the evaluation range required for video marketing for video duration similarity; numbering the users to be investigated, and obtaining individual portrait similarity coefficients based on video type similarity and video duration similarity, obtaining portrait similarity coefficients based on individual portrait similarity coefficients, the individual portrait similarity coefficient indicating the portrait similarity coefficient of a single user to be investigated, and the portrait similarity coefficient is calculated using the following formula:

[0044]

[0045] Where f is the number of users to be investigated, f = 1, 2,,, F, F is the total number of users to be investigated, VTS f VDS represents the video type similarity of the fth user to be investigated. f represents the video duration similarity of the fth user to be investigated, μ represents the video type evaluation threshold, ρ represents the video duration evaluation threshold, ISC f represents the individual portrait similarity coefficient of the fth user to be investigated, and PSC represents the portrait similarity coefficient of the user to be investigated in the current preset time period in the short video platform to be analyzed.

[0046] In this embodiment, the algorithm combines the evaluation data, video type similarity and video duration similarity for comprehensive analysis to obtain the individual portrait similarity coefficient, and then obtains the portrait similarity coefficient based on the individual portrait similarity coefficient; wherein, each variable is independent but jointly affects the changing trend of the individual portrait similarity coefficient, and due to the division of the evaluation data, the expression of the algorithm is divided into two cases, when the video type similarity is less than the video type evaluation threshold, or the video duration similarity is less than the video duration evaluation threshold, it is the first case, if the video type similarity is not less than the video type evaluation threshold, and the video duration similarity is not less than the video duration evaluation threshold, it is the second case, the two cases obtain the value of the individual portrait similarity coefficient by different methods, the first case obtains the individual portrait similarity coefficient according to the formula, wherein the video type similarity and the video duration similarity are positively correlated with the individual portrait similarity coefficient, as shown in Figure 2 As shown, Figure 2 Schematic diagram of the change of the similarity coefficient of individual user portraits. The figure shows that with the increase of the values ​​of video type similarity and video duration similarity, the image shows a gradually rising trend and approaches the maximum value of 1, indicating that the value of the individual portrait similarity coefficient gradually increases and tends to 1; it can be obtained that the larger the values ​​of video type similarity and video duration similarity, the closer the user portrait constructed by the platform is to the user's actual interest in watching videos, then the larger the value of the individual portrait similarity coefficient, the larger the value of the portrait similarity, and the more conducive to the video marketing of the short video platform to be analyzed. Therefore, the size of the portrait similarity coefficient is conducive to accurately and efficiently judging whether the current video marketing situation meets the requirements of video marketing, and timely taking video marketing measures to improve the timeliness of user portraits.

[0047] Specifically, assuming that the data of ten users to be investigated are analyzed, and assuming that the video type evaluation threshold is 1.5, the video duration evaluation threshold is also 1.5, and the value range of video type similarity and video duration similarity is between 0 and 2, then the portrait similarity coefficient data change table obtained according to the formula is shown in Table 1:

[0048] Table 1. Data change table of image similarity coefficient

[0049]

[0050] It can be seen from the table that as the video type similarity and video duration similarity increase, the individual portrait similarity coefficient also increases. Correspondingly, the larger the individual portrait similarity coefficient, the larger the portrait similarity coefficient. Among them, when the video type similarity is not less than the video type evaluation threshold of 1.5, and the video duration similarity is not less than the video duration evaluation threshold of 1.5, the individual portrait similarity coefficient reaches the maximum value of 1; however, if the two independent variables are not less than the corresponding thresholds at the same time, the maximum value cannot be reached. For example, the two rows of data numbered 8 and 9 for the user to be investigated, in the 8th row of data, although the video type similarity is not less than 1.5, the individual portrait similarity coefficient reaches the maximum value of 1. The similarity coefficient of individual portraits cannot reach the maximum value of 1. Similarly, in the 9th row of data, the video type similarity is less than the video type evaluation threshold of 1.5, but the video length similarity is not less than the video length evaluation threshold of 1.5. Finally, the individual portrait similarity coefficient still cannot reach the maximum value of 1. It can be seen that when analyzing the portrait similarity coefficient, the individual portrait similarity coefficient is mainly analyzed. When analyzing the individual portrait similarity coefficient, the consideration of the two factors of video type similarity and video length similarity cannot be ignored.

[0051] Specifically, the video type evaluation threshold is obtained from a preset database. In a specific embodiment, the video type evaluation threshold is generally the value of the video type similarity corresponding to the video conversion rate reaching the standard. For example, if the requirement for the video conversion rate to reach the standard is that the video type similarity reaches 0.85, then 0.85 can be used as the video type evaluation threshold.

[0052] Specifically, the video duration evaluation threshold is obtained from a preset database. In a specific embodiment, the video duration evaluation threshold is generally the value of the video duration similarity corresponding to when the video conversion rate meets the standard. For example, if the requirement for the video conversion rate to meet the standard is that the video type similarity reaches 0.90, then 0.90 is the video type evaluation threshold.

[0053] Furthermore, the process of obtaining the comparison result is as follows: a similarity threshold is obtained from a preset database, and the obtained similarity threshold is compared with the portrait similarity coefficient: if the portrait similarity coefficient is not less than the similarity threshold, the comparison result is that the portrait meets the standard; if the portrait similarity coefficient is less than the similarity threshold, the comparison result is that the portrait does not meet the standard.

[0054] In this embodiment, the portrait similarity coefficient represents the degree of match between the user portrait and the feedback portrait, and the coefficient is calculated for two dimensions: video type and video duration. The system extracts a similarity threshold from a preset database, and then compares this threshold with the portrait similarity coefficient calculated previously. If the portrait similarity coefficient is not less than the threshold, it means that the similarity between the user portrait and the user's self-reported preferences has reached or exceeded the minimum standard, and such a portrait is considered to be "up to standard". If the portrait similarity coefficient is less than the threshold, it means that the similarity between the user portrait and the user's self-reported preferences is lower than the standard, and such a portrait is considered to be "not up to standard". By comparing the calculated portrait similarity coefficient with the preset similarity threshold, the short video platform to be analyzed can determine whether the user portrait meets the standard. This judgment process helps to ensure the accuracy of the user portrait, thereby improving the accuracy of content recommendations, user experience, and the overall service quality of the platform.

[0055] Specifically, the similarity threshold is obtained from a preset database. In a specific embodiment, the similarity threshold is generally the minimum requirement of the user portrait similarity coefficient required by marketing in past successful marketing activities based on historical data analysis. For example, if a box plot is made based on the portrait similarity coefficient in successful marketing activities in the past year, the similarity threshold is represented by the difference between the first quartile of the box plot and 1.5 times the interquartile range.

[0056] Furthermore, the specific process of adopting the portrait data collection strategy based on the comparison result is as follows: if the comparison result is that the portrait meets the standards, indicating that the similarity between the current user portrait and the actual user portrait meets the video marketing requirements, then the portrait data collection strategy adopted is to continue to use the current user portrait; if the comparison result is that the portrait does not meet the standards, indicating that the similarity between the current user portrait and the actual user portrait does not meet the video marketing requirements, then the portrait data collection strategy adopted is to judge the impact of the current data collection delay on the construction of the user portrait based on the data collection delay index. The data collection delay indicates the time delay for collecting and processing user portrait data when constructing a user portrait based on big data portraits. The data collection delay includes data processing delay and data transmission delay. The data collection delay index is used to quantify the impact of data collection delay on video portrait data collection.

[0057] In this embodiment, the process of adjusting the user portrait data collection strategy according to the comparison results, especially the processing method when judging whether the current portrait meets the video marketing requirements, when the comparison result shows that the user portrait meets the standards, it means that the similarity between the current user portrait and the actual user portrait has met the requirements of video marketing. In this case, the short video platform to be analyzed believes that the current data collection and processing methods are effective, and therefore chooses to continue using the existing user portrait without the need for additional data collection or adjustment; when the comparison result shows that the user portrait does not meet the standards, it means that there is a large inconsistency between the current user portrait and the actual user portrait, which cannot meet the requirements of video marketing. At this time, the system needs to take further measures to improve the accuracy of the portrait; by adjusting the portrait data collection strategy according to the comparison results (meeting the standards or not), the short video platform to be analyzed can ensure the accuracy of the user portrait, and ultimately improve the quality of the user portrait, thereby improving the effect of video marketing.

[0058] It should be understood that data collection delay refers to the time delay from data collection to completion of processing when building a user portrait. It includes two parts: data processing delay is the time required to process data, such as preprocessing of raw data; data transmission delay is the time required to transfer data from one point to another, such as from the user device to the server, and then to the database. The data collection delay index is an indicator used to quantify the impact of data collection delay on user portrait construction, which can help evaluate whether the delay has led to inaccurate portraits.

[0059] Furthermore, the process of obtaining the data collection delay index is as follows: obtaining data processing delay and data transmission delay through the log records of the short video platform to be analyzed, the data processing delay represents the time delay generated when the short video platform to be analyzed processes video portrait data, the data transmission delay represents the time delay generated when the short video platform to be analyzed transmits video portrait data, and the log records are used to record the real-time data of data processing and transmission of the short video platform to be analyzed; obtaining reference data from a preset database, the reference data includes average data processing delay, average data transmission delay, processing delay evaluation value and transmission delay evaluation value, the average data processing delay represents the average time taken by the short video platform to be analyzed to process video portrait data, the average data transmission delay represents the average time taken by the short video platform to be analyzed to transmit video portrait data, the processing delay evaluation value represents the maximum value of the data processing delay allowed, and the transmission delay evaluation value represents the maximum value of the data transmission delay allowed; obtaining an individual delay index based on the data collection delay and the reference data, and then obtaining a data collection delay index based on the individual delay index, the individual delay index represents the data collection delay index of a single user to be investigated, and the data collection delay index is calculated using the following formula:

[0060]

[0061] In the formula, f represents the number of users to be investigated, f=1,2,,,F, F represents the total number of users to be investigated, CD f represents the data processing delay of the fth user to be investigated, represents the average data processing delay, α represents the processing delay evaluation value, TD f represents the data transmission delay of the fth user to be investigated, represents the average data transmission delay, β represents the transmission delay evaluation value, and IDI f represents the individual delay index of the fth user to be investigated, and DCI represents the data collection delay index within a preset time period.

[0062] In this embodiment, the algorithm combines the data collection delay and the reference data for comprehensive analysis to obtain the individual delay index, and then obtains the data collection delay index based on the individual delay index; in the formula, each variable is independent but jointly affects the change trend of the individual delay index, and due to the division of the reference data, the expression of the algorithm is divided into two cases, when the data processing delay is not less than the processing delay evaluation value, or the data transmission delay is not less than the transmission delay evaluation value, it is the first case, if the data processing delay is less than the processing delay evaluation value, and the data transmission delay is less than the transmission delay evaluation value, it is the second case, the two cases obtain the value of the individual delay index by different methods, the first case obtains the individual delay index according to the formula, in the formula, the data processing delay and the data transmission delay are positively correlated with the individual delay index, with the increase of the absolute value of the difference between the data processing delay and the average data processing delay and the increase of the absolute value of the difference between the data transmission delay and the average data transmission delay, the individual delay index increases, and the second case can directly obtain the value of the individual delay index as 0; it can be obtained that reducing the data processing delay and the data transmission delay is conducive to building a more accurate user portrait and formulating appropriate video marketing measures.

[0063] Specifically, the average data processing delay is represented by the average value of the data processing delay of the historical data in the log record.

[0064] Specifically, the average data transmission delay is represented by the average value of the data transmission delay of the historical data in the log record.

[0065] Specifically, the processing delay evaluation value is obtained from a preset database. In a specific embodiment, the processing delay evaluation value is generally the maximum limit of the concentrated distribution of data processing delays in successful marketing activities. It is assumed that statistical analysis finds that in successful marketing activities, a corresponding box plot is drawn according to the data processing delay, and then the processing delay evaluation value is represented by the sum of the third quartile of the box plot and 1.5 times the interquartile range.

[0066] Specifically, the transmission delay evaluation value is obtained from a preset database. In a specific embodiment, the transmission delay evaluation value is generally the maximum limit of the concentrated distribution of data transmission delays in successful marketing activities. It is assumed that statistical analysis finds that in successful marketing activities, a corresponding box plot is drawn according to the data transmission delay, and then the transmission delay evaluation value is represented by the sum of the third quartile of the box plot and 1.5 times the interquartile range.

[0067] Furthermore, a data collection delay index is obtained, and then the data collection delay index is compared with a data collection delay index threshold obtained from a preset database to obtain a judgment result, as follows: the data collection delay index threshold is used to judge the impact of the current data collection delay on the portrait construction, and the judgment result includes normal data collection delay and abnormal data collection delay; the specific method for obtaining the judgment result is as follows: if the data collection delay index is not greater than the data collection delay index threshold, the judgment result is that the data collection delay is normal; if the data collection delay index is greater than the data collection delay index threshold, the judgment result is that the data collection delay is abnormal.

[0068] In this embodiment, the impact of data collection delay on user portrait construction is determined by comparing the data collection delay index with the data collection delay index threshold, and by comparing the data collection delay index with the data collection delay index threshold, it is possible to accurately evaluate whether the current delay is within an acceptable range; this evaluation helps to promptly identify potential problems and ensure the timeliness and accuracy of data collection; then, based on clear threshold standards, the system can automatically and accurately determine whether the delay will affect user portrait construction, which reduces the interference of subjective judgment and improves the degree of automation of data processing; the process of this embodiment ensures the accuracy and timeliness of user portraits, thereby supporting more accurate video recommendations and marketing strategies.

[0069] Specifically, the data collection delay index threshold is obtained from a preset database. In a specific embodiment, the data collection delay index threshold is generally the mode of the data collection delay index in successful marketing activities. For example, if it is found based on historical data statistical analysis that in successful marketing activities, the mode of the data collection delay index is 0.95, then the data collection delay index threshold can be set to 0.95.

[0070] Furthermore, a judgment result is obtained, and then feedback on the judgment result is included. The specific process is as follows: if the judgment result is that the data collection delay is normal, indicating that the current data collection delay does not affect the data collection for portrait construction, then a portrait update judgment is performed based on the user portrait update frequency, and the portrait update judgment is used to determine whether the user portrait update frequency has an impact on video marketing; if the judgment result is that the data collection delay is abnormal, indicating that the current data collection delay affects the data collection for portrait construction, then feedback is given to the preset analyst that the factor causing the portrait similarity coefficient to not meet the video marketing requirements is the data collection delay, and a command prompt for reducing the data collection delay is sent to the preset analyst.

[0071] In this embodiment, it is described how to provide feedback on the results and take corresponding measures after judging whether the data collection delay is normal. By judging whether the data collection delay is normal and adjusting the user portrait update frequency according to the judgment result, it can be ensured that the portrait data is synchronized with the user behavior, thereby improving the accuracy and effectiveness of video recommendations; through the automated judgment and feedback process, the system can quickly and accurately identify and handle data collection delay problems, reduce human intervention, and improve the efficiency and reliability of data processing; through the dynamic judgment and feedback mechanism, the short video platform to be analyzed can ensure the timely update of user portraits, optimize video marketing strategies, and solve delay problems in a timely manner. This mechanism not only improves the degree of automation of data processing, but also ensures the accuracy of user portraits and the effectiveness of video marketing.

[0072] Furthermore, the specific steps for judging the update of user portraits based on the update frequency of user portraits are as follows: Step 1, obtain the current user portrait update frequency from the log records of the short video platform to be analyzed, and obtain the average user portrait update frequency from the preset database. The user portrait update frequency represents the frequency with which the short video platform to be analyzed updates the user portrait in the current preset time period, and the average user portrait update frequency represents the average value of the user portrait update frequency of the short video platform to be analyzed; Step 2, obtain the update frequency evaluation value from the preset database, and obtain the individual user portrait update frequency index based on the user portrait update frequency, the average user portrait update frequency and the update frequency evaluation value, and calculate the update frequency index based on the individual user portrait. The update frequency index obtains the user portrait update frequency index. The update frequency evaluation value indicates the minimum value of the evaluation range within which the user portrait update frequency can meet the requirements of video marketing. The individual user portrait update frequency index indicates the user portrait update frequency index of a single user to be investigated. The user portrait update frequency index is used to quantify the impact of the user portrait update frequency on video marketing. Step three: compare the user portrait update frequency index with the user portrait update frequency index threshold obtained from the preset database and provide update feedback. The user portrait update frequency index threshold is used to determine the impact of the current user portrait update frequency on video marketing. The user portrait update frequency index is calculated using the following formula:

[0073]

[0074] Where f is the number of users to be investigated, f = 1, 2,,, F, F is the total number of users to be investigated, UF f represents the update frequency of the user profile of the fth user to be investigated, represents the average user profile update frequency, θ represents the update frequency evaluation value, and IUI f It represents the update frequency index of the individual user portrait of the fth user to be investigated, and UFI represents the update frequency index of the user portrait of the short video platform to be analyzed within the preset time period.

[0075] In this embodiment, the algorithm combines the user portrait update frequency, the average user portrait update frequency and the update frequency evaluation value for comprehensive analysis to obtain the individual user portrait update frequency index, and then obtains the user portrait update frequency index based on the individual user portrait update frequency index; wherein, the user portrait update frequency independently affects the change trend of the individual user portrait update frequency index, and due to the division of the update frequency evaluation value, the expression of the algorithm is divided into two cases, when the user portrait update frequency is less than the update frequency evaluation value, it is the first case, and when the user portrait update frequency is not less than the update frequency evaluation value, it is the second case. The two cases derive the individual user portrait update frequency index through different methods. The value of the frequency index. In the first case, the individual user portrait update frequency index is obtained according to the formula, in which the user portrait update frequency is negatively correlated with the individual user portrait update frequency index. As the absolute value of the difference between the user portrait update frequency and the average user portrait update frequency increases, the individual user portrait update frequency index decreases. In the second case, it can be directly concluded that the value of the individual user portrait update frequency index is 0; if the value of the individual user portrait update frequency index is smaller, then the value of the user portrait update frequency index is also smaller. Therefore, the analysis of the user portrait update frequency index is helpful to monitor the user portrait update frequency and help improve the timeliness of user portrait updates.

[0076] Specifically, the average user portrait update frequency is calculated based on the actual number of updates within a period of time recorded in historical data. For example, assuming that the Douyin short video platform updates a user's portrait 5 times within a week, the average user portrait update frequency is 7 days / 5 intervals = 1.4 days / time.

[0077] Specifically, the update frequency evaluation value is a measurement indicator used to evaluate whether the actual update frequency meets the expected frequency. The update frequency evaluation value is usually set based on business needs and user behavior patterns. For example, assuming that the Douyin short video platform sets a target update frequency of 1.5 days / time based on business needs analysis, the update frequency evaluation value is 1.5 days / time.

[0078] Furthermore, the specific process of comparing the user portrait update frequency index with the user portrait update frequency index threshold obtained from the preset database and performing update feedback is as follows: if the user portrait update frequency index is not less than the user portrait update frequency index threshold, feedback is given to the preset analyst that there is no problem with data collection delay and user portrait update frequency; if the user portrait update frequency index is less than the user portrait update frequency index threshold, feedback is given to the preset analyst that there is a problem with the user portrait update frequency, and a command prompt for increasing the user portrait update frequency is sent to the preset analyst.

[0079] In this embodiment, it is described how to judge whether the user portrait update frequency meets the requirements by comparing the user portrait update frequency index with the user portrait update frequency index threshold, and to feedback and handle the problem based on the judgment result; the user portrait update frequency index is an indicator that measures the update speed of the user portrait, that is, the frequency with which the user portrait reflects the user behavior data. The higher this index is, the more timely the user portrait can reflect the user's latest behavior.

[0080] It is important to understand that if the user portrait update frequency index is not less than the preset user portrait update frequency index threshold, it means that the frequency of user portrait updates is high enough to accurately reflect the user's latest behavior. In this case, the system will provide feedback to the preset analyst, indicating that there is no problem with the update frequency of the user portrait, and there are no factors such as data collection delay and user portrait update frequency that cause the portrait similarity coefficient to fail to meet the video marketing requirements; this means that if there is a problem with the portrait similarity coefficient, the cause may be in other aspects.

[0081] It should be understood that if the user portrait update frequency index is less than the preset user portrait update frequency index threshold, it means that the portrait update is not timely enough, which may lead to inaccurate portraits; Feedback content: The system will feedback to the preset analyst, indicating that there are factors related to the user portrait update frequency, resulting in the portrait similarity coefficient not meeting the video marketing requirements; at the same time, the system will send a command prompt to increase the user portrait update frequency, and suggest that the analyst take measures to increase the portrait update speed step by step to speed up the portrait update speed to improve the accuracy of the portrait. Feedback to the preset analyst on specific issues can help them accurately locate the root cause of the problem and take targeted measures to solve the problem. For example, if it is confirmed that the portrait update frequency is too low, they can focus on improving the update speed without wasting time on other aspects. The system sends command prompts to remind analysts of the importance of portrait update frequency and guide them to focus on measures to increase the update frequency in subsequent operations; this helps ensure that the user portrait can reflect the user's latest behavior in a timely manner, thereby enhancing the accuracy of video marketing.

[0082] Specifically, the user portrait update frequency index threshold is a preset standard value used to determine whether the update frequency of the user portrait is within an acceptable range. This threshold is set based on historical data analysis or business needs. If the update frequency index is lower than this threshold, it means that the portrait update may not be timely enough, thus affecting the accuracy of the portrait; for example, suppose Company B is a short video e-commerce platform that needs to capture users' shopping behaviors in real time for accurate recommendations. After business needs analysis, the company decides that the user portrait should be updated every 6 hours, so 6 hours is set as the user portrait update frequency index threshold.

[0083] To sum up, the embodiment of the present application collects video portrait data within a preset time period, obtains a portrait similarity coefficient based on the collected video portrait data, and then compares the portrait similarity coefficient with a similarity threshold obtained from a preset database to obtain a comparison result. Finally, a portrait data collection strategy is adopted based on the comparison result, thereby realizing dynamic updating of user portraits, thereby improving the timeliness of updating user portrait data in marketing data, and effectively solving the problem of untimely updating of user portrait data in marketing data in the prior art.

[0084] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0085] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0086] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0087] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0088] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0089] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A marketing data analysis method based on big data portrait, characterized in that: The following steps are involved: Collecting video portrait data within a preset time period, and obtaining a portrait similarity coefficient based on the collected video portrait data, wherein the video portrait data includes feedback portrait data and user portrait data, and the portrait similarity coefficient is used to measure the similarity between the current user portrait data and the feedback portrait data; Comparing the portrait similarity coefficient with a similarity threshold obtained from a preset database to obtain a comparison result, wherein the comparison result includes whether the portrait meets the standard or not; A portrait data collection strategy is adopted according to the comparison result, and the portrait data collection strategy is used to improve the timeliness of the user portrait.

2. A marketing data analysis method based on big data portrait as claimed in claim 1, characterized in that: The process of obtaining the video image data is as follows: Collect feedback portrait data by sending a portrait questionnaire to the users to be investigated in the short video platform to be analyzed within a preset time period, and transmit the obtained feedback portrait data to a preset database for storage, wherein the feedback portrait data includes feedback video tags and feedback video duration; Obtaining user portrait data of the user to be investigated from the short video platform to be analyzed, wherein the user portrait data includes user video tags and user video duration, and the content category of the user portrait data is consistent with the feedback portrait data; The video type similarity between the feedback video tag and the user video tag is obtained through a hash algorithm, and the video duration similarity between the feedback video duration and the user video duration is obtained through a hash algorithm. The video type similarity indicates the similarity between the user video tag and the feedback video tag, and the video duration similarity indicates the similarity between the user video duration and the feedback video duration.

3. A marketing data analysis method based on big data portrait as claimed in claim 2, characterized in that: The process of obtaining the image similarity coefficient is as follows: Acquire evaluation data from a preset database, wherein the evaluation data includes a video type evaluation threshold and a video duration evaluation threshold; The users to be investigated are numbered, and the individual portrait similarity coefficients are obtained based on the video type similarity and the video duration similarity. The portrait similarity coefficient is obtained according to the individual portrait similarity coefficients. The individual portrait similarity coefficient represents the portrait similarity coefficient of a single user to be investigated. The portrait similarity coefficient is calculated using the following formula: Where f is the number of users to be investigated, f = 1, 2,,, F, F is the total number of users to be investigated, VTS f VDS represents the video type similarity of the fth user to be investigated. f represents the video duration similarity of the fth user to be investigated, μ represents the video type evaluation threshold, ρ represents the video duration evaluation threshold, ISC f represents the individual portrait similarity coefficient of the fth user to be investigated, and PSC represents the portrait similarity coefficient of the user to be investigated in the current preset time period in the short video platform to be analyzed.

4. A marketing data analysis method based on big data portrait as claimed in claim 3, characterized in that: The process of obtaining the comparison result is as follows: Get the similarity threshold from the preset database and compare the obtained similarity threshold with the portrait similarity coefficient: If the portrait similarity coefficient is not less than the similarity threshold, the comparison result is that the portrait meets the standard; If the portrait similarity coefficient is less than the similarity threshold, the comparison result is that the portrait does not meet the standard.

5. A marketing data analysis method based on big data portrait as claimed in claim 4, characterized in that: The specific process of adopting the portrait data collection strategy according to the comparison results is as follows: If the comparison result shows that the profile meets the requirements, the current user profile will continue to be used; If the comparison result is that the portrait does not meet the standard, the impact of the current data collection delay on the construction of the user portrait is judged based on the data collection delay index. The data collection delay includes data processing delay and data transmission delay. The data collection delay index is used to quantify the impact of data collection delay on video portrait data collection.

6. A marketing data analysis method based on big data portrait as claimed in claim 5, characterized in that: The process of obtaining the data collection delay index is as follows: The data processing delay and the data transmission delay are obtained through the log records of the short video platform to be analyzed. The data processing delay represents the time delay generated by the short video platform to be analyzed when processing the video portrait data. The data transmission delay represents the time delay generated by the short video platform to be analyzed in the process of transmitting the video portrait data. Acquire reference data from a preset database, the reference data including an average data processing delay, an average data transmission delay, a processing delay evaluation value, and a transmission delay evaluation value; An individual delay index is obtained based on the data collection delay and the reference data, and then a data collection delay index is obtained based on the individual delay index. The individual delay index represents the data collection delay index of a single user to be investigated, and the data collection delay index is calculated using the following formula: In the formula, f represents the number of users to be investigated, f=1,2,,,F, F represents the total number of users to be investigated, CD f represents the data processing delay of the fth user to be investigated, represents the average data processing delay, α represents the processing delay evaluation value, TD f represents the data transmission delay of the fth user to be investigated, represents the average data transmission delay, β represents the transmission delay evaluation value, and IDI f represents the individual delay index of the fth user to be investigated, and DCI represents the data collection delay index within a preset time period.

7. A marketing data analysis method based on big data portrait as claimed in claim 6, characterized in that: The data collection delay index is obtained, and then the data collection delay index is compared with the data collection delay index threshold obtained from the preset database to obtain a judgment result, which is specifically as follows: The judgment result includes normal data collection delay and abnormal data collection delay; The specific method for obtaining the judgment result is as follows: If the data collection delay index is not less than the data collection delay index threshold, the result is that the data collection delay is normal; If the data collection delay index is less than the data collection delay index threshold, the judgment result is that the data collection delay is abnormal.

8. A marketing data analysis method based on big data portrait as claimed in claim 7, characterized in that: The determination result is obtained, and then the determination result is fed back. The specific process is as follows: If the result of the judgment is that the data collection delay is normal, a portrait update judgment is performed based on the user portrait update frequency, and the portrait update judgment is used to determine whether the user portrait update frequency has an impact on video marketing; If the judgment result is that the data collection delay is abnormal, the factor that causes the portrait similarity coefficient to fail to meet the video marketing requirements is the data collection delay, and a command prompt for reducing the data collection delay is sent to the preset analyst.

9. A marketing data analysis method based on big data portrait as claimed in claim 8, characterized in that: The specific steps of determining the update of the user portrait according to the update frequency of the user portrait are as follows: Step 1: Obtain the current user portrait update frequency from the log records of the short video platform to be analyzed, and obtain the average user portrait update frequency from a preset database, wherein the user portrait update frequency indicates the frequency with which the short video platform to be analyzed updates the user portrait in the current preset time period; Step 2: Obtain an update frequency evaluation value from a preset database, and obtain an individual user portrait update frequency index based on the user portrait update frequency, the average user portrait update frequency and the update frequency evaluation value, and obtain a user portrait update frequency index based on the individual user portrait update frequency index. The individual user portrait update frequency index represents the user portrait update frequency index of a single user to be investigated, and the user portrait update frequency index is used to quantify the impact of the user portrait update frequency on video marketing. Step 3: compare the user portrait update frequency index with the user portrait update frequency index threshold obtained from a preset database and provide update feedback; The user portrait update frequency index is calculated using the following formula: Where f is the number of users to be investigated, f = 1, 2,,, F, F is the total number of users to be investigated, UF f represents the update frequency of the user profile of the fth user to be investigated, represents the average user profile update frequency, θ represents the update frequency evaluation value, and IUI f It represents the update frequency index of the individual user portrait of the fth user to be investigated, and UFI represents the update frequency index of the user portrait of the short video platform to be analyzed within the preset time period.

10. A marketing data analysis method based on big data portrait as claimed in claim 9, characterized in that: The specific process of comparing the user portrait update frequency index with the user portrait update frequency index threshold obtained from the preset database and performing update feedback is as follows: If the user portrait update frequency index is not less than the user portrait update frequency index threshold, feedback is given to the preset analyst that there is no problem with data collection delay and user portrait update frequency; If the user portrait update frequency index is less than the user portrait update frequency index threshold, a command prompt for increasing the user portrait update frequency is sent to the preset analyst.

Images