Big data-based live broadcast e-commerce content management method and system

Abstract

The invention relates to the field of network content management, and discloses a live broadcast e-commerce content management method and system based on big data, and the system comprises a live broadcast content data collection module which collects live broadcast content data on a live broadcast e-commerce platform in real time through a plurality of data collection interfaces; preprocessing the live broadcast content data, and carrying out preprocessing operations of cleaning, duplicate removal and format conversion on the collected live broadcast content data; the live broadcast content analysis module analyzes the preprocessed live broadcast content data, the live broadcast content data comprises active content data and passive content data, and the active content data and the passive content data are integrated and analyzed to obtain an active violation index and a passive violation index; judging whether the live broadcast content is illegal or not according to the active violation index and the passive violation index; and the live broadcast content real-time monitoring module monitors the live broadcast process in real time according to a content analysis result, and takes corresponding management measures in time when discovering violation.

Description

Technical Field

[0001] The present invention relates to the field of network content management, and in particular to a live e-commerce content management method and system based on big data. Background Art

[0002] With the rapid development of Internet technology, live e-commerce, as an emerging business model, has gradually become an important part of the e-commerce industry. Live e-commerce combines product display, sales and interaction through real-time video broadcasting, greatly improving users' shopping experience and merchants' sales efficiency. However, as the scale of live e-commerce continues to expand, content management issues have become increasingly prominent.

[0003] Traditional live content management methods mainly rely on manual review and simple rule filtering, which are difficult to cope with massive live data and complex content scenarios. They are prone to problems such as lax content review and the spread of illegal content. These problems will not only harm the rights and interests of consumers, but also affect the healthy development of the live e-commerce industry. Therefore, how to quickly identify the live broadcast rooms that violate regulations among the numerous live broadcast rooms on the live e-commerce platform is a technical problem that needs to be solved urgently. Summary of the invention

[0004] The purpose of the present invention is to provide a live e-commerce content management method and system based on big data to solve the above technical problems.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] A live e-commerce content management system based on big data, including:

[0007] The live content data collection module is used to collect live content data on the live e-commerce platform in real time through multiple data collection interfaces;

[0008] Live content data preprocessing, which is used to perform preprocessing operations such as cleaning, deduplication and format conversion on the collected live content data;

[0009] A live content analysis module analyzes the pre-processed live content data, wherein the live content data includes active content data and passive content data, and obtains active violation indicators and passive violation indicators by analyzing the active content data and the passive content data, and then determines whether the live content violates the regulations based on the active violation indicators and the passive violation indicators;

[0010] The live broadcast content real-time monitoring module monitors the live broadcast process in real time based on the results of content analysis, and takes corresponding management measures in a timely manner when violations are found;

[0011] Among them, the active data includes the host's language expression and product introduction content, and the passive data includes the audience's language expression.

[0012] As a further technical solution, the live content analysis module includes:

[0013] Active content analysis unit, used to perform natural language processing and semantic analysis on the host's language expression and product introduction content to identify whether there are any illegal words, sensitive information or false propaganda;

[0014] Passive content analysis unit, used to detect violations of audience language expressions and identify whether there are advertisements or inappropriate remarks;

[0015] The violation index calculation unit is used to calculate the active violation index and the passive violation index according to the results of the active content analysis unit and the passive content analysis unit, and comprehensively judge whether the live broadcast content violates the regulations.

[0016] As a further technical solution, the process of calculating the active violation index is:

[0017] S1. Data processing:

[0018] Convert the host's voice into text, perform word segmentation on the extracted text, build a stop word list, and remove stop words from the word segmentation results;

[0019] S2. Feature extraction:

[0020] Build a library of illegal words that include false propaganda words and sensitive information words, traverse the text after word segmentation, and count the number of illegal words that match;

[0021] Use pre-trained BERT or GPT language models to perform semantic analysis to determine whether the text contains false advertising or exaggerated efficacy; use the TF-IDF algorithm to extract key product information and core expressions in the text as keywords;

[0022] S3. Calculation of violation indicators:

[0023] Different weights are assigned to different types of illegal words, and the illegal word score S is calculated based on the matched illegal words and their weights. word ; The expression is:

[0024] Where n is the number of words in the illegal vocabulary, α i is the weight coefficient of the i-th illegal word, C i is the number of times the illegal word is matched;

[0025] According to the results of semantic analysis, if it is determined that there is a semantic violation, a fixed semantic violation score S is given. yu ;

[0026] By formula: A = S word *β 1 +S yu *β 2 The active violation index A is calculated; where β 1 , β 2 is the weight coefficient, β 1 +β 2 =1.

[0027] As a further technical solution, the process of calculating the passive violation index is:

[0028] The language expression text data of the audience in the chat area during the live broadcast is collected in real time and processed by word segmentation;

[0029] Traverse the audience comments after word segmentation and count the number of illegal words matched as N f ;

[0030] By formula: The passive violation index P is calculated; where N span is the screen-sweeping score, m is the number of words in the comment, γ 1 , γ 2 is the weight coefficient, γ 1 +γ 2 =1.

[0031] As a further technical solution, the process of obtaining the screen-swiping score is as follows:

[0032] Get the number of word segments M of the comment, count the frequency K of comments sent by the same viewer within the set time period, and substitute into the following formula:

[0033]

[0034] Calculate the screen-swiping score N span , where M th is the word count threshold for comments, K th The frequency threshold for the same viewer to send comments within a set period of time.

[0035] As a further technical solution, the process of judging whether the live broadcast content violates the law based on the active violation indicators and the passive violation indicators is as follows:

[0036] By formula: F = (σ 1 *A+σ 2 *P) D The comprehensive violation assessment index F is calculated;

[0037] Among them, σ 1 , σ 2 is the weight coefficient, σ 1 +σ 2 =1, D is the historical impact coefficient;

[0038] The calculated comprehensive violation assessment index F is compared with the preset assessment threshold F T To compare:

[0039] If F>F T , then the live broadcast content is judged to be illegal; if F≤F T , then the live broadcast content is judged not to be in violation of regulations.

[0040] As a further technical solution, the calculation formula of the historical impact coefficient is:

[0041]

[0042] Among them, M sub-periods are obtained continuously according to the preset interval length, j is the jth sub-period, F j is the average value of the comprehensive violation evaluation index in the jth sub-period, θ j is the impact factor of the live broadcast stage; the expected total live broadcast period is divided into three stages: opening, mid-term and ending, and the impact factor of the opening stage is greater than the impact factor of the mid-term stage and greater than the impact factor of the ending stage. max 、F min are the maximum and minimum values ​​of the comprehensive violation assessment indicators in the M sub-periods, respectively.

[0043] As a further technical solution, the system further includes:

[0044] The live content data storage module is used to store the pre-processed live content data in the big data storage module to provide an analysis basis for subsequent content management.

[0045] A live e-commerce content management method based on big data, the steps are as follows:

[0046] S100, collect live content data on the live e-commerce platform in real time through multiple data collection interfaces;

[0047] S200, performing pre-processing operations of cleaning, deduplication and format conversion on the collected live content data;

[0048] S300, analyzing the pre-processed live content data, wherein the live content data includes active content data and passive content data, analyzing the active content data and the passive content data to obtain active violation indicators and passive violation indicators, and then judging whether the live content violates the regulations based on the active violation indicators and the passive violation indicators;

[0049] S400. Based on the results of content analysis, the live broadcast process is monitored in real time, and when violations are found, corresponding management measures are taken in a timely manner.

[0050] Beneficial effects of the present invention:

[0051] The present invention analyzes the pre-processed live content data through a live content analysis module, and divides the live content data into two categories: active content data and passive content data, so as to provide a basis for further processing after the live broadcast violation is determined. For example, if it is an active content violation, the anchor is processed, and if it is a passive content violation, the audience is processed. Then, the active content data and the passive content data are analyzed comprehensively to obtain active violation indicators and passive violation indicators, and then it is judged whether the live broadcast content has violated the regulations based on the active violation indicators and the passive violation indicators. Therefore, it is possible to comprehensively judge whether the live broadcast content is compliant from the two aspects of the audience and the anchor, so as to achieve the purpose of quickly locating the live broadcast room with illegal behaviors, and reduce the probability of erroneously processing the anchor due to the illegal behaviors of the audience. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The present invention will be further described below in conjunction with the accompanying drawings.

[0053] Figure 1 is a system structure diagram of the present invention;

[0054] Figure 2 It is a diagram of the method steps of the present invention. DETAILED DESCRIPTION

[0055] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] See also Figure 1 As shown, the present invention is a live e-commerce content management system based on big data, comprising:

[0057] The live content data collection module is used to collect live content data on the live e-commerce platform in real time through multiple data collection interfaces;

[0058] Live content data preprocessing, which is used to perform preprocessing operations such as cleaning, deduplication and format conversion on the collected live content data;

[0059] A live content analysis module analyzes the pre-processed live content data, wherein the live content data includes active content data and passive content data, and obtains active violation indicators and passive violation indicators by analyzing the active content data and the passive content data, and then determines whether the live content violates the regulations based on the active violation indicators and the passive violation indicators;

[0060] The live broadcast content real-time monitoring module monitors the live broadcast process in real time based on the results of content analysis, and takes corresponding management measures in a timely manner when violations are found;

[0061] Among them, the active data includes the host's language expression and product introduction content, and the passive data includes the audience's language expression.

[0062] The system further comprises:

[0063] The live content data storage module is used to store the pre-processed live content data in the big data storage module to provide an analysis basis for subsequent content management.

[0064] In this embodiment, the pre-processed live content data is analyzed by the live content analysis module, and the live content data is divided into two categories: active content data and passive content data, so as to provide a basis for further processing after the live broadcast violation is determined. For example, if it is an active content violation, the anchor will be processed, and if it is a passive content violation, the audience will be processed. The active content data and the passive content data are then analyzed to obtain active violation indicators and passive violation indicators, and then it is judged whether the live content violates the regulations based on the active violation indicators and the passive violation indicators. Therefore, it is possible to make a comprehensive judgment on whether the live content is compliant from the perspectives of the audience and the anchor, so as to achieve the purpose of quickly locating the live broadcast room with violations, and reduce the probability of incorrectly processing the anchor due to violations of audience behavior, so as to provide an accurate basis for taking corresponding management measures in a timely manner when violations are found.

[0065] The live content analysis module includes:

[0066] Active content analysis unit, used to perform natural language processing and semantic analysis on the host's language expression and product introduction content to identify whether there are any illegal words, sensitive information or false propaganda;

[0067] Passive content analysis unit, used to detect violations of audience language expressions and identify whether there are advertisements or inappropriate remarks;

[0068] The violation index calculation unit is used to calculate the active violation index and the passive violation index according to the results of the active content analysis unit and the passive content analysis unit, and comprehensively judge whether the live broadcast content violates the regulations.

[0069] The process of calculating the active violation indicator is:

[0070] S1. Data processing:

[0071] Convert the voice of the host into text, perform word segmentation on the extracted text, use the jieba word segmentation tool in the Chinese scenario, and use the NLTK library in the English scenario, both of which are existing technologies; then construct a stop word list and remove stop words from the word segmentation results; stop words refer to words that frequently appear in the text but are not substantially helpful for semantic understanding, such as "de", "le", "a", etc.; by removing stop words, the interference of irrelevant information on subsequent analysis can be reduced; a stop word list can also be constructed according to specific language characteristics and then stop words can be filtered out from the word segmentation results.

[0072] S2. Feature extraction:

[0073] Construct a violation vocabulary library containing false propaganda words and sensitive information words, traverse the word-segmented text, and count the number of matching violation words;

[0074] Adopt a pre-trained BERT or GPT language model for semantic analysis to determine whether there is false propaganda or exaggerated efficacy in the text; use the TF-IDF algorithm to extract key product information and core expressions in the text as keywords; the above semantic model and TF-IDF algorithm are both existing technologies and will not be elaborated;

[0075] S3. Violation index calculation:

[0076] Assign different weights to different types of violation words, and calculate the violation word score S according to the matching violation words and their weights word ; The expression is:

[0077] where n is the number of words in the violation vocabulary library, α i is the weight coefficient of the i-th violation word, and C i is the number of times the violation word is matched;

[0078] According to the result of semantic analysis, if a violation semantics is determined, a fixed semantic violation score S yu is given;

[0079] Through the formula: A = S word *β 1 + S yu *β 2 Calculate the active violation index A; where β 1 , β 2 are weight coefficients, and β 1 + β 2 = 1.

[0080] In this embodiment, during the live broadcast, the active content analysis unit can obtain the anchor's language expression and product introduction content in real time, and quickly analyze and process it, ensuring that the quantified active violation index value can be output in a short time, so that the anchor's violation behavior can be discovered in time, and an early warning can be issued immediately, providing strong support for the real-time management of the live broadcast platform; once the active violation index value exceeds the preset threshold, corresponding management measures can be taken quickly, such as suspending the live broadcast, warning the anchor, blocking related content, etc., through timely intervention, it can effectively avoid the further spread of illegal content, protect the rights and interests of consumers, and maintain the good order and image of the live broadcast platform; finally, the quantified active violation index value is used to objectively reflect the severity of the anchor's violation during the live broadcast. Compared with the traditional subjective judgment method, the quantitative evaluation is more scientific and accurate, and avoids the interference of human factors; the live broadcast platform can perform graded management of the anchor according to the index, warn and educate the anchors with less violations, and take more severe punishment measures for the anchors with serious violations.

[0081] The process of calculating the passive violation indicator is:

[0082] The language expression text data of the audience in the chat area during the live broadcast is collected in real time and processed by word segmentation;

[0083] Traverse the audience comments after word segmentation and count the number of illegal words matched as N f ;

[0084] By formula: The passive violation index P is calculated; where N span is the screen-sweeping score, m is the number of words in the comment, γ 1 , γ 2 is the weight coefficient, γ 1 +γ 2 =1.

[0085] The process of obtaining the screen-swiping score is as follows:

[0086] Get the number of word segments M of the comment, count the frequency K of comments sent by the same viewer within the set time period, and substitute into the following formula:

[0087]

[0088] Calculate the screen-swiping score N span , where M th is the word count threshold for comments, K th The frequency thresholds for the same audience to send comments within a set period of time, and the above two thresholds are determined based on the analysis of historical data and experience data.

[0089] In this embodiment, the passive content analysis unit can collect and analyze the audience's language expression data in real time during the live broadcast process, and output the passive violation index value in time, so that the live broadcast platform can grasp the violation situation of the audience interaction link in real time. Once a violation is found, it can quickly take corresponding intervention measures, such as blocking illegal comments, warning illegal audiences, etc., to maintain the normal order of the live broadcast; at the same time, by timely discovering and handling the audience's violations, the passive content analysis unit helps to create a healthy and positive live broadcast interactive environment; to avoid malicious comments, advertising screens and inappropriate remarks that have a negative impact on the viewing experience of other viewers, and to improve the audience's satisfaction and loyalty to the live broadcast platform; finally, the quantified passive violation index value can objectively reflect the severity of the violation situation in the audience interaction link. Compared with the traditional subjective judgment method, the quantitative evaluation is more scientific and accurate, and avoids the interference of human factors; and the live broadcast platform can classify the audience's violations according to the indicator, warn and educate the audience with a lighter degree of violation, and take severe measures such as banning the account for the audience with serious violations.

[0090] The process of judging whether the live broadcast content violates the rules based on active and passive violation indicators is as follows:

[0091] By formula: F = (σ 1 *A+σ 2 *P) D The comprehensive violation assessment index F is calculated;

[0092] Among them, σ 1 , σ 2 is the weight coefficient, σ 1 +σ 2 =1, D is the historical impact coefficient;

[0093] The calculated comprehensive violation assessment index F is compared with the preset assessment threshold F T To compare:

[0094] If F>F T , then the live broadcast content is judged to be illegal; if F≤F T , then the live broadcast content is judged not to be in violation of regulations;

[0095] The calculation formula of the historical impact coefficient is:

[0096]

[0097] Among them, M sub-periods are obtained continuously according to the preset interval length, j is the jth sub-period, F j is the average value of the comprehensive violation evaluation index in the jth sub-period, θ jis the impact factor of the live broadcast stage; the expected total live broadcast period is divided into three stages: opening, mid-term and ending, and the impact factor of the opening stage is greater than the impact factor of the mid-term stage and greater than the impact factor of the ending stage. max 、F min are the maximum and minimum values ​​of the comprehensive violation assessment index in the M sub-periods, respectively. μ is the attenuation coefficient, and the initial setting is μ = 0.1. As time goes by, the impact of violations in earlier time periods on the comprehensive violation assessment index will gradually decrease.

[0098] In this embodiment, the comprehensive violation assessment index is used in the live e-commerce content management to not only consider the active violation behaviors of the anchor, such as false propaganda and improper expression, but also the passive violation behaviors of the audience, such as malicious comments and screen swiping, so as to avoid the one-sidedness of single-dimensional assessment;

[0099] At the same time, when calculating the comprehensive violation index evaluation, through F = (σ 1 *A+σ 2 *P) D and The influencing factors are refined and the parameter formula of time attenuation is introduced: e -μ(M-j) To highlight the immediate impact of recent violations, e -μ(M-j) The smaller it is, the smaller the impact of violations in earlier time periods on the comprehensive violation evaluation index will gradually decrease. 1-τ*(F max -F min ) is smaller, it means (F max -F min ) is larger, that is, the extreme value difference of the comprehensive violation assessment index of the current live broadcast room is larger, and there is a large fluctuation range, so strict monitoring is required;

[0100] In addition, the weight of the live broadcast stage is set according to the different stages of the live broadcast (opening, halftime, and ending) to make the evaluation more realistic. At the same time, the comprehensive violation assessment index can realize real-time dynamic monitoring and management, and real-time calculation can be performed during the live broadcast to detect violation risks in a timely manner. Once the score exceeds the preset threshold, an alarm will be issued, which is convenient for management personnel to take measures such as suspending the live broadcast and issuing warnings in a timely manner. Management strategies can also be adjusted dynamically according to real-time scores, such as strengthening comment monitoring in response to increased violations by viewers. In addition, it provides an objective basis for platform management decisions, facilitates the sorting and classification of live broadcast content, helps understand the distribution of violations to formulate long-term strategies, optimizes resource allocation, and reasonably arranges monitoring resources; it can also play a restraining and normative role on anchors and viewers, prompting anchors to standardize language and introduction content, and viewers to reduce bad interactive behaviors, creating a healthy and orderly environment, while also enhancing user trust, ensuring the authenticity, legality and health of live broadcast content, making consumers more assured, and improving user satisfaction and loyalty to the platform.

[0101] See also Figure 2 As shown, a live e-commerce content management method based on big data, the steps are as follows:

[0102] S100, collect live content data on the live e-commerce platform in real time through multiple data collection interfaces;

[0103] S200, performing pre-processing operations of cleaning, deduplication and format conversion on the collected live content data;

[0104] S300, analyzing the pre-processed live content data, wherein the live content data includes active content data and passive content data, analyzing the active content data and the passive content data to obtain active violation indicators and passive violation indicators, and then judging whether the live content violates the regulations based on the active violation indicators and the passive violation indicators;

[0105] S400. Based on the results of content analysis, the live broadcast process is monitored in real time, and when violations are found, corresponding management measures are taken in a timely manner.

[0106] It should be noted that the calculation formulas and various parameters involved in the calculation in the present invention have been dimensionally processed in advance, and the process of dimensionless processing is well known in the industry and will not be described here.

[0107] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A live e-commerce content management system based on big data, characterized in that: include: The live content data collection module is used to collect live content data on the live e-commerce platform in real time through multiple data collection interfaces; Live content data preprocessing, which is used to perform preprocessing operations such as cleaning, deduplication and format conversion on the collected live content data; A live content analysis module analyzes the pre-processed live content data, wherein the live content data includes active content data and passive content data, and obtains active violation indicators and passive violation indicators by analyzing the active content data and the passive content data, and then determines whether the live content violates the regulations based on the active violation indicators and the passive violation indicators; The live broadcast content real-time monitoring module monitors the live broadcast process in real time based on the results of content analysis, and takes corresponding management measures in a timely manner when violations are found; Among them, the active data includes the host's language expression and product introduction content, and the passive data includes the audience's language expression.

2. The live broadcast e-commerce content management system based on big data according to claim 1 is characterized in that: The live content analysis module includes: Active content analysis unit, used to perform natural language processing and semantic analysis on the host's language expression and product introduction content to identify whether there are any illegal words, sensitive information or false propaganda; Passive content analysis unit, used to detect violations of audience language expressions and identify whether there are advertisements or inappropriate remarks; The violation index calculation unit is used to calculate the active violation index and the passive violation index according to the results of the active content analysis unit and the passive content analysis unit, and comprehensively judge whether the live broadcast content violates the regulations.

3. The live broadcast e-commerce content management system based on big data according to claim 2 is characterized in that: The process of calculating the active violation indicator is: S1. Data processing: Convert the host's voice into text, perform word segmentation on the extracted text, build a stop word list, and remove stop words from the word segmentation results; S2. Feature extraction: Build a library of illegal words that include false propaganda words and sensitive information words, traverse the text after word segmentation, and count the number of illegal words that match; Use pre-trained BERT or GPT language models to perform semantic analysis to determine whether the text contains false advertising or exaggerated efficacy; use the TF-IDF algorithm to extract key product information and core expressions in the text as keywords; S3. Calculation of violation indicators: Different weights are assigned to different types of illegal words, and the illegal word score S is calculated based on the matched illegal words and their weights. word ; The expression is: Where n is the number of words in the illegal vocabulary, α i is the weight coefficient of the i-th illegal word, C i is the number of times the illegal word is matched; According to the results of semantic analysis, if it is determined that there is a semantic violation, a fixed semantic violation score S is given. yu ; By formula: A = S word *β1+S yu *β2 is used to calculate the active violation index A; where β1 and β2 are weight coefficients, β1+β2=1.

4. The live broadcast e-commerce content management system based on big data according to claim 3 is characterized in that: The process of calculating the passive violation indicator is: The language expression text data of the audience in the chat area during the live broadcast is collected in real time and processed by word segmentation; Traverse the audience comments after word segmentation and count the number of illegal words matched as N f ; By formula: The passive violation index P is calculated; where N span is the screen-sweeping score, m is the number of words in the comment, γ1 and γ2 are weight coefficients, and γ1+γ2=1.

5. The live broadcast e-commerce content management system based on big data according to claim 4 is characterized in that: The process of obtaining the screen-swiping score is as follows: Get the number of word segments M of the comment, count the frequency K of comments sent by the same viewer within the set time period, and substitute into the following formula: Calculate the screen-swiping score N span , where M th is the word count threshold for comments, K th The frequency threshold for the same viewer to send comments within a set period of time.

6. The live broadcast e-commerce content management system based on big data according to claim 4 is characterized in that: The process of judging whether the live broadcast content violates the rules based on active and passive violation indicators is as follows: By formula: F = (σ1*A+σ2*P) D The comprehensive violation assessment index F is calculated; Among them, σ1 and σ2 are weight coefficients, σ1+σ2=1, and D is the historical impact coefficient; The calculated comprehensive violation assessment index F is compared with the preset assessment threshold F T To compare: If F>F T , then the live broadcast content is judged to be illegal; if F≤F T , then the live broadcast content is judged not to be in violation of regulations.

7. The live broadcast e-commerce content management system based on big data according to claim 6 is characterized in that: The calculation formula of the historical impact coefficient is: Among them, M sub-periods are obtained continuously according to the preset interval length, j is the jth sub-period, F j is the average value of the comprehensive violation evaluation index in the jth sub-period, θ j is the impact factor of the live broadcast stage; the expected total live broadcast period is divided into three stages: opening, mid-term and ending, and the impact factor of the opening stage is greater than the impact factor of the mid-term stage and greater than the impact factor of the ending stage. max 、F min are the maximum and minimum values ​​of the comprehensive violation assessment indicators in the M sub-periods, respectively.

8. The live broadcast e-commerce content management system based on big data according to claim 1 is characterized in that: The system further comprises: The live content data storage module is used to store the pre-processed live content data in the big data storage module to provide an analysis basis for subsequent content management.

9. A live e-commerce content management method based on big data, which is implemented based on the live e-commerce content management system based on big data according to claim 1, characterized in that: Here are the steps: S100, collect live content data on the live e-commerce platform in real time through multiple data collection interfaces; S200, performing pre-processing operations of cleaning, deduplication and format conversion on the collected live content data; S300, analyzing the pre-processed live content data, wherein the live content data includes active content data and passive content data, analyzing the active content data and the passive content data to obtain active violation indicators and passive violation indicators, and then judging whether the live content violates the regulations based on the active violation indicators and the passive violation indicators; S400. Based on the results of content analysis, the live broadcast process is monitored in real time, and when violations are found, corresponding management measures are taken in a timely manner.

Images