A market supervision data management system and method based on large model text

Through large-scale text, analyzing user purchase records and browsing information, calculating impact factors and risk values, providing risk reminders, solving the problem of false information affecting user purchase decisions, reducing purchase risks, and improving information authenticity and purchase success rate.

CN118733774BActive Publication Date: 2024-11-29上海市大数据中心
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202410869647.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-01
Publication Date
2024-11-29
Estimated Expiration
2044-07-01

AI Technical Summary

Technical Problem

There is a large amount of false information and insider information on the market, which makes it difficult for users to effectively evaluate risks when making purchase decisions, and is susceptible to information and lead to loss of benefits.

Method used

A market supervision data management system based on large model text is adopted to analyze users' historical purchase records and browse information, extract keywords, calculate impact factors and risk values, judge purchase risks in real time, and provide risk reminders.

Benefits of technology

Effectively evaluate the risk value of users' purchases, reduce the probability of profit loss due to factors, help users distinguish the authenticity of information, reduce losses, and improve the purchase success rate.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118733774B_ABST
    Figure CN118733774B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of market data technology, and specifically to a market supervision data management system and method based on large model text, the management method comprising the following steps: classifying a plurality of words extracted from product information to obtain a plurality of word sets; selecting a plurality of words as influencing factors of product purchase records, and calculating the comprehensive proportion of user purchases affected by various influencing factors; analyzing the information content browsed by the user before each product purchase record, determining the number of times each keyword appears, and calculating the degree of influence of the keyword on the user's purchase; analyzing the degree of influence of each keyword on the user and the number of times each keyword appears, and obtaining the risk value of a product purchase record of the user; obtaining the minimum risk value that can be purchased according to the income of each product purchase record in the history of the user; calculating the risk value of the user's current purchase in real time, and judging whether to issue a risk reminder for the user's current purchase.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of market data technology, and in particular to a market supervision data management system and method based on large model text. Background Art

[0002] Large model text usually refers to large-scale text generation models built based on artificial intelligence and machine learning technologies, such as the GPT model series; these large model texts can process and generate large amounts of natural language text data and have strong language understanding and generation capabilities; they are trained through large-scale training data sets and deep learning technologies to learn language patterns, semantic understanding, and text generation capabilities. They can automatically generate coherent and logical text content, answer questions, generate articles, conduct conversations, and other natural language processing tasks;

[0003] There is a large amount of false information and various purchasing insider information in the market. Users are often influenced by this information when making purchasing decisions. Users are unable to conduct effective risk assessments on their purchasing behaviors and are easily influenced by various information in the market, resulting in losses to their own interests. Summary of the invention

[0004] The purpose of the present invention is to provide a market supervision data management system and method based on large model text to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a market supervision data management method based on a large model text, the management method comprising the following steps:

[0006] Step S100: Obtain product information of each product purchase record of the user in history, classify several words extracted from the product information to obtain several categories of word sets; analyze the words in the several categories of word sets, select several words as influencing factors of the user's product purchase record, obtain the proportion of abnormal purchases of the user affected by the influencing factors based on the total number of purchases containing the influencing factors in the user's product purchase record, and calculate the comprehensive proportion of the user's purchase affected by each influencing factor;

[0007] Step S200: Analyze the information content browsed by the user before each product purchase record to obtain a number of keywords; replace some keywords by comparing the similarity between each keyword and other keywords, and determine the number of times each keyword appears; analyze the influence of each keyword on the user in the historical purchase, and calculate the influence of the keyword on the user's purchase;

[0008] Step S300: Analyze the influence of each keyword on the user and the number of times each keyword appears in the information content of each product purchase record browsed by the user in history, and obtain the risk value of the user's product purchase record; according to the profit of each product purchase record in history of the user, obtain the minimum risk value that can be purchased;

[0009] Step S400: Acquire the information content currently browsed by the user in real time, calculate the risk value of the user's current purchase based on the extraction of each keyword in the information content, and determine whether to issue a risk reminder for the user's current purchase.

[0010] Furthermore, step S100 includes the following steps:

[0011] Step S101: Set the product information of the user's i-th product purchase record as M i , generate product information M i The text content in the text is segmented to obtain a number of words; any two words are selected, the similarity between the two words is compared, and a similarity threshold is set. If the similarity between the two words exceeds the similarity threshold, the two words are classified into one category to obtain a number of categories of word sets;

[0012] Step S102: Count the number of words in each word set, and sort the word sets from large to small according to the number of words; select the first c word sets as the influencing word sets of the i-th category purchase; arbitrarily select a word in the j-th influencing word set as the reference word of the j-th influencing word set, obtain the similarity between each remaining word in the j-th influencing word set and the reference word, and calculate the average value to obtain the average similarity of the reference words; obtain the average similarity when each word in the j-th influencing word set is used as the reference word, select a word with the largest average similarity as the target word of the j-th influencing word set, and use the target word as an influencing factor of the user's i-th product purchase record; by determining the influencing factor, several factors of the user's purchase can be determined, which is conducive to the subsequent analysis of the user to analyze the influence of each factor on the user's purchase;

[0013] Step S103: Get the number of after-sales service times of the i-th product purchase record as G i , if G i >0, the i-th product purchase record is marked as an abnormal purchase; the d-th influencing factor of the user's i-th product purchase record is set to Get historical existence impact factor Every product purchase record, count the number of purchases with abnormal purchase marks Set the user history influence factor The total number of purchases is Calculate the user's impact factor The proportion of abnormal purchases affected is

[0014] Step S104: Setting the impact factor The abnormal purchases accounted for the largest proportion, and the abnormal purchases accounted for According to the formula:

[0015]

[0016] Among them, e is the e-th influencing factor in the i-th product purchase record, g is the number of influencing factors in the i-th product purchase record, is the proportion of abnormal purchases of users affected by the e-th influencing factor, α is a constant and The comprehensive proportion of the factors affecting the user's i-th product purchase record is calculated to be α i ; By comparing the relationship between each influencing factor and the set constant, the influence degree of the influencing factor can be judged. If the influence degree of the influencing factor is low, it can be said that the influence of this influencing factor on user purchases is small, which can reduce the overall influence degree; and select the influencing factor with the largest proportion of abnormal purchases as the basic value, because this influencing factor must be the largest factor affecting user purchases, and the final comprehensive proportion can be more accurate.

[0017] Further, step S200 includes the following steps:

[0018] Step S201: Obtain the information content browsed by the user during the i-th product purchase record, and set the number of browsed information contents as U i ;

[0019] Select the kth information content as Extract information content Keywords, generate keyword sets;

[0020] Step S202: Obtain the information content that the user browsed most recently before the i-th product purchase record. Extract information content Keyword collection Set the selected keyword set The pth keyword in is Compare Keywords The similarity between the keywords in the rest of the information content. If the similarity between two keywords exceeds the set similarity threshold, each keyword will be replaced with the keyword Get the keywords in the information content of the user's i-th product purchase record browsing The number of occurrences is

[0021] Step S203: Obtain the influence factor of the user's i-th product purchase record and calculate the keyword The similarity between each influencing factor is set as Keywords and impact factor The similarity is S pd , according to the formula:

[0022]

[0023] in, Factors that affect users The abnormal purchase ratio affected by the IF() function. If G i >0, then IF(G i <0)=1, otherwise IF(G i <0)=0, Influencing factors exist in the vocabulary set The number of purchases; calculate the keywords Among the information browsed by users, the influence on users' purchase is By analyzing the number of occurrences of each keyword and the number of times each keyword caused a successful purchase in historical purchases, the selected keywords are obtained from the most recent browsing history before the purchase. Because the user's purchase is very likely to be influenced by the keywords in the most recent browsing history, therefore, the above method is used to calculate the degree of influence of each keyword on the user's purchase, which can more effectively calculate the subsequent risk value.

[0024] Further, step S300 includes the following steps:

[0025] Step S301: Set the pth keyword in the information content browsed in the i-th product purchase record of the user The impact on user purchases is According to the formula:

[0026]

[0027] Among them, A is the baseline risk value, w is the number of keywords in the browsing information content, The pth keyword The number of times it appears in the message content; The information that the user has browsed in the past contains keywords Number of purchases; The information that the user has browsed in the past contains keywords The number of purchases with after-sales service times greater than zero in each purchase record; the risk value of the user's i-th product purchase record in history is calculated as Risk i ;

[0028] Step S302: Obtain all the product purchase records of the user with the number of after-sales service times greater than zero, and select the one with the least impact factor as the comprehensive proportion α min , according to the formula:

[0029] Risk min =A×(1+α min )

[0030] The minimum risk value that the user can purchase is calculated as Risk min .

[0031] Furthermore, step S400 includes the following steps:

[0032] Step S401: Obtain all the information content currently browsed by the user, extract several keywords from the information content; select the pth keyword Key p , the keyword Key p Compare the similarity with each word in the information content, and replace the words whose similarity exceeds the similarity threshold with the keyword Key p , and count the keywords in the information content p The number of occurrences is Q p ;

[0033] Step S402: Obtain the keyword Key p The number of times a keyword is used as a keyword in a user's purchase history is num(Key p ), and in num(Key p ) product purchase records, the number of purchases with a revenue less than zero is num(Key p )'; According to the formula:

[0034]

[0035] Among them, A is the benchmark risk value, Y p Key p The degree of influence on user purchase; after calculating the current browsing information content, the risk value of the user making a purchase is Risk;

[0036] Step S403: Set the minimum risk value that can be purchased as Risk min , if Risk>Risk min , then the user will be reminded of the risks of the current purchase.

[0037] In order to better implement the above method, a market supervision data management system based on large model text is also proposed. The management system includes a historical purchase behavior analysis module, an information risk assessment module, a purchase risk assessment module and a real-time purchase analysis module;

[0038] The historical purchase behavior analysis module is used to obtain product information of each product purchase record of the user in history, classify several words extracted from the product information to obtain several categories of word sets; by analyzing the words in the several categories of word sets, several words are selected as the influencing factors of the user's product purchase record, based on the total number of purchases containing the influencing factors in the user's product purchase record, the proportion of abnormal purchases of the user affected by the influencing factors is obtained, and the comprehensive proportion of the user's purchase affected by each influencing factor is calculated;

[0039] The information risk assessment module is used to analyze the information content browsed by the user before each product purchase record to obtain a number of keywords; replace some keywords by comparing the similarity between each keyword and the remaining keywords, and determine the number of times each keyword appears; analyze the impact of each keyword on the user in the historical purchase, and calculate the impact of the keyword on the user's purchase;

[0040] The purchase risk assessment module is used to analyze the influence of each keyword on the user and the number of times each keyword appears in the information content of each product purchase record browsed by the user in history, and obtain the risk value of the user's product purchase record; according to the income of each product purchase record in history of the user, the minimum risk value that can be purchased is obtained;

[0041] The real-time purchase analysis module is used to obtain the information content currently browsed by the user in real time, calculate the risk value of the user's current purchase based on the extraction of each keyword in the information content, and determine whether to issue a risk reminder for the user's current purchase.

[0042] Furthermore, the historical purchase behavior analysis module includes a vocabulary set acquisition unit and an influence ratio calculation unit;

[0043] The vocabulary set acquisition unit is used to obtain product information of each product purchase record of the user in history, and classify several words extracted from the product information to obtain several categories of vocabulary sets; the influence ratio calculation unit is used to analyze the words in the several categories of vocabulary sets, select several words as influence factors of the user's product purchase record, obtain the proportion of abnormal purchases of the user affected by the influence factors based on the total number of purchases containing the influence factors in the user's product purchase record, and calculate the comprehensive proportion of the user's purchase affected by each influence factor.

[0044] Further, the information risk assessment module includes a keyword analysis unit and a keyword impact analysis unit;

[0045] The keyword analysis unit is used to analyze the information content browsed by the user before each product purchase record to obtain a number of keywords; by comparing the similarity between each keyword and the remaining keywords, some keywords are replaced and the number of times each keyword appears is determined; the keyword impact analysis unit is used to analyze the impact of each keyword on the user in the historical purchase and calculate the impact of the keyword on the user's purchase.

[0046] Further, the purchase risk assessment module includes a risk value calculation unit and a minimum risk calculation unit;

[0047] The risk value calculation unit is used to analyze the influence of each keyword on the user and the number of times each keyword appears in the information content of each product purchase record browsed by the user in history, and obtain the risk value of the user's product purchase record; the minimum risk calculation unit is used to obtain the minimum risk value that can be purchased based on the profit of each product purchase record in history of the user.

[0048] Furthermore, the real-time purchase analysis module includes a real-time purchase evaluation unit and a risk warning judgment unit;

[0049] The real-time purchase evaluation unit is used to obtain the information content currently browsed by the user in real time, and calculate the risk value of the user's current purchase based on the extraction of each keyword in the information content; the risk reminder judgment unit is used to judge whether to give a risk reminder to the user's current purchase based on the risk value of the user's current purchase.

[0050] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention analyzes the product information at the time of purchase by the user, captures the factors that affect the user's purchase according to the purchase content, and obtains the reasons for the user's purchase, which can help to effectively evaluate the risk value of the user's purchase in the future, thereby reducing the probability of the user's loss of profit due to the influence of some factors; (2) The present invention analyzes the information content browsed by the user before the purchase, analyzes the factors that prompt the user to make a purchase, and judges whether there is a risk in the user's purchase according to the influence of each factor in the historical data, which can help the user distinguish the authenticity and usefulness of the information, promote the user's purchase success, and reduce the user's loss; (3) The present invention directly helps the user judge the risk level of the purchase by calculating the risk value of the user's product purchase record, and reminds the user, which can effectively help the user to carry out the purchase. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0052] Figure 1 It is a schematic diagram of the steps of a market supervision data management method based on a large model text;

[0053] Figure 2 It is a structural diagram of a market supervision data management system based on large model text. DETAILED DESCRIPTION

[0054] 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.

[0055] See also Figure 1 to Figure 2 The present invention provides a technical solution: a market supervision data management method based on a large model text, the management method comprising the following steps:

[0056] Step S100: Obtain product information of each product purchase record of the user in history, classify several words extracted from the product information to obtain several categories of word sets; analyze the words in the several categories of word sets, select several words as influencing factors of the user's product purchase record, obtain the proportion of abnormal purchases of the user affected by the influencing factors based on the total number of purchases containing the influencing factors in the user's product purchase record, and calculate the comprehensive proportion of the user's purchase affected by each influencing factor;

[0057] Wherein, step S100 includes the following steps:

[0058] Step S101: Set the product information of the user's i-th product purchase record as M i , generate product information M i The text content in the text is segmented to obtain a number of words; any two words are selected, the similarity between the two words is compared, and a similarity threshold is set. If the similarity between the two words exceeds the similarity threshold, the two words are classified into one category to obtain a number of categories of word sets;

[0059] Step S102: Count the number of words in each word set, and sort the word sets from large to small according to the number of words; select the first c word sets as the influencing word sets of the i-th category purchase; arbitrarily select a word in the j-th influencing word set as the reference word of the j-th influencing word set, obtain the similarity between each remaining word in the j-th influencing word set and the reference word, and calculate the average value to obtain the average similarity of the reference words; obtain the average similarity of each word in the j-th influencing word set when it is used as a reference word, select a word with the largest average similarity as the target word of the j-th influencing word set, and use the target word as an influencing factor of the user's i-th product purchase record;

[0060] Step S103: Get the number of after-sales service times of the i-th product purchase record as G i , if G i >0, the i-th product purchase record is marked as an abnormal purchase; the d-th influencing factor of the user's i-th product purchase record is set to Get historical existence impact factor Every product purchase record, count the number of purchases with abnormal purchase marks Set the user history influence factor The total number of purchases is Calculate the user's impact factor The proportion of abnormal purchases affected is

[0061] Step S104: Setting the impact factor The abnormal purchases accounted for the largest proportion, and the abnormal purchases accounted for According to the formula:

[0062]

[0063] Among them, e is the e-th influencing factor in the i-th product purchase record, g is the number of influencing factors in the i-th product purchase record, is the proportion of abnormal purchases of users affected by the e-th influencing factor, α is a constant and The comprehensive proportion of the factors affecting the user's i-th product purchase record is calculated to be α i .

[0064] Step S200: Analyze the information content browsed by the user before each product purchase record to obtain a number of keywords; replace some keywords by comparing the similarity between each keyword and other keywords, and determine the number of times each keyword appears; analyze the influence of each keyword on the user in the historical purchase, and calculate the influence of the keyword on the user's purchase;

[0065] Wherein, step S200 includes the following steps:

[0066] Step S201: Obtain the information content browsed by the user during the i-th product purchase record, and set the number of browsed information contents as U i ;

[0067] Select the kth information content as Extract information content Keywords, generate keyword sets;

[0068] Step S202: Obtain the information content that the user browsed most recently before the i-th product purchase record. Extract information content Keyword collection Set the selected keyword set The pth keyword in is Compare Keywords The similarity between the keywords in the rest of the information content. If the similarity between two keywords exceeds the set similarity threshold, each keyword will be replaced with the keyword Get the keywords in the information content of the user's i-th product purchase record browsing The number of occurrences is

[0069] Step S203: Obtain the influence factor of the user's i-th product purchase record and calculate the keyword The similarity between each influencing factor is set as Keywords and impact factor The similarity is S pd , according to the formula:

[0070]

[0071] in, Factors that affect users The abnormal purchase ratio affected by the IF() function. If G i >0, then IF(G i <0)=1, otherwise IF(G i <0)=0, Influencing factors exist in the vocabulary set The number of purchases; calculate the keywords Among the information browsed by users, the influence on users' purchase is

[0072] Step S300: Analyze the influence of each keyword on the user and the number of times each keyword appears in the information content of each product purchase record browsed by the user in history, and obtain the risk value of the user's product purchase record; according to the profit of each product purchase record in history of the user, obtain the minimum risk value that can be purchased;

[0073] Wherein, step S300 includes the following steps:

[0074] Step S301: Set the pth keyword in the information content browsed in the i-th product purchase record of the user The impact on user purchases is According to the formula:

[0075]

[0076] Among them, A is the baseline risk value, w is the number of keywords in the browsing information content, The pth keyword The number of times it appears in the message content; The information that the user has browsed in the past contains keywords Number of purchases; The information that the user has browsed in the past contains keywords The number of purchases with after-sales service times greater than zero in each purchase record; the risk value of the user's i-th product purchase record in history is calculated as Risk i ;

[0077] Step S302: Obtain all the product purchase records of the user with the number of after-sales service times greater than zero, and select the one with the least impact factor as the comprehensive proportion α min , according to the formula:

[0078] Risk min =A×(1+α min )

[0079] The minimum risk value that the user can purchase is calculated as Risk min .

[0080] Step S400: obtaining the information content currently browsed by the user in real time, calculating the risk value of the user's current purchase according to the extraction of each keyword in the information content, and determining whether to issue a risk reminder for the user's current purchase;

[0081] Wherein, step S400 includes the following steps:

[0082] Step S401: Obtain all the information content currently browsed by the user, extract several keywords from the information content; select the pth keyword Keyp , the keyword Key p Compare the similarity with each word in the information content, and replace the words whose similarity exceeds the similarity threshold with the keyword Key p , and count the keywords in the information content p The number of occurrences is Q p ;

[0083] Step S402: Obtain the keyword Key p The number of times a keyword is used as a keyword in a user's purchase history is num(Key p ), and in num(Key p ) product purchase records, the number of purchases with a revenue less than zero is num(Key p )'; According to the formula:

[0084]

[0085] Among them, A is the benchmark risk value, Y p Key p The degree of influence on user purchase; after calculating the current browsing information content, the risk value of the user making a purchase is Risk;

[0086] Step S403: Set the minimum risk value that can be purchased as Risk min , if Risk>Risk min , then the user will be reminded of the risks of the current purchase.

[0087] A market supervision data management system based on large model text, the management system includes a historical purchase behavior analysis module, an information risk assessment module, a purchase risk assessment module and a real-time purchase analysis module;

[0088] The historical purchase behavior analysis module is used to obtain product information of each product purchase record of the user in history, classify several words extracted from the product information to obtain several categories of word sets; by analyzing the words in the several categories of word sets, several words are selected as the influencing factors of the user's product purchase record, based on the total number of purchases containing the influencing factors in the user's product purchase record, the proportion of abnormal purchases of the user affected by the influencing factors is obtained, and the comprehensive proportion of the user's purchase affected by each influencing factor is calculated;

[0089] The information risk assessment module is used to analyze the information content browsed by the user before each product purchase record to obtain a number of keywords; replace some keywords by comparing the similarity between each keyword and the remaining keywords, and determine the number of times each keyword appears; analyze the impact of each keyword on the user in the historical purchase, and calculate the impact of the keyword on the user's purchase;

[0090] The purchase risk assessment module is used to analyze the influence of each keyword on the user and the number of times each keyword appears in the information content of each product purchase record browsed by the user in history, and obtain the risk value of the user's product purchase record; according to the income of each product purchase record in history of the user, the minimum risk value that can be purchased is obtained;

[0091] The real-time purchase analysis module is used to obtain the information content currently browsed by the user in real time, calculate the risk value of the user's current purchase based on the extraction of each keyword in the information content, and determine whether to issue a risk reminder for the user's current purchase.

[0092] Among them, the historical purchase behavior analysis module includes a vocabulary set acquisition unit and an influence ratio calculation unit;

[0093] The vocabulary set acquisition unit is used to obtain product information of each product purchase record of the user in history, and classify several words extracted from the product information to obtain several categories of vocabulary sets; the influence ratio calculation unit is used to analyze the words in the several categories of vocabulary sets, select several words as influence factors of the user's product purchase record, obtain the proportion of abnormal purchases of the user affected by the influence factors based on the total number of purchases containing the influence factors in the user's product purchase record, and calculate the comprehensive proportion of the user's purchase affected by each influence factor.

[0094] Among them, the information risk assessment module includes a keyword analysis unit and a keyword impact analysis unit;

[0095] The keyword analysis unit is used to analyze the information content browsed by the user before each product purchase record to obtain a number of keywords; by comparing the similarity between each keyword and the remaining keywords, some keywords are replaced and the number of times each keyword appears is determined; the keyword impact analysis unit is used to analyze the impact of each keyword on the user in the historical purchase and calculate the impact of the keyword on the user's purchase.

[0096] Among them, the purchase risk assessment module includes a risk value calculation unit and a minimum risk calculation unit;

[0097] The risk value calculation unit is used to analyze the influence of each keyword on the user and the number of times each keyword appears in the information content of each product purchase record browsed by the user in history, and obtain the risk value of the user's product purchase record; the minimum risk calculation unit is used to obtain the minimum risk value that can be purchased based on the profit of each product purchase record in history of the user.

[0098] Among them, the real-time purchase analysis module includes a real-time purchase evaluation unit and a risk reminder judgment unit;

[0099] The real-time purchase evaluation unit is used to obtain the information content currently browsed by the user in real time, and calculate the risk value of the user's current purchase based on the extraction of each keyword in the information content; the risk reminder judgment unit is used to judge whether to give a risk reminder to the user's current purchase based on the risk value of the user's current purchase.

[0100] The usage scenario is that users purchase some goods by browsing market information, and then sell the purchased goods to obtain profits;

[0101] The product information of a user's historical product purchase record is obtained as follows: "100K screen clarity, permanent battery life; price: only 50, original price 500", and the content extracted from the text information is "clarity, 100K, battery life, permanent, only for sale", and "clarity", "battery life" and "permanent" are selected as influencing factors; the user's profit after purchase is -100 yuan, so this product purchase record is marked as an abnormal purchase; in each product purchase record with the influencing factor "permanent" in the history, the number of purchases with abnormal purchase marks is counted as 4 times, and the total number of purchases with the influencing factor "permanent" in the user's history is 5 times, and it is calculated that the abnormal purchase ratio of the user affected by the influencing factor "permanent" is 80%;

[0102] Set the largest influencing factor of abnormal purchases to "permanent", according to the formula:

[0103]

[0104] =80%×(1-20%)=64%, the abnormal purchase ratio of the influencing factor "endurance" is 20%; the comprehensive ratio of the user's product purchase record affected by the influencing factors is 64%;

[0105] Get the information content that the user browsed most recently before making a purchase, extract the keywords "battery life" and "permanent", get the number of times "battery life" appears in all the browsed information content before the purchase is 3 times, and the number of times "permanent" appears in all the browsed information content before the purchase is 3 times, and the similarity between the two keywords and the influencing factors of the product information is 100%, according to the formula:

[0106]

[0107] Among them, the abnormal purchases affected by the factor "endurance" accounted for 20%, and the abnormal purchases affected by the factor "permanent" accounted for 80%, and the historical revenues of these two keywords were both less than zero. Therefore, it was calculated that the keyword "permanent" had an influence of 80% on the user's purchase in the information browsed by the user, and the keyword "endurance" had an influence of 20% on the user's purchase in the information browsed by the user.

[0108] Set the risk baseline value to 100, according to the formula:

[0109]

[0110] The risk value of the user's purchase is 100×(2.4+0.6)=300; the comprehensive proportion of the factors that have the least impact on the user is set to 50%, and the minimum risk value that the user can purchase is obtained as 100×1.5=150. Because 300 is greater than 150, there is a high risk in the user's purchase, and a risk reminder is issued.

[0111] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0112] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A market supervision data management method based on large model text, characterized by: The management method comprises the following steps: Step S100: Obtain product information of each product purchase record of the user in history, classify several words extracted from the product information to obtain several categories of word sets; analyze the words in the several categories of word sets, select several words as influencing factors of the user's product purchase record, obtain the proportion of abnormal purchases of the user affected by the influencing factors based on the total number of purchases containing the influencing factors in the user's product purchase record, and calculate the comprehensive proportion of the user's purchase affected by each influencing factor; Step S200: Analyze the information content browsed by the user in each product purchase record to obtain a number of keywords; replace some keywords by comparing the similarity between each keyword and other keywords, and determine the number of times each keyword appears; analyze the influence of each keyword on the user in the historical purchase, and calculate the influence of the keyword on the user's purchase; Step S300: Analyze the influence of each keyword on the user and the number of times each keyword appears in the information content browsed by the user in each product purchase record in history, and obtain the risk value of the user's product purchase record; according to the after-sales situation of each product purchase record in history of the user, obtain the minimum risk value that can be purchased; Step S400: Acquire the information content currently browsed by the user in real time, calculate the risk value of the user's current purchase based on the extraction of each keyword in the information content, and determine whether to issue a risk reminder for the user's current purchase.

2. According to claim 1, a market supervision data management method based on large model text is characterized by: The step S100 includes the following steps: Step S101: Set the product information of the user's i-th product purchase record as M i , generate product information M i The text content in the text is segmented to obtain a number of words; any two words are selected, the similarity between the two words is compared, and a similarity threshold is set. If the similarity between the two words exceeds the similarity threshold, the two words are classified into one category to obtain a number of categories of word sets; Step S102: Count the number of words in each word set, and sort the word sets from large to small according to the number of words; select the first c word sets as the influencing word sets of the i-th category purchase; arbitrarily select a word in the j-th influencing word set as the reference word of the j-th influencing word set, obtain the similarity between each remaining word in the j-th influencing word set and the reference word, and calculate the average value to obtain the average similarity of the reference words; obtain the average similarity of each word in the j-th influencing word set when it is used as a reference word, select a word with the largest average similarity as the target word of the j-th influencing word set, and use the target word as an influencing factor of the user's i-th product purchase record; Step S103: Get the number of after-sales service times of the i-th product purchase record as G i , if G i >0, the i-th product purchase record is marked as an abnormal purchase; the d-th influencing factor of the user's i-th product purchase record is set to Get historical existence impact factor Every product purchase record, count the number of purchases with abnormal purchase marks Set the user history influence factor The total number of purchases is Calculate the user's impact factor The proportion of abnormal purchases affected is Step S104: Setting the impact factor The abnormal purchases accounted for the largest proportion, and the abnormal purchases accounted for According to the formula: Among them, e is the e-th influencing factor in the i-th product purchase record, g is the number of influencing factors in the i-th product purchase record, is the proportion of abnormal purchases of users affected by the e-th influencing factor, α is a constant and The comprehensive proportion of the factors affecting the user's i-th product purchase record is calculated to be α i .

3. The market supervision data management method based on large model text according to claim 2 is characterized by: The step S200 includes the following steps: Step S201: Obtain the information content browsed by the user during the i-th product purchase record, and set the number of browsed information contents as U i ; Select the kth information content as Extract information content Keywords, generate keyword sets; Step S202: Obtain the information content that the user browsed most recently before the i-th product purchase record. Extract information content Keyword collection Set the selected keyword set The pth keyword in is Compare Keywords The similarity between the keywords in the rest of the information content. If the similarity between two keywords exceeds the set similarity threshold, each keyword will be replaced with the keyword Get the keywords in the information content of the user's i-th product purchase record browsing The number of occurrences is Step S203: Obtain the influence factor of the user's i-th product purchase record and calculate the keyword The similarity between each influencing factor is set as Keywords and impact factor The similarity is S pd , according to the formula: in, Factors that affect users The abnormal purchase ratio affected by the IF() function. If G i >0, then IF(G i <0)=1, otherwise IF(G i <0)=0, Influencing factors exist in the vocabulary set The number of purchases; calculate the keywords Among the information browsed by users, the influence on users' purchase is 4. The market supervision data management method based on large model text according to claim 3 is characterized by: The step S300 includes the following steps: Step S301: Set the pth keyword in the information content browsed in the i-th product purchase record of the user The impact on user purchases is According to the formula: Among them, A is the baseline risk value, w is the number of keywords in the browsing information content, The pth keyword The number of times it appears in the message content; The information that the user has browsed in the past contains keywords Number of purchases; The information that the user has browsed in the past contains keywords The number of purchases with after-sales service times greater than zero in each purchase record; the risk value of the user's i-th product purchase record in history is calculated as Risk i ; Step S302: Obtain all the product purchase records of the user with the number of after-sales service times greater than zero, and select the one with the least impact factor as the comprehensive proportion α min , according to the formula: Risk min =A×(1+α min ) The minimum risk value that the user can purchase is calculated as Risk min .

5. The market supervision data management method based on large model text according to claim 4 is characterized by: The step S400 includes the following steps: Step S401: Obtain all the information content currently browsed by the user, extract several keywords from the information content; select the pth keyword Key p , the keyword Key p Compare the similarity with each word in the information content, and replace the words whose similarity exceeds the similarity threshold with the keyword Key p , and count the keywords in the information content p The number of occurrences is Q p ; Step S402: Obtain the keyword Key p The number of times a keyword is used as a keyword in a user's purchase history is num(Key p ), and in num(Key p ) product purchase records, the number of purchases with after-sales times greater than zero is num(Key p )'; According to the formula: Among them, A is the benchmark risk value, Y p Key p The degree of influence on user purchase; after calculating the current browsing information content, the risk value of the user making a purchase is Risk; Step S403: Set the minimum risk value that can be purchased as Risk min , if Risk>Risk min , then the user will be reminded of the risks of the current purchase.

6. A market supervision data management system applied to a market supervision data management method based on a large model text according to any one of claims 1 to 5, characterized in that: The management system includes a historical purchase behavior analysis module, an information risk assessment module, a purchase risk assessment module and a real-time purchase analysis module; The historical purchase behavior analysis module is used to obtain product information of each product purchase record of the user in history, classify several words extracted from the product information to obtain several categories of word sets; by analyzing the words in the several categories of word sets, several words are selected as the influencing factors of the user's product purchase record, based on the total number of purchases containing the influencing factors in the user's product purchase record, the proportion of abnormal purchases of the user affected by the influencing factors is obtained, and the comprehensive proportion of the user's purchase affected by each influencing factor is calculated; The information risk assessment module is used to analyze the information content browsed by the user before each product purchase record to obtain a number of keywords; replace some keywords by comparing the similarity between each keyword and the remaining keywords, and determine the number of times each keyword appears; analyze the impact of each keyword on the user in the historical purchase, and calculate the impact of the keyword on the user's purchase; The purchase risk assessment module is used to analyze the influence of each keyword on the user and the number of times each keyword appears in the information content of each product purchase record browsed by the user in history, and obtain the risk value of the user's product purchase record; according to the after-sales situation of each product purchase record in history of the user, the minimum risk value that can be purchased is obtained; The real-time purchase analysis module is used to obtain the information content currently browsed by the user in real time, calculate the risk value of the user's current purchase based on the extraction of each keyword in the information content, and determine whether to issue a risk reminder for the user's current purchase.

7. The market supervision data management system according to claim 6, characterized in that: The historical purchase behavior analysis module includes a vocabulary set acquisition unit and an influence ratio calculation unit; The vocabulary set acquisition unit is used to obtain product information of each product purchase record of the user in history, and classify several words extracted from the product information to obtain several categories of vocabulary sets; the influence ratio calculation unit is used to analyze the words in the several categories of vocabulary sets, select several words as influence factors of the user's product purchase record, obtain the proportion of abnormal purchases of the user affected by the influence factors based on the total number of purchases containing the influence factors in the user's product purchase record, and calculate the comprehensive proportion of the user's purchase affected by each influence factor.

8. The market supervision data management system according to claim 6, characterized in that: The information risk assessment module includes a keyword analysis unit and a keyword impact analysis unit; The keyword analysis unit is used to analyze the information content browsed by the user before each product purchase record to obtain a number of keywords; by comparing the similarity between each keyword and the remaining keywords, some keywords are replaced and the number of times each keyword appears is determined; the keyword impact analysis unit is used to analyze the impact of each keyword on the user in the historical purchase and calculate the impact of the keyword on the user's purchase.

9. The market supervision data management system according to claim 6, characterized in that: The purchase risk assessment module includes a risk value calculation unit and a minimum risk calculation unit; The risk value calculation unit is used to analyze the influence of each keyword on the user and the number of times each keyword appears in the information content of each product purchase record browsed by the user in history, and obtain the risk value of the user's product purchase record; the minimum risk calculation unit is used to obtain the minimum purchase risk value based on the after-sales situation of each product purchase record in history of the user.

10. The market supervision data management system according to claim 6, characterized in that: The real-time purchase analysis module includes a real-time purchase evaluation unit and a risk reminder judgment unit; The real-time purchase evaluation unit is used to obtain the information content currently browsed by the user in real time, and calculate the risk value of the user's current purchase based on the extraction of each keyword in the information content; the risk reminder judgment unit is used to judge whether to give a risk reminder to the user's current purchase based on the risk value of the user's current purchase.

Citation Information

Patent Citations

  • E-commerce order interception method and device and electronic equipment

    CN112396504A

  • Transaction information reading method and system based on online shopping

    CN116402577A

  • Techniques for generating a psychographic profile

    US20170270544A1

  • Personal information assistant computing system

    US20180082683A1