Customer identification methods, devices, and electronic equipment based on multi-dimensional data tagging

By using a multi-dimensional data tagging method, internal and external data are collected and analyzed, abnormal texts are filtered and tagged, and abnormal customer groups are identified. This solves the problems of time-consuming, labor-intensive, and inaccurate identification in existing technologies, and realizes full-cycle abnormal customer group identification and security assurance.

CN115760149BActive Publication Date: 2026-11-17SHANGHAI QIYU INFORMATION TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202211263248.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-10-14
Publication Date
2026-11-17
Estimated Expiration
2042-10-14

AI Technical Summary

Technical Problem

In existing technologies, methods for identifying abnormal customer groups are time-consuming, labor-intensive, and cannot identify them in a timely and accurate manner, thus threatening data security and property security.

Method used

By using a multi-dimensional data labeling method, internal and external data are collected separately. Text analysis models are used to filter target-related texts. Automated sentiment analysis and preset rules are combined to extract effective information, segment and analyze customer groups, and identify abnormal behaviors.

Benefits of technology

It enables full-cycle, timely, and comprehensive identification of abnormal customer groups, improving the timeliness and accuracy of identification and ensuring the platform's data and property security.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115760149B_ABST
    Figure CN115760149B_ABST
Patent Text Reader

Abstract

This invention discloses a customer group identification method, device, and electronic device based on multi-source data tagging. The method includes: collecting multi-source data from internal and external data to obtain initial information; filtering target-related text from the initial information using a text analysis model; extracting effective information from the target-related text using automated text sentiment analysis and preset rules, and tagging the effective information to obtain tag data; locating the target corresponding to each tag data, and dividing each target into multiple first target customer groups; and analyzing preset indicators and their changing trends for each first target customer group to determine second target customer groups. This invention enables the effective utilization of internal and external information related to business scenarios and the full-link monitoring of abnormal customer groups within business scenarios, improving the timeliness and accuracy of customer group identification, thereby providing timely and comprehensive protection for platform data and property security.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of text processing technology, and more specifically, to a method, apparatus, system, and computer-readable medium for customer group identification based on multi-source data tagging. Background Technology

[0002] With the development of the internet, various internet service platforms have emerged. These platforms can provide products or services that require payment after use (for example, rental platforms can provide products for rent first, and payment is made upon return; shopping platforms can provide products first, and payment is made after satisfaction, etc.), bringing great convenience to people's lives. However, at the same time, many abnormal customer groups have emerged who exploit the vulnerabilities of various platforms to carry out illegal activities such as network attacks, cheating, and fraud, threatening the data security and property security of the platforms.

[0003] Currently, abnormal customer groups are mainly identified through manual review before product delivery or after payment settlement. However, manual review before delivery is time-consuming and labor-intensive, and its accuracy cannot be guaranteed due to data limitations. Discovering abnormal customers after payment settlement means missing the opportunity for timely control and loss recovery. Therefore, there is an urgent need for a time-saving, labor-saving, highly efficient method capable of identifying abnormal customers throughout the entire product lifecycle. Summary of the Invention

[0004] In view of this, the main objective of the present invention is to propose a customer group identification method, apparatus, electronic device and computer-readable medium based on multi-source data tagging, in order to at least partially solve at least one of the above-mentioned technical problems.

[0005] To address the aforementioned technical problems, the first aspect of this invention proposes a customer group identification method based on multi-dimensional data labeling, the method comprising:

[0006] We collect data from both internal and external sources to obtain initial information relevant to the business scenario.

[0007] The target related text in the initial information is filtered out using a text analysis model;

[0008] By combining automated text sentiment analysis and preset rules, effective information is extracted from the target associated text, and the effective information is tagged to obtain tag data;

[0009] Locate the target corresponding to each tag data, and divide each target into multiple primary target customer groups;

[0010] Analyze the preset indicators and the changing trends of the preset indicators for each primary target customer group to determine the secondary target customer group.

[0011] According to a preferred embodiment of the present invention, the step of analyzing preset indicators and the changing trends of preset indicators for each first target customer group to determine the second target customer group includes:

[0012] Quantitative analysis of preset indicators for each primary target customer group determines the discriminative power of the corresponding tag data;

[0013] Track and analyze the changing trends of preset indicators for each primary target customer group to determine the current security performance trend and security performance level of the corresponding primary target customer group;

[0014] The second target customer group is selected from each of the first target customer groups by combining the preset indicators, discrimination, security performance and security performance trend.

[0015] According to a preferred embodiment of the present invention, the method further includes:

[0016] Obtain target data for each target within each of the second target customer groups;

[0017] Based on the target data, a second indicator is determined for the abnormal behavior of each of the second target customer groups;

[0018] The final target customer group is determined in each of the second target customer groups based on the second indicator.

[0019] According to a preferred embodiment of the present invention, the method further includes:

[0020] The final target customer group is monitored through preset strategies or preset models, and business log data of each target in the final target customer group is obtained in the business scenario.

[0021] Determine the percentage of targets exhibiting abnormal behavior based on the aforementioned business log data;

[0022] The preset indicators and their changing trends are adjusted and optimized based on the stated percentage.

[0023] According to a preferred embodiment of the present invention, the multi-source data acquisition of external data includes:

[0024] Select target external data channels based on business scenarios and information popularity;

[0025] Collect text data related to the pre-selected keywords from the target external data channels;

[0026] Analyze the relevance of the text data to the business scenario, and select target keywords from the pre-selected keywords based on the relevance.

[0027] Based on the target keywords, obtain external initial information related to the business scenario from external channels.

[0028] According to a preferred embodiment of the present invention, after filtering out the target related text in the initial information using a text analysis model, the method further includes:

[0029] Analyze the target-related text;

[0030] Configure corresponding quantitative indicators based on the analysis results;

[0031] Monitor and display the changes in the quantitative indicators.

[0032] According to a preferred embodiment of the present invention, the analysis of the target associated text includes:

[0033] Classify the target-related text.

[0034] Determine the first category corresponding to the target associated text;

[0035] Find business cases that match the first category from internal data;

[0036] The aforementioned business case will be analyzed.

[0037] To address the aforementioned technical problems, a second aspect of the present invention provides a customer group identification device based on multi-source data tagging, the device comprising:

[0038] The data acquisition module is used to collect multi-source data from internal and external data to obtain initial information related to the business scenario.

[0039] The filtering module is used to filter out the target related text in the initial information using a text analysis model;

[0040] The tagging module is used to extract effective information from the target associated text by combining automated text sentiment analysis and preset rules, and to tag the effective information to obtain tag data;

[0041] The segmentation module is used to locate the target corresponding to each tag data and divide each target into multiple primary target customer groups;

[0042] The analysis and determination module is used to analyze the preset indicators and the changing trends of the preset indicators of each first target customer group to determine the second target customer group.

[0043] According to a preferred embodiment of the present invention, the analysis and determination module includes:

[0044] The quantitative analysis module is used to quantitatively analyze the preset indicators of each primary target customer group to determine the distinguishability of the corresponding tag data;

[0045] The tracking and analysis module is used to track and analyze the changing trends of preset indicators for each first target customer group to determine the current security performance trend and security performance level of the corresponding first target customer group.

[0046] The comprehensive screening module is used to select a second target customer group from each first target customer group by comprehensively considering the preset indicators, discrimination, security performance and security performance trend.

[0047] According to a preferred embodiment of the present invention, the apparatus further includes:

[0048] The acquisition module is used to acquire target data for each target in each of the second target customer groups;

[0049] The evaluation module is used to determine a second indicator of abnormal behavior in each of the second target customer groups based on the target data;

[0050] The selection module is used to determine the final target customer group among each of the second target customer groups based on the second indicator.

[0051] According to a preferred embodiment of the present invention, the apparatus further includes:

[0052] The monitoring and acquisition module is used to monitor the final target customer group through a preset strategy or preset model, and acquire the business log data of each target in the final target customer group under the business scenario.

[0053] The sub-determination module is used to determine the percentage of targets exhibiting abnormal behavior based on the business log data.

[0054] The adjustment module is used to adjust and optimize the preset indicators and the trend of preset indicator changes based on the quantity ratio.

[0055] According to a preferred embodiment of the present invention, the acquisition module includes:

[0056] The first selection module is used to select target external data channels based on business scenarios and information popularity.

[0057] The sub-collection module is used to collect text data related to the pre-selected keywords from external data channels of the target.

[0058] The analysis and filtering module is used to analyze the relevance between the text data and the business scenario, and to filter target keywords from the pre-selected keywords based on the relevance.

[0059] The sub-acquisition module is used to acquire external initial information related to the business scenario from external channels based on the target keywords.

[0060] According to a preferred embodiment of the present invention, the apparatus further includes:

[0061] The first analysis module is used to analyze the target associated text;

[0062] The configuration module is used to configure the corresponding quantitative indicators based on the analysis results.

[0063] The monitoring and display module is used to monitor and display the changes in the quantitative indicators.

[0064] According to a preferred embodiment of the present invention, the first analysis module includes:

[0065] The sub-classification module is used to classify the target-related text.

[0066] The sub-classification module is used to determine the first category corresponding to the target associated text;

[0067] The sub-search module is used to find business cases that match the first category from internal data;

[0068] The sub-analysis module is used to analyze the business cases.

[0069] To solve the above-mentioned technical problems, a third aspect of the present invention provides an electronic device, comprising:

[0070] Processor; and

[0071] A memory storing computer-executable instructions, which, when executed, cause the processor to perform the method described in any of the preceding descriptions.

[0072] To address the aforementioned technical problems, a fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores one or more programs that, when executed by a processor, implement the above-described method.

[0073] This invention collects multi-source data from both internal and external sources to obtain initial public opinion information related to business scenarios. It then uses a text analysis model to filter target-related text from this initial information. Combined with automated text sentiment analysis and preset rules, it extracts effective information from the target-related text and tags this effective information to obtain multi-source tagged data containing both internal and external information. It locates the targets corresponding to each tagged data and divides each target into multiple first target customer groups. Finally, it determines second target customer groups by analyzing preset indicators and their changing trends for each first target customer group. Specifically, the multi-source tagged data of external information obtained through real-time multi-source data collection, filtering, and tagging can detect abnormal public opinion before product delivery and comprehensively and promptly identify abnormal customer groups. Similarly, the multi-source tagged data of internal information obtained through real-time multi-source data collection, filtering, and tagging can comprehensively and promptly identify abnormal customer groups after product delivery. This enables effective utilization of internal and external information related to business scenarios and full-link monitoring of abnormal customer groups within business scenarios, improving the timeliness and accuracy of customer group identification and thus providing timely and comprehensive protection for platform data and property security. Attached Figure Description

[0074] To make the technical problems solved by this invention, the technical means adopted, and the technical effects achieved clearer, specific embodiments of this invention will be described in detail below with reference to the accompanying drawings. However, it should be noted that the drawings described below are merely drawings of exemplary embodiments of this invention. Those skilled in the art can obtain drawings of other embodiments based on these drawings without any creative effort.

[0075] Figure 1 This is a flowchart illustrating a customer group identification method based on multi-source data tagging according to an embodiment of the present invention;

[0076] Figure 2 This is a schematic diagram illustrating the tag data obtained in an embodiment of the present invention;

[0077] Figure 3 This is a schematic diagram illustrating the customer group positioning in an embodiment of the present invention;

[0078] Figure 4 This is a schematic diagram of the structural framework of a customer group identification device based on multi-source data tagging according to an embodiment of the present invention;

[0079] Figure 5 This is a structural block diagram of an exemplary embodiment of an electronic device according to the present invention;

[0080] Figure 6 This is a schematic diagram of an embodiment of a computer-readable medium according to the present invention. Detailed Implementation

[0081] Exemplary embodiments of the invention will now be described more fully with reference to the accompanying drawings. While these exemplary embodiments can be implemented in various specific ways, they should not be construed as limiting the invention to the embodiments set forth herein. Rather, these exemplary embodiments are provided to make the content of the invention more complete and to facilitate a full communication of the inventive concept to those skilled in the art.

[0082] Subject to the inventive concept, the structures, performance, effects or other features described in a particular embodiment may be combined in any suitable manner with one or more other embodiments.

[0083] Please see Figure 1 , Figure 1 This invention provides a customer group identification method based on multi-dimensional data labeling, such as... Figure 1 As shown, the method includes:

[0084] S1. Collect multiple data from internal and external data respectively to obtain initial information related to the business scenario;

[0085] Internal data refers to data in the platform's internal database, primarily involving text records related to various aspects of the platform's business, and is clearly related to business scenarios. This can include records of browsing, applying for, or using platform products or services, records of product production, transportation, use, and settlement, etc. Figure 2 Different recorded data can be configured with different annotation information, and the corresponding text record data can be collected as internal initial information through the annotation information; for example, the text record data of product settlement can be collected as internal initial information based on the settlement annotation. This embodiment can obtain internal initial information by collecting internal data.

[0086] External data refers to public data other than internal data. This can be data publicly available on public platforms such as public social media platforms, public opinion platforms that match the business scenario, and designated public websites. Diversified collection of external data yields initial external information. The aforementioned initial internal information and initial external information together constitute the initial information.

[0087] For example, external data can be collected from various public platforms using distributed data acquisition technology. The multi-source data acquisition of external data includes:

[0088] S11. Select target external data channels based on business scenarios and information popularity;

[0089] In this embodiment, to pre-filter information strongly related to the business scenario, it is necessary to consider both business scenario relevance and information popularity to ensure that the information data source has a high degree of abnormal topic aggregation and high popularity of topics related to the platform's business. At the same time, based on considerations for user privacy, such as... Figure 2 It can select one or more platforms from public platforms that are relevant to the business scenario and have high information popularity based on preset business keywords and platform topic popularity as target external data channels. Among them, public platforms can be: public social platforms, public opinion platforms that match the business scenario, designated public websites, etc.

[0090] S12. Collect text data related to the pre-selected keywords from the target external data channels;

[0091] For example, a keyword set can be pre-configured, which includes multiple pre-selected keywords that may be mistakenly hit but are highly relevant to the business scenario, such as black market, fraud, and cyberattack. Based on the pre-selected keywords, a certain amount of keyword text can be collected from external data channels as text data related to the pre-selected keywords.

[0092] S13. Analyze the relevance between the text data and the business scenario, and select target keywords from the pre-selected keywords based on the relevance.

[0093] For example, the similarity between keyword text and business scenario can be analyzed to determine their relevance. In addition, the relevance between keyword text and business scenario can be calculated using the Word2Vec model. Furthermore, the validity of the relevance data can be judged by combining empirical models, and then the pre-selected keywords corresponding to keyword text with a relevance greater than the threshold can be selected as target keywords based on the valid relevance data. At the same time, false hits with low relevance can be excluded from the text data.

[0094] S14. Obtain external initial information related to the business scenario from external channels based on the target keywords.

[0095] External channels can include various public information platforms such as public social media platforms, public opinion platforms that match the business scenario, designated public websites, and the aforementioned target external channels.

[0096] For example, such as Figure 2Distributed data acquisition technology can be used to collect data comprehensively from multiple external channels (such as various important external public social platforms, abnormal public opinion platforms, etc.), thereby enabling real-time or near-real-time diversified data collection from various external channels. Specifically, distributed data acquisition technology involves installing data acquisition programs on multiple computers, sharing queues, and removing duplicate data, ensuring that multiple data acquisition programs do not retrieve already acquired content, thus achieving joint data collection.

[0097] Thus, we have collected initial external information related to business scenarios from diversified external channels.

[0098] S2. Filter out the target related text from the initial information using a text analysis model;

[0099] For example, such as Figure 2 Natural Language Processing (NLP) technology can be used to build a text analysis model. By training the text analysis model with key information of customer group identification, it can accurately identify strongly related text, thereby filtering out strongly related target text from the initial information.

[0100] Furthermore, this step can also perform quantitative analysis and visualization of the target-related text. The quantitative analysis results can be clearly presented to staff, who can directly configure preset indicators based on the quantitative analysis. Therefore, after step S2, the method may further include:

[0101] S21. Analyze the target associated text;

[0102] For example, this step may include:

[0103] S201. Classify the target-related text.

[0104] For example, by pre-setting keywords of different levels, the target related text is decomposed, and the level of keyword hits of the decomposed target related text is determined. Target related texts that hit the same level of keywords are grouped into one category. The target related texts of the same category and their hit level keywords are evaluated to determine the accuracy of the classification. The evaluation data is automatically stored in the underlying data table of the database for easy subsequent management and quantitative analysis.

[0105] S202. Determine the first category corresponding to the target associated text;

[0106] After obtaining the classification, the order of magnitude of target-related texts in each category in the recent period (within the current preset time period) is collected and analyzed. Categories with an order of magnitude greater than the preset number are designated as the first category, indicating that there are more target-related texts in the first category and they cover a large amount of information.

[0107] S203. Find business cases that match the first category from internal data;

[0108] For example: Each business case in the internal data has labeled information, such as... Figure 2 This allows matching the target-related text in the first category with the annotation information to obtain business cases that match the first category.

[0109] S204. Analyze the aforementioned business cases.

[0110] For example, it can analyze various metrics of business cases, and users can configure quantitative metrics based on the analysis results.

[0111] S22. Configure corresponding quantitative indicators based on the analysis results;

[0112] For example, configure quantitative metrics for historical internal cases, such as product application approval rate, product usage rate, product settlement rate, etc.

[0113] S23. Monitor and display the changes in the quantitative indicators.

[0114] for example Figure 2 One-stop self-service BI tools can be used to visualize changes in quantitative indicators.

[0115] The interface provides a simple preset indicator configuration interface by visually displaying the changes in quantitative indicators. Users can configure various quantitative indicators that need to be analyzed in this interface and view the changes in the quantitative indicators, which makes it convenient to configure preset indicators.

[0116] S3. Combine automated text sentiment analysis and preset rules to extract effective information from the target associated text, and perform tagging processing on the effective information to obtain tag data;

[0117] The process involves automated text sentiment analysis, which combines natural language processing, text mining, and computer language techniques to extract sentiment information from target-related text. For example, keywords can be labeled first, and then different models can be used to expand these keywords to other new words, generating a sentiment dictionary for sentiment polarity classification to obtain the sentiment information of the target-related text. This sentiment information reflects the reliability level of the target-related text. Preset rules are used to determine the reliability level of the related text information. These rules can be configured based on experience; for example, different levels of reliable keywords can be preset, and the reliability level of the related text information is determined based on the frequency and level of the reliable keywords. Finally, the reliability level obtained from automated text sentiment analysis is merged with the reliability level obtained from the preset rules to obtain the final reliability level of the target-related text. Based on the final reliability level, valid information is selected; for example, target-related text with a final reliability level greater than the preset reliability level is considered valid information. The valid information is then labeled according to its reliability level to obtain labeled data.

[0118] Thus, we obtained tag data that allows for accurate targeting of customer groups.

[0119] S4. Locate the target corresponding to each tag data, and divide each target into multiple primary target customer groups;

[0120] The target can be the terminal device used to apply for, use, or settle payments on the platform's products, or it can be the user account used to apply for, use, or settle payments on the platform's products. For example... Figure 3 This step utilizes a text classification model to locate the target corresponding to each tag. This text classification model, trained on a large amount of text, can classify text based on the aforementioned tags, thereby obtaining the target corresponding to each tag. For example, device text data related to applying for, using, or settling on a platform product can be input into the text classification model to obtain the target corresponding to each tag. Targets corresponding to the same tag are grouped into the same first target customer group, thus each tag corresponds to a first target customer group.

[0121] S5. Analyze the preset indicators and the changing trends of the preset indicators for each first target customer group to determine the second target customer group.

[0122] In this embodiment, the preset indicator is used to determine whether the tag differentiation of the first target customer group is significant. The trend of the preset indicator is used to determine whether the current abnormal behavior of the first target customer group is significant. If the target in the first target customer group is a terminal, the abnormal behavior may be illegal data acquisition or unauthorized data acquisition. If the target in the first target customer group is a user account, the abnormal behavior may be frequent login in different locations or IP switching. The preset indicator may be a business data indicator related to the abnormal behavior of the first target customer group, such as: product usage rate, i.e., the proportion of the first target customer group that applies for and uses the product to the total number of target customers who apply for the product; product settlement rate, i.e., the proportion of the first target customer group that uses and settles the product to the total number of target customers who use the product, etc. In specific applications, one or more preset indicators can be configured. When there are multiple preset indicators, a corresponding weight ratio can be configured for each preset indicator, and all preset indicators are weighted and summed to obtain the final preset indicator.

[0123] For example, this step may include:

[0124] S51. Quantitatively analyze the preset indicators of each primary target customer group to determine the distinguishability of the corresponding tag data;

[0125] For example, different threshold values ​​for different discrimination levels can be pre-selected. Statistical tools can be used to quantitatively analyze the preset indicators of each first target customer group. The preset indicators of each first target customer group can be compared with the threshold values ​​of each discrimination level to determine the discrimination level of the preset indicators of each first target customer group, which is the label discrimination level of each first target customer group.

[0126] S52. Track and analyze the changing trends of preset indicators for each first target customer group to determine the current safety performance trend and safety performance level of the corresponding first target customer group;

[0127] For example, we can analyze the changing trends of preset indicators in each primary target customer group at multiple points in time within the current preset time period to determine the changing trends of preset indicators. Based on these trends, we can determine the current safety performance trend and safety performance level. For example, a continuous increase in preset indicators corresponds to the first level of safety performance, a decrease followed by an increase corresponds to the second level of safety performance, an increase followed by a decrease corresponds to the third level of safety performance, and a continuous decrease in preset indicators corresponds to the fourth level of safety performance.

[0128] S53. Based on the preset indicators, discrimination, security performance and security performance trend, a second target customer group is selected from each first target customer group.

[0129] When providing services to a customer group, given certain conditions, the more stable the data of the target within that group, the better it meets the system requirements. Therefore, to select a second target customer group from the first target customer group, we can configure indicator thresholds, discrimination thresholds, security performance thresholds, and security performance trend requirements. From the first target customer group, we can select the second target customer group whose preset indicators are greater than the indicator threshold, discrimination is greater than the discrimination threshold, security performance is greater than the security performance threshold, and security performance trend conforms to the security performance trend threshold range.

[0130] After identifying the second target customer group, the final target customer group can be further identified based on the device data of the second target customer group to improve the positioning accuracy of the customer group. Therefore, the method may further include:

[0131] S61. Obtain target data for each target in each of the second target customer groups;

[0132] If the target in the second target customer group is a terminal device, then the target data includes: device model, device data selected by the user for public disclosure, and the device data may include: device location, device user information, device communication information, etc. If the target in the second target customer group is a user account, then the target data includes: account IP address, account login address, account login time, etc.

[0133] S62. Determine a second indicator of abnormal behavior for each of the second target customer groups based on the target data;

[0134] Among them, the second indicator can be used to predict the abnormal behavior of the target. For example, the data of each target can be input into the abnormal behavior model for analysis, the second indicator of each target can be output, and the average value of the second indicators of each target can be used as the second indicator of the second target customer group, thereby determining the probability of abnormal behavior of each second target customer group.

[0135] S63. Determine the final target customer group in each of the second target customer groups based on the second indicator.

[0136] For example, the second target customer group whose second indicator is greater than the preset indicator is taken as the final target customer group, thereby accurately locating the customer group with abnormal behavior.

[0137] Furthermore, the present invention can also adjust and optimize preset indicators and preset indicator change trend threshold ranges based on the business logs of the final target customer group, further improving positioning accuracy. The method further includes:

[0138] S71. Monitor the final target customer group through a preset strategy or preset model, and obtain the business log data of each target in the final target customer group under the business scenario;

[0139] Specifically: Preset strategies can monitor target behavior within the final target customer group. For example, a preset strategy can be configured with multiple abnormal behavior judgment rules. When the real-time behavior of a target matches a certain judgment rule, an alarm is triggered, allowing the platform to control the abnormal behavior. Preset models can predict abnormal behavior within the final target customer group. After inputting the target data of the final target customer group into the preset model, it outputs the probability value of abnormal behavior occurring in the final target customer group. When this probability value is greater than the preset probability, an alarm is triggered, allowing the platform to control the abnormal behavior.

[0140] Without triggering an alarm, technologies such as data tracking can be used to automatically obtain business log data of each target within the final target customer group in the stated business scenario.

[0141] S72. Determine the percentage of targets exhibiting abnormal behavior based on the business log data;

[0142] In this embodiment, if the target in the first target customer group is a terminal, the abnormal behavior could be illegal data acquisition or unauthorized data acquisition. If the target in the first target customer group is a user account, the abnormal behavior could be frequent logins from different locations or IP switching. This step parses the data acquisition method or IP login method in the business log data to determine the number of targets exhibiting abnormal behavior and obtain the percentage of targets exhibiting abnormal behavior.

[0143] S73. Adjust and optimize the preset indicators and the preset indicator change trend threshold range according to the quantity ratio.

[0144] In this embodiment, a percentage threshold can be preset, and the quantity percentage can be compared with the percentage threshold. When the quantity percentage is less than the percentage threshold, it indicates that the abnormal behavior of the final target customer group is not prominent enough, and the preset indicators for screening the second target customer group need to be adjusted, thereby adjusting the second target customer group and then adjusting the final target customer group.

[0145] For example, when there is only one preset indicator and the proportion of the quantity is less than the proportion threshold, another preset indicator is used to replace the original preset indicator, or at least one other preset indicator is added. When there are multiple preset indicators and the proportion of the quantity is less than the proportion threshold, the weight proportion of different preset indicators can be adjusted. By adjusting the weight proportion, the focus on the threshold range of the preset indicator change trend can be adjusted, thereby adjusting the range requirement of the preset indicator change trend and thus completing the adjustment of the second target customer group.

[0146] Figure 4 This invention relates to a customer group identification device based on multi-source data tagging, such as... Figure 4 As shown, the device includes:

[0147] The data acquisition module 41 is used to collect multi-source data from internal and external data respectively to obtain initial information related to the business scenario.

[0148] The filtering module 42 is used to filter out the target related text in the initial information through a text analysis model;

[0149] The tagging module 43 is used to extract effective information from the target associated text by combining automated text sentiment analysis and preset rules, and to tag the effective information to obtain tag data;

[0150] The segmentation module 44 is used to locate the target corresponding to each tag data and divide each target into multiple first target customer groups;

[0151] The analysis and determination module 45 is used to analyze the preset indicators and the changing trends of the preset indicators of each first target customer group to determine the second target customer group.

[0152] According to a preferred embodiment of the present invention, the analysis and determination module 45 includes:

[0153] The quantitative analysis module is used to quantitatively analyze the preset indicators of each primary target customer group to determine the distinguishability of the corresponding tag data;

[0154] The tracking and analysis module is used to track and analyze the changing trends of preset indicators for each first target customer group to determine the current security performance trend and security performance level of the corresponding first target customer group.

[0155] The comprehensive screening module is used to select a second target customer group from each first target customer group by comprehensively considering the preset indicators, discrimination, security performance and security performance trend.

[0156] Furthermore, the device also includes:

[0157] The acquisition module is used to acquire target data for each target in each of the second target customer groups;

[0158] The evaluation module is used to determine a second indicator of abnormal behavior in each of the second target customer groups based on the target data;

[0159] The selection module is used to determine the final target customer group among each of the second target customer groups based on the second indicator.

[0160] Furthermore, the device also includes:

[0161] The monitoring and acquisition module is used to monitor the final target customer group through a preset strategy or preset model, and acquire the business log data of each target in the final target customer group under the business scenario.

[0162] The sub-determination module is used to determine the percentage of targets exhibiting abnormal behavior based on the business log data.

[0163] The adjustment module is used to adjust and optimize the preset indicators and the trend of preset indicator changes based on the quantity ratio.

[0164] Optionally, the acquisition module 41 includes:

[0165] The first selection module is used to select target external data channels based on business scenarios and information popularity.

[0166] The sub-collection module is used to collect text data related to the pre-selected keywords from external data channels of the target.

[0167] The analysis and filtering module is used to analyze the relevance between the text data and the business scenario, and to filter target keywords from the pre-selected keywords based on the relevance.

[0168] The sub-acquisition module is used to acquire external initial information related to the business scenario from external channels based on the target keywords.

[0169] Furthermore, the device also includes:

[0170] The first analysis module is used to analyze the target associated text;

[0171] The configuration module is used to configure the corresponding quantitative indicators based on the analysis results.

[0172] The monitoring and display module is used to monitor and display the changes in the quantitative indicators.

[0173] The first analysis module includes:

[0174] The sub-classification module is used to classify the target-related text.

[0175] The sub-classification module is used to determine the first category corresponding to the target associated text;

[0176] The sub-search module is used to find business cases that match the first category from internal data;

[0177] The sub-analysis module is used to analyze the business cases.

[0178] Those skilled in the art will understand that the modules in the above-described device embodiments can be distributed throughout the device as described, or they can be modified accordingly and distributed in one or more devices different from the above embodiments. The modules in the above embodiments can be combined into one module, or they can be further divided into multiple sub-modules.

[0179] Figure 5This is a structural block diagram of an exemplary embodiment of an electronic device according to the present invention. Figure 5 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of the present invention.

[0180] like Figure 5 As shown, the electronic device 500 of this exemplary embodiment is manifested in the form of a general data processing device. The components of the electronic device 500 may include, but are not limited to: at least one processing unit 510, at least one storage unit 520, a bus 530 connecting different electronic device components (including storage unit 520 and processing unit 510), a display unit 540, etc.

[0181] The storage unit 520 stores a computer-readable program, which may be source code or read-only code. The program can be executed by the processing unit 510, causing the processing unit 510 to perform the steps of various embodiments of the present invention. For example, the processing unit 510 can perform actions such as... Figure 1 The steps are shown.

[0182] Bus 530 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0183] Figure 6 This is a schematic diagram of a computer-readable medium embodiment of the present invention. Figure 6 As shown, the computer program may be stored on one or more computer-readable media.

[0184] The computer-readable medium can be a readable signal medium or a readable storage medium. When the computer program is executed by one or more data processing devices, the computer-readable medium enables the implementation of the above-described method of the present invention, namely: performing multi-source data collection on internal and external data respectively to obtain initial information related to the business scenario; filtering out target-related text in the initial information through a text analysis model; extracting effective information from the target-related text by combining automated text sentiment analysis and preset rules, and tagging the effective information to obtain tag data; locating the target corresponding to each tag data, and dividing each target into multiple first target customer groups; analyzing the preset indicators and the changing trends of the preset indicators of each first target customer group to determine the second target customer group.

[0185] This invention can be implemented by a method, apparatus, electronic device, or computer-readable medium that executes a computer program. In practice, some or all of the functions of this invention can be implemented using general-purpose data processing devices such as microprocessors or digital signal processors (DSPs).

[0186] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the present invention is not inherently related to any specific computer, virtual device, or electronic device, and various general-purpose devices can also implement the present invention. The above descriptions are merely specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for customer recognition based on multi-dimensional data tagging, characterized in that, The method includes: The internal data is collected from multiple sources to obtain initial internal data related to the business scenario; based on preset business keywords and topic popularity, multiple platforms related to the business scenario and with information popularity higher than a threshold are selected from public social platforms or public opinion platforms that match the business scenario as target external data channels, and multiple external data channels are collected from multiple target external data channels to obtain initial external information related to the business scenario; the internal initial information and the external initial information are combined to form initial information. The target related text in the initial information is filtered out using a text analysis model; Sentiment information reflecting the reliability level of target-related text information is obtained based on automated text sentiment analysis; the reliability level of related text information is determined based on preset rules, and the reliability level reflected by the sentiment information and the reliability level determined by the preset rules are merged to obtain the final reliability level of the target-related text; effective information in the target-related text is extracted based on the final reliability level, and the effective information is tagged to obtain diversified external information tag data that can detect abnormal public opinion and identify targets before the product is provided, and diversified internal information tag data that can identify targets after the product is provided. Locate the target corresponding to each tag data, and divide each target into multiple primary target customer groups; The second target customer group is determined by analyzing the preset indicators and the trend of the preset indicators of each first target customer group; the preset indicators are used to determine whether the label differentiation of the first target customer group is obvious, and the trend of the preset indicators is used to determine whether the current abnormal behavior of the first target customer group is obvious.

2. The method according to claim 1, characterized in that, The analysis of preset indicators and their changing trends for each first target customer group to determine the second target customer group includes: Quantitative analysis of preset indicators for each primary target customer group determines the discriminative power of the corresponding tag data; Track and analyze the changing trends of preset indicators for each primary target customer group to determine the current security performance trend and security performance level of the corresponding primary target customer group; The second target customer group is selected from each of the first target customer groups by combining the preset indicators, discrimination, security performance and security performance trend.

3. The method according to claim 2, characterized in that, The method further includes: Obtain target data for each target within each of the second target customer groups; Based on the target data, a second indicator is determined for the abnormal behavior of each of the second target customer groups; The final target customer group is determined in each of the second target customer groups based on the second indicator.

4. The method according to claim 3, characterized in that, The method further includes: The final target customer group is monitored through preset strategies or preset models, and business log data of each target in the final target customer group is obtained in the business scenario. Determine the percentage of targets exhibiting abnormal behavior based on the aforementioned business log data; The preset indicators and their changing trends are adjusted and optimized based on the stated proportions.

5. The method according to claim 1, characterized in that, The method of obtaining initial external information related to the business scenario through diversified data collection from multiple target external data channels includes: Collect text data related to the pre-selected keywords from the target external data channels; Analyze the relevance of the text data to the business scenario, and select target keywords from the pre-selected keywords based on the relevance. Based on the target keywords, obtain external initial information related to the business scenario from external channels.

6. The method according to claim 1, characterized in that, After filtering out the target related text from the initial information using a text analysis model, the method further includes: Analyze the target-related text; Configure corresponding quantitative indicators based on the analysis results; Monitor and display the changes in the quantitative indicators.

7. The method according to claim 6, characterized in that, The analysis of the target associated text includes: Classify the target associated text and determine the first category corresponding to the target associated text; Find business cases that match the first category from internal data; The aforementioned business case will be analyzed.

8. A customer group identification device based on multi-source data labeling, characterized in that, The device includes: The data acquisition module is used to collect internal data from multiple sources to obtain internal initial data related to the business scenario; based on preset business keywords and topic popularity, it selects multiple platforms from public social platforms or public opinion platforms that match the business scenario and whose information popularity is higher than a threshold as target external data channels, and collects external initial information related to the business scenario from multiple target external data channels; and combines the internal initial information and external initial information to form initial information. The filtering module is used to filter out the target related text in the initial information using a text analysis model; The tagging module is used to obtain sentiment information reflecting the reliability level of target-related text information based on automated text sentiment analysis; determine the reliability level of related text information based on preset rules; merge the reliability level reflected by the sentiment information and the reliability level determined by the preset rules to obtain the final reliability level of the target-related text; extract effective information from the target-related text based on the final reliability level; and perform tagging processing on the effective information to obtain diversified external information tag data that can detect abnormal public opinion and identify targets before the product is provided, and diversified internal information tag data that can identify targets after the product is provided. The segmentation module is used to locate the target corresponding to each tag data and divide each target into multiple primary target customer groups; The analysis and determination module is used to analyze the preset indicators and the trend of preset indicator changes of each first target customer group to determine the second target customer group; the preset indicators are used to determine whether the label differentiation of the first target customer group is obvious, and the trend of preset indicator changes is used to determine whether the current abnormal behavior of the first target customer group is obvious.

9. The apparatus according to claim 8, characterized in that, The analysis and determination module includes: The quantitative analysis module is used to quantitatively analyze the preset indicators of each primary target customer group to determine the distinguishability of the corresponding tag data; The tracking and analysis module is used to track and analyze the changing trends of preset indicators for each first target customer group to determine the current security performance trend and security performance level of the corresponding first target customer group. The comprehensive screening module is used to select a second target customer group from each first target customer group by comprehensively considering the preset indicators, discrimination, security performance and security performance trend.

10. The apparatus according to claim 9, characterized in that, The device further includes: The acquisition module is used to acquire target data for each target in each of the second target customer groups; The evaluation module is used to determine a second indicator of abnormal behavior in each of the second target customer groups based on the target data; The selection module is used to determine the final target customer group among each of the second target customer groups based on the second indicator.

11. The apparatus according to claim 10, characterized in that, The device further includes: The monitoring and acquisition module is used to monitor the final target customer group through a preset strategy or preset model, and acquire the business log data of each target in the final target customer group under the business scenario. The sub-determination module is used to determine the percentage of targets exhibiting abnormal behavior based on the business log data. The adjustment module is used to adjust and optimize the preset indicators and the trend of preset indicator changes based on the quantity ratio.

12. The apparatus according to claim 8, characterized in that, The acquisition module includes: The sub-collection module is used to collect text data related to the pre-selected keywords from external data channels of the target. The analysis and filtering module is used to analyze the relevance between the text data and the business scenario, and to filter target keywords from the pre-selected keywords based on the relevance. The sub-acquisition module is used to acquire external initial information related to the business scenario from external channels based on the target keywords.

13. The apparatus according to claim 8, characterized in that, The device further includes: The first analysis module is used to analyze the target associated text; The configuration module is used to configure the corresponding quantitative indicators based on the analysis results. The monitoring and display module is used to monitor and display the changes in the quantitative indicators.

14. The apparatus according to claim 13, characterized in that, The first analysis module includes: The sub-classification module is used to classify the target-related text. The sub-classification module is used to determine the first category corresponding to the target associated text; The sub-search module is used to find business cases that match the first category from internal data; The sub-analysis module is used to analyze the business cases.

15. An electronic device comprising: processor; as well as A memory storing computer-executable instructions, which, when executed, cause the processor to perform the method according to any one of claims 1 to 7.

16. A computer-readable storage medium, wherein, The computer-readable storage medium stores one or more programs that, when executed by a processor, implement the method of any one of claims 1 to 7.

Citation Information

Patent Citations

  • Risk early warning method and device for predicting customer group qualification change and electronic equipment

    CN110335138A

  • Creation method and device of specified tag customer group, electronic equipment and storage medium

    CN114003704A