TWI854521B - Attention list collating and analyzing method and system - Google Patents

Abstract

An attention list collating and analyzing method, configured to analyze and compare news articles from at least one news media source to generate an attention list, comprises the following steps of: a news collection module collecting news articles from the at least one news media source; a concerned subject comparison module identifying whether the news articles have a concerned subject and a concerned issue, and capturing at least one keyword from news articles having the concerned subject and the concerned issue; a duplicate information comparison module comparing the received news articles having the concerned subject and the concerned issue and the at least one keyword to the destination file respectively, and updating the destination file according to the result of the comparison between each news article and the destination file; and, an integration module generates the attention list and related information according to the destination file.

Description

Focus list compilation and analysis method and system

The present invention relates to an information analysis method and system, in particular to a method and system for compiling and analyzing a follow list for analyzing characters appearing in news media information sources.

With the globalization of the economy and the rapid development of information technology, financial crime cases such as money laundering and fraud have increased year by year and are gradually becoming more organized and professional. In order to respond to the anti-money laundering plan and reduce the number of money laundering and fraud cases, banks have begun to strengthen the Know Your Customer/Customer Due Diligence (KYC/CDD) process. KYC/CDD is the process by which banks get to know each customer. It generally refers to the process by which banks collect documents about customers and search online databases before customers first open an account with financial institutions or sign a business agreement to understand the customer's identity, background, source of income and funds, financial status, and whether they have been sanctioned and have related negative news, etc. In other words, the main function of KYC/CDD is that financial institutions conduct risk assessments based on the information obtained from the documents submitted by customers, and can determine whether to approve the customer's account opening application or business agreement based on the risk assessment results. When a customer is classified as a high-risk category, the account opening may be refused, the business agreement may be rejected, or enhanced review may be conducted. If after review, the financial institution approves the customer's account opening application or business agreement, but it is still a high-risk person for suspicious transactions, it may restrict its future financial transactions and regularly track its financial transactions and search for negative news related to it.

The above-mentioned method of tracking relevant news about customers is generally to search directly on the Internet, especially on news websites that provide a large amount of news, which are more likely to have news about the aforementioned high-risk customers with suspicious transactions. However, due to the huge amount of information on the Internet today, when entering keywords on the Internet platform to search for negative news related to the aforementioned high-risk customers with suspicious transactions, too much complex data is often obtained, and it is not easy to quickly and accurately obtain important information that meets the search needs, and the search results are usually accompanied by advertising information that has very little relevance to the information to be queried.

The news results obtained by searching for keywords mentioned above are currently obtained through manual comparison, screening, and elimination of duplicates to obtain the required information. The above-mentioned comparison, screening, and elimination of duplicate processes include: (i) When there are multiple news articles and media data about the same event, the content of the news articles you want to focus on is highly repetitive, but the content of each piece of data is slightly different, so it is necessary to perform data comparison to screen out the list related to the same news event and eliminate duplicate data; (ii) The names of people appearing in news media reports, news articles, and data do not all match the news figures related to the search target, so they need to be reviewed and eliminated; (iii) The new list overlaps with the previous list, and the same name may be It can appear repeatedly, so it is necessary to judge whether each piece of data belongs to the same news event; (iv) As the same news event develops over time, the status and identity of the same person may change. For example, a person who was originally only a suspect may become a criminal after being prosecuted, sentenced or convicted. Therefore, it is necessary to collect litigation information and mark specific negative news persons who meet the preset conditions; (v) When the subject of negative news is a non-natural person, whether the status of the non-natural person is still valid and whether it has been dissolved or reorganized needs to be verified through relevant public information channels and platforms. The above process is quite complicated. If it relies solely on manual judgment and repeated comparison, it is very likely to have a high error rate and consume a lot of time.

Therefore, it is necessary to develop an information analysis system that can quickly obtain useful and required news media data from Internet search results to solve the problems of previous technologies.

In view of this, the present invention provides a method and system for sorting and analyzing a list of concerns to solve the above-mentioned problems.

According to a specific embodiment of the present invention, the method for sorting and analyzing the follow list is used to analyze and compare a plurality of news articles from at least one news media information source to generate a follow list and related information, which includes the following steps: a news collection module collects news articles from at least one news media information source; a follow subject comparison module uses natural language technology to identify whether the news articles have a follow subject, and then identifies whether the content of the news articles with the follow subject meets the requirements of the news content. Focus on the topic, and extract at least one keyword from the news articles with the focus subject and matching the focus topic; the duplicate information matching module receives each news article with the focus subject and matching the focus topic and the corresponding at least one keyword from the focus subject matching module, and matches the received news articles with the target file respectively, and updates the target file according to the comparison results of each news article and the target file; and the integration module generates a list of focus personnel and related information according to the target file.

Among them, the method for sorting and analyzing the follow list further includes the following steps: the automatic delivery module sends the follow person list and related information generated by the target file to the outside in the form of a message notification.

Among them, users input the subject of attention and set the topic of attention through the user interface.

Among them, the subject comparison module compares the news articles with the cross-comparison information sources through the subject comparison model, and the subject comparison model is generated through the machine learning algorithm; the cross-comparison information sources further include business databases, value-added information databases, and websites for querying digital footprints.

Among them, the duplicate information comparison module further includes the following steps according to the comparison results of each news article and the target file: when the comparison result of the first news article in the news article and the target file is that the focus subject is repeated and there is no information difference, the duplicate information comparison module deletes the first news article; when the comparison result of the first news article and the target file is that the focus subject is repeated and there is an information change, the duplicate information comparison module adds the information change to the target file; when the comparison result of the first news article and the target file is that the focus subject is not repeated, the duplicate information comparison module adds the first news article to the target file.

Among them, the attention list sorting and analysis method further includes the following steps: the repeated information comparison module compares the target files respectively with the news articles received through the information comparison model, wherein the information comparison model is generated through a supervised machine learning algorithm.

Another scope of the present invention is to provide a follow list sorting and analysis system. According to another specific embodiment, the follow list sorting and analysis system is used to analyze a plurality of news articles from at least one news media information source to generate a follow list and related information thereof, and includes a news collection module, a follow subject comparison module, a duplicate information comparison module, and an integration module. The news collection module is used to connect to at least one news media information source to collect news articles from at least one news media information source. The follow subject comparison module is coupled to the news collection module to receive news articles. The subject comparison module is used to identify whether a news article has a subject of interest through natural language technology, and then identify whether the content of the news article with the subject of interest meets the topic of interest, and extract at least one keyword from the news article with the subject of interest and meeting the topic of interest. The duplicate information comparison module is coupled to the subject of interest comparison module. The duplicate information comparison module is used to receive each news article with the subject of interest and meeting the topic of interest and the corresponding at least one keyword from the subject of interest comparison module, and compare the received news articles with a target file, and update the target file according to the comparison results of each news article and the target file. The integration module is coupled to the duplicate information comparison module. The integration module is used to generate a list of concerned personnel and their related information based on the target file.

Among them, the follow list sorting and analysis system further includes an automatic delivery module. The automatic delivery module is coupled with the integration module to send the follow list and related information generated by the target file to the outside in the form of message notification.

Among them, the subject comparison module is coupled to the cross-comparison information source. The subject comparison module compares the news article with the cross-comparison information source. The cross-comparison information source further includes a business database, a value-added information database, and a website for querying digital footprints.

The subject comparison module includes a subject comparison model. The duplicate information comparison module includes an information comparison model. The subject comparison model and the information comparison model are generated by a supervised machine learning algorithm. The subject comparison model is used to compare the news articles with the cross-comparison information source, and the information comparison model is used to compare the received news articles with the target files respectively.

In summary, the present invention provides a method and system for sorting and analyzing a follow list, wherein a follow subject comparison module selects news articles that match the user's follow subject and follow topic, and then a duplicate information comparison module compares the selected news articles with a target file, and updates the target file according to the comparison result. The present invention can quickly obtain useful and required news media data information from network search results through a follow list sorting and analysis method and system. In addition, by introducing artificial intelligence methods, supervised machine learning is used to reduce the error rate and duplication rate of the follow list and exclude invalid information, thereby reducing the human error rate and the time spent on manual interpretation, thereby improving efficiency and reducing time and labor costs. In addition, users can receive the latest list of people to follow and their related information in a timely or regular manner through the automatic delivery module, so users do not need to spend extra time searching again, thereby reducing manpower and time costs and improving the overall search and retrieval efficiency.

FIG1 is a functional block diagram of a focus list sorting and analysis system according to one specific embodiment of the present invention.

FIG2 is a flowchart showing the steps of a method for sorting and analyzing a list of concerns according to a specific embodiment of the present invention.

FIG3 is a flowchart showing the steps of a method for sorting and analyzing a list of concerns according to another specific embodiment of the present invention.

FIG4 is a flowchart showing a method for sorting and analyzing a list of concerns according to another specific embodiment of the present invention.

In order to make the advantages, spirit and features of the present invention easier and clearer to understand, the following will be described and discussed in detail with reference to the attached drawings using specific embodiments. It is worth noting that these specific embodiments are only representative specific embodiments of the present invention, and the specific methods, devices, conditions, materials, etc. cited therein are not used to limit the present invention or the corresponding specific embodiments. In addition, the devices in the figure are only used to express their relative positions and are not drawn according to their actual proportions, which should be explained in advance.

Please refer to FIG1. FIG1 is a functional block diagram of a follow list sorting and analysis system 1 according to a specific embodiment of the present invention. The specific embodiment provides a follow list sorting and analysis method and system for analyzing and comparing news articles of news media information sources to generate a follow list and related information. As shown in FIG1, in the specific embodiment, the follow list sorting and analysis system 1 includes a news collection module 11, a follow subject comparison module 12, a duplicate information comparison module 13 and an integration module 14. The news collection module 11 is used to couple at least one external news media information source 21. The follow subject comparison module 12 is coupled to the news collection module 11, the duplicate information comparison module 13 is coupled to the follow subject comparison module 12, and the integration module 14 is coupled to the duplicate information comparison module 13. In addition, the subject comparison module 12 in this specific embodiment is further coupled to the external cross-comparison information source 22. The subject comparison module 12 includes a subject comparison model 121; the duplicate information comparison module 13 includes an information comparison model 131. Through the follow list sorting and analysis system 1, news articles can be collected and analyzed from the news media information source 21 to obtain relevant information about the people or customers you want to follow from these news articles, as a basis for monitoring high-risk customers, and even to achieve the effect of money laundering or fraud prevention. The following will be used in conjunction with the follow list sorting and analysis method to explain the functions of each module in the follow list sorting and analysis system 1 of this specific embodiment. In practice, the news collection module of the attention list sorting and analysis system 1 can be a network module in a computer system to connect to external news media information sources; the attention subject matching module, duplicate information matching module and integration module can be integrated into the central processing unit of the computer system or cloud system, or integrated into an integrated chip.

Please refer to FIG. 1 and FIG. 2 together. FIG. 2 is a flowchart showing the steps of the method for sorting and analyzing the attention list according to a specific embodiment of the present invention. Please note that the steps of the method for sorting and analyzing the attention list in FIG. 2 can be achieved through the system for sorting and analyzing the attention list 1 in FIG. 1 , and therefore, the following describes the steps of the method for sorting and analyzing the attention list in the specific embodiment of FIG. 2 by using the structure of the system for sorting and analyzing the attention list 1 in FIG. 1 . As shown in FIG2 , in this specific embodiment, the method for sorting and analyzing the attention list includes step S1: the news collection module 11 collects news articles from the news media information source 21; step S2: the attention subject comparison module 12 uses natural language technology to identify whether the news articles have attention subjects, and then identifies whether the content of the news articles with attention subjects meets the attention topics, and compares the news articles with attention subjects and meeting the attention topics. The news article of the subject is extracted to extract keywords; Step S3: The duplicate information matching module 13 receives each news article and corresponding keywords that has a subject of interest and meets the subject of interest from the subject of interest matching module 12, and compares the received news articles with the target file respectively, and updates the target file according to the comparison results of each news article with the target file; and Step S4: The integration module 14 generates a list of followers and their related information according to the target file.

In practice, the news collection module 11 of the attention list sorting and analysis system 1 can be connected to one or more news media information sources 21, or it can also automatically collect news articles and data from various news media information sources on the Internet through a web crawler program. If non-text data such as audio and video files appear, the relevant data can be converted into text in a timely manner to obtain text content, and a news article list can be further generated through the collected news articles and related data. In addition, when the subject comparison module 12 performs identification and analysis on the collected news articles and related data, if the subject does not appear in the content of the news article, the identification and analysis of the next news article in the news article list is performed; if the subject appears, the identification and analysis is performed on whether the content of this news article conforms to the topic of interest, so as to screen out news articles that have both the key subject and the key topic. In addition, in practice, the types of news media information sources 21 connected to the news collection module 11 can be news text content and news website URLs, such as online news media platforms and online newspaper and magazine platforms; the relevant data formats can be text files, voice files, video files, and photo files; the data source and search scope can be domestic media and global media. The number of news media information sources 21 connected to the news collection module 11 can be one, as shown in FIG1 , or multiple news media information sources 21 can be connected to obtain news articles in these news media information sources 21 at the same time, and the number can be adjusted according to user needs. In addition, the integration module 14 further aggregates the target files of news articles with a focus subject and in line with the focus topic to generate a list of followers followed by the user and their related information.

Please refer to FIG. 1 and FIG. 3 together. FIG. 3 is a flow chart showing the steps of a method for sorting and analyzing a follow list according to another specific embodiment of the present invention. Similarly, in this specific embodiment, each step of the method of FIG. 3 can be achieved through the system architecture of FIG. 1, so the steps of FIG. 3 are described below by referring to the system architecture in FIG. 1. As shown in FIG. 3, the difference between this specific embodiment and the aforementioned specific embodiment is that the method for sorting and analyzing a follow list of this specific embodiment further includes step S5 executed after step S4: the automatic delivery module 15 sends the follow list of persons to be followed and the related information thereof generated by the destination file generated by the integration module 14 to the outside in the form of a message notification. Please note that the other steps in the method for sorting and analyzing the concerned list of this specific embodiment are roughly the same as the corresponding steps in the aforementioned specific embodiment, so they will not be elaborated here.

In practice, the automatic delivery module 15 can notify and send the list of people to be followed and their related information generated by the target file through email, mobile phone messages and social platforms. In addition, users can set up regular or timely message notifications and transmissions according to their needs, so as to receive the latest and updated list of people to be followed and their related information in a timely or regular manner, without having to spend extra time searching again, thereby reducing manpower and time costs and improving overall efficiency.

Furthermore, as shown in FIG3 , in this specific embodiment, the attention list sorting and analysis method further includes step S11 executed before step S2: the user inputs the attention subject and sets the attention topic through the user interface. In detail, through the user interface (not shown) in the attention list sorting and analysis system, the user can input the attention subject and attention topic to be followed in advance, so that the attention topic comparison module 12 can identify and analyze the attention subject and attention topic according to the content of the news article. Among them, the attention subject can be a person, organization, company, and foundation; the attention topic can be money laundering, fraud, corruption, and other illegal acts. For example, if a user wants to investigate whether a person has a criminal record of money laundering, fraud, and corruption, he or she can enter the name of the person as the subject of attention and set the topics of attention to money laundering, fraud, and corruption. In practice, the types of subjects and topics of attention that users want to pay attention to are not limited to these, and the number of subjects and topics of attention can be set according to user needs.

Please refer to FIG. 1 and FIG. 4 together. FIG. 4 is a flowchart showing a method for sorting and analyzing a list of interests according to another specific embodiment of the present invention. Similarly, in this specific embodiment, each step of the method of FIG. 4 can be achieved through the system architecture of FIG. 1, so the steps of FIG. 4 are explained below by using the system architecture in FIG. 1. As shown in FIG4 , the difference between this embodiment and the aforementioned embodiment is that the method for sorting and analyzing the attention list of this embodiment further includes step S2' executed after step S1: the attention subject comparison module 12 compares the news article with the cross-comparison information source 22 through the subject comparison model 121, and then identifies whether the content of the news article with the attention subject meets the attention topic, and extracts keywords from the news article with the attention subject and meeting the attention topic. Among them, the subject comparison model 121 is generated through a machine learning algorithm, and the cross-comparison information source 22 further includes a business database, a value-added information database, and a website for querying digital footprints. Please note that the other steps in the method for sorting and analyzing the concerned list of this specific embodiment are roughly the same as the corresponding steps in the aforementioned specific embodiment, so they will not be elaborated here.

In this specific embodiment, the subject comparison model 121 learns through a machine learning algorithm to obtain related words with people, places, time, organizations, titles, and behavior orientations from a large number of articles to establish an intelligent model of the correlation between the information content describing the situation and the subject of interest. In practice, the types of cross-comparison information sources 22 used to compare the subject of interest with the news article are not limited to this, and the number of cross-comparison information sources 22 selected can be adjusted according to user needs.

In another specific embodiment, the cross-reference information source 22 can be an authorized global business database, business registration query platform, and other public channel platforms. For those with unclear information, the cross-reference information source 22 is used to provide supplementary data sources, such as the judgment results of the subject of the news article and whether the legal status of the subject company or business has been dissolved or reorganized. However, in practice, the types of cross-reference information sources 22 are not limited to this, and the types and quantities of cross-reference information sources can be adjusted according to user needs.

In addition, another difference between the present embodiment and the above-mentioned embodiment is that the attention list sorting and analysis method of the present embodiment further includes step S31: the duplicate information comparison module 13 receives each news article and corresponding keywords having an attention subject and meeting the attention topic from the attention subject comparison module 12, and compares the received news articles with the target file respectively. When the comparison result between the first news article in the news article and the target file is that the attention subject is repeated and there is no information difference, the duplicate information comparison module 13 deletes the first news article; step S32: the duplicate information comparison module 13 receives each news article having an attention subject and meeting the attention topic from the attention subject comparison module 12. A news article and corresponding keywords, and the received news articles are compared with the target file respectively. When the comparison result of the first news article in the news article with the target file is that the subject of attention is repeated and there is information change, the duplicate information comparison module 13 adds the information change to the target file; and step S33: the duplicate information comparison module 13 receives each news article with a subject of attention and a corresponding keyword that meets the topic of attention from the subject of attention comparison module 12, and compares the received news articles with the target file respectively. When the comparison result of the first news article in the news article with the target file is that the subject of attention is not repeated, the duplicate information comparison module 13 adds the first news article to the target file. Through steps S31~S33, the duplicate information comparison module 13 can update the target file with the comparison result.

In this specific embodiment, the repeated information comparison module 13 may further include an information comparison model 131 to perform the aforementioned comparison process. The information comparison model 131 is generated by a supervised machine learning algorithm. In practice, supervised machine learning uses a training data set for machine learning and generates a supervised machine learning algorithm. The training data set includes labels and features. The label is the prediction target, and the feature is the input data used by the model to predict the label. In this embodiment, the training data set of supervised machine learning uses features such as people, organizations, names, locations, job titles, and time, and classifies them into labels such as duplicate or valid/invalid, and the content of the training data set may include news articles, news article labels, news article classification, and news article analysis. However, in actual applications, it is not limited to this.

As mentioned above, the subject comparison model 121 in the subject comparison module 12 and the information comparison model 131 in the duplicate information comparison module 13 can be generated by a supervised machine learning algorithm. The subject comparison model 121 is used to compare the news article with the cross-comparison information source 22, and the information comparison model 131 is used to compare the received news article with the target file. Among them, the supervised learning algorithm in the subject comparison model 121 and the information comparison model 131 is a DNN-LSTM model formed by combining two different algorithms, deep neural network (DNN) and long short-term memory (LSTM). In other words, the subject comparison model 121" and the information comparison model 131" in this specific embodiment combine the long-term memory ability of LSTM and the deep learning ability of DNN to perform identification and comparison analysis of news articles. However, in actual applications, the supervised learning algorithm used in the attention list sorting and analysis system is not limited to this. Thus, the attention list sorting and analysis system of the present invention introduces artificial intelligence and simplifies the repeated comparison process and reduces the list error rate and repetition rate through the supervised learning algorithm.

In summary, the present invention provides a method and system for sorting and analyzing a follow list, wherein a follow subject comparison module is used to screen out news articles that meet the user's concerns based on the follow subject and the follow topic, and then a duplicate information comparison module is used to compare the screened news articles with the target file, and the target file is updated according to the comparison result. The present invention can quickly obtain useful and required news media data information from the network search results through the follow list sorting and analysis method and system. In addition, by introducing artificial intelligence methods, supervised machine learning is used to reduce the error rate and duplication rate of the follow list and exclude invalid information, thereby reducing the human error rate and the time spent on manual interpretation, thereby improving efficiency and reducing time and labor costs. In addition, users can receive the latest list of people to follow and their related information in a timely or regular manner through the automatic delivery module, so users do not need to spend extra time searching again, thereby reducing manpower and time costs and improving the overall search and retrieval efficiency.

The above detailed description of the preferred specific embodiments is intended to more clearly describe the features and spirit of the present invention, rather than to limit the scope of the present invention by the preferred specific embodiments disclosed above. On the contrary, the purpose is to cover various changes and arrangements with equivalents within the scope of the patent application for the present invention. Therefore, the scope of the patent application for the present invention should be interpreted in the broadest sense based on the above description, so as to cover all possible changes and arrangements with equivalents.

S1~S4: Steps

Claims (10)

A method for sorting and analyzing a follow list is used to analyze and compare a plurality of news articles from at least one news media information source to generate a follow list and related information thereof. The method for sorting and analyzing a follow list comprises the following steps: a news collection module collects the news articles from the at least one news media information source; a follow subject comparison module uses a natural language technology to identify whether the news articles have a follow subject, and then identifies whether the content of the news articles with the follow subject meets a follow topic, and extracts at least one keyword from the news articles with the follow subject and meeting the follow topic; a duplicate information comparison module extracts at least one keyword from the follow subject; The subject matching module receives each of the news articles having the concerned subject and matching the concerned topic and the corresponding at least one keyword, and matches the received news articles with a target file respectively; when the comparison result of a first news article among the news articles with the target file is that the concerned subject is repeated and there is no information difference, the duplicate information matching module deletes the first news article; when the comparison result of the first news article with the target file is that the concerned subject is not repeated, the duplicate information matching module adds the first news article to the target file to update the target file; and the unification module generates the concerned person list and related information according to the target file. The method for sorting and analyzing the concerned list as described in Item 1 of the patent application scope further includes the following steps: an automatic delivery module sends the concerned person list and related information generated by the target file to the outside in the form of a message notification. As described in Item 1 of the patent application scope, a method for sorting and analyzing a concerned list, wherein a user inputs the concerned subject and sets the concerned topic through a user interface. As described in item 1 of the patent application scope, the concerned subject comparison module compares the concerned subject of the news articles with a cross-comparison information source through a subject comparison model, and the subject comparison model is generated through a machine learning algorithm; wherein the cross-comparison information source further includes a business database, a value-added information database, and a website for querying digital footprints. The method for sorting and analyzing the attention list as described in Item 1 of the patent application scope further includes the following steps: when the comparison result between the first news article and the target file is that the attention subject is repeated and there is an information change, the duplicate information comparison module adds the information change to the target file to update the target file. The method for sorting and analyzing the attention list as described in item 1 of the patent application scope further comprises the following steps: the duplicate information comparison module compares the news articles received by an information comparison model with the target file respectively, wherein the information comparison model is generated by a supervised machine learning algorithm. A follow list sorting and analysis system is used to analyze a plurality of news articles from at least one news media information source to generate a follow list and related information thereof. The follow list sorting and analysis system comprises: a news collection module, used to connect to the at least one news media information source to collect the news articles from the at least one news media information source; a follow subject comparison module, coupled to the news collection module to receive the news articles, The subject-of-interest matching module is used to identify whether the news articles have a subject of interest through a natural language technology, and then identify whether the content of the news articles having the subject of interest conforms to a topic of interest, and extract at least one keyword from the news articles having the subject of interest and conforming to the topic of interest; a duplicate information matching module is coupled to the subject-of-interest matching module, and the duplicate information matching module The module is used to receive each of the news articles having the subject of interest and matching the topic of interest and the corresponding at least one keyword from the subject of interest comparison module, and to compare the received news articles with a target file respectively, and to update the target file according to the comparison result between each of the news articles and the target file, wherein when the comparison result between a first news article among the news articles and the target file is that the subject of interest is repeated and there is no When there is information difference, the duplicate information matching module deletes the first news article, and when the comparison result between the first news article and the target file is that the subject of attention is not repeated, the duplicate information matching module adds the first news article to the target file to update the target file; and a unification module, coupled to the duplicate information matching module, the unification module is used to generate the list of attention persons and related information according to the target file. The concerned list sorting and analysis system as described in Item 7 of the patent application scope further includes an automatic delivery module coupled to the integration module, which is used to send the concerned person list and related information generated by the target file to the outside in the form of a message notification. As described in item 7 of the patent application scope, the concerned list sorting and analysis system, wherein the concerned subject comparison module is coupled to a cross-comparison information source, and the concerned subject comparison module compares the concerned subject of the news articles with the cross-comparison information source, and the cross-comparison information source further includes a business database, a value-added information database, and a website for querying digital footprints. As described in item 7 of the patent application scope, the concerned list sorting and analysis system, wherein the concerned subject comparison module includes a subject comparison model, and the duplicate information comparison module includes an information comparison model, wherein the subject comparison model and the information comparison model are generated by a supervised machine learning algorithm, the subject comparison model is used to compare the concerned subject with the news articles and a cross-comparison information source, and the information comparison model is used to compare the received news articles with the target file respectively.

Images