KR20220102181A - Apparatus for Searching Fraudulent News and Driving Method Thereof, and Computer Readable Recording Medium - Google Patents

Abstract

The present invention relates to a negative news search device, a driving method of the device, and a computer-readable recording medium. The negative news search device according to an embodiment of the present invention includes: a storage unit for classifying negative words and positive words by parts of speech and storing a list; and a control unit for filtering the previously collected news articles based on the part-of-speech information related to the parts of speech in the stored list, using the negative and positive words of the filtered news article and the negative and positive words in the (pre-)stored parts of speech list to calculate the negative and positive indices, respectively, and operating the calculated negative and positive indices to calculate the final negative index.

Description

Apparatus for Searching Fraudulent News and Driving Method Thereof, and Computer Readable Recording Medium

The present invention relates to a fraudulent news search device, a method of driving the device, and a computer-readable recording medium, and more particularly, to select and select fraudulent news using news articles to prepare for, for example, a national disaster situation. It relates to a device for detecting fraudulent news, a method of driving the device, and a computer-readable recording medium, which is intended to be used to overcome a disaster situation by using the pointed out items as content.

With the recent development of the Internet, big data analysis is developing. Data mining refers to the process of discovering useful correlations hidden among large amounts of data, extracting actionable information in the future, and using it for decision-making. In other words, it is to discover information as if finding a vein by finding hidden patterns and relationships in data. Here, information discovery is the process of finding useful patterns and relationships by applying advanced statistical analysis and modeling techniques to data, and it can be said to be a core technology of database marketing.

For example, by analyzing data from a sales database in a department store, what products are selling well on Friday mornings and what kind of correlation there are between selling products, etc., are reflected in marketing. Therefore, an essential element of data mining is sufficient data with high reliability. This is because reliable and sufficient data make accurate predictions possible. However, too much data can lower the predictive ability of data mining, so it is necessary to secure meaningful data that can produce optimal results.

Although data mining in Korea is still in its infancy, it is expected to become common soon. The reason is that many data warehouses, which are optimal systems for data mining, have already been established in Korea. In addition, the development of data mining is inevitable because the company's requirements are going towards database marketing, which mainly focuses on customer management.

However, at the national level, particularly in relation to disaster situations, a useful system such as using news articles to reflect policies and the like has not been established, so the establishment of a system for this is urgently required.

Korean Patent Publication No. 10-1741509 (2017.05.24) Korean Patent Publication No. 10-1599675 (2016.02.25) Korean Patent Publication No. 10-2104316 (2020.04.20)

In an embodiment of the present invention, for example, in order to prepare for a national disaster, fraudulent news is selected by using news articles, and the illegal news search device and the device thereof are intended to be utilized to overcome a disaster situation by using the points pointed out in the selected fraudulent news as content. It is an object of the present invention to provide a method of driving and a computer-readable recording medium.

A negative news search apparatus according to an embodiment of the present invention includes a storage unit for storing a list by classifying negative and positive words by part-of-speech, and filtering previously collected news articles based on the stored part-of-speech information related to the part-of-speech. And, by using the words of the filtered news article and the stored negative and positive words in the list of parts of speech, respectively, a negative index and a positive index are calculated, and the calculated negative index and the calculated positive index are calculated, A control unit for calculating the final negative index is included.

The control unit calculates the negative index by further filtering in consideration of the importance in the news article with respect to the words of the filtered news article, and to determine the importance in the news article, TF-IDF (Term Frequency - Inverse) Document Frequency) model can be used.

The control unit calculates the negative index by measuring cosine similarity by comparing the words obtained by the additional filtering with negative words and positive words in the list of parts of speech, and measures the cosine similarity for each negative word Thus, the sum of the representative values having the negative word having the highest value as the representative value can be calculated as the negative index.

The controller may divide the calculated negative index by the body length of the news article to normalize it, and calculate the calculated positive index to the normalized negative index to calculate the final negative index.

In addition, the driving method of the negative news search apparatus according to an embodiment of the present invention includes the steps of: a storage unit classifying negative words and positive words by part-of-speech and storing the list; and the control unit, storing part-of-speech information related to the stored part of speech information Filters news articles collected based on and calculating a final negative index by calculating the index and the calculated positive index.

The calculating step further includes calculating the negative index by further filtering in consideration of the importance in the news article with respect to the words of the filtered news article, and TF to determine the importance in the news article. -IDF model is available.

The calculating step includes calculating the negative index by measuring the cosine similarity by comparing the additionally filtered words with negative and positive words in the list of parts of speech, and measuring the cosine similarity for each negative word. The method may further include calculating, as the negative index, a sum of representative values using the negative word having the highest value as the representative value.

The calculating step includes normalizing the calculated negative index by dividing the length of the body of the news article, and calculating the final negative index by calculating the calculated positive index on the normalized negative index. may include more.

On the other hand, a computer-readable recording medium according to an embodiment of the present invention is a computer-readable recording medium including a program for executing a driving method of a fraudulent news search apparatus, wherein the driving method of the fraudulent news search apparatus includes, and storing the list by classifying positive words by part-of-speech, and filtering news articles collected based on the stored part-of-speech information related to the part-of-speech, negative words and positive words of the filtered news article, and the A negative index and a positive index are respectively calculated using the negative and positive words of the stored list of parts-of-speech, and the final negative index is calculated by calculating the calculated negative index and the calculated positive index.

According to an embodiment of the present invention, when policy is reflected in relation to a national disaster, etc., it is possible to increase the efficiency of the policy by utilizing the points pointed out in the media article.

1 is a view showing a fraudulent news search system according to an embodiment of the present invention;
2 is a block diagram illustrating the detailed structure of the user terminal device of FIG. 1 as a fraudulent news search apparatus according to an embodiment of the present invention;
3 is a block diagram illustrating another detailed structure of the fraudulent news search unit of FIG. 2, and
4 is a flowchart illustrating a driving process of a fraudulent news search apparatus according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

1 is a diagram illustrating a fraudulent news search system according to an embodiment of the present invention.

As shown in FIG. 1 , the fraudulent news search system 90 according to an embodiment of the present invention includes a part or all of a user terminal device 100 , a communication network 110 , and a news service device 120 .

Here, at least one of the user terminal device 100 and the news service device 120 may operate as a fraudulent news search device according to an embodiment of the present invention, and when two devices operate as a fraudulent news search device can operate in cooperation with each other, and "including some or all" means that some components such as the communication network 110 are omitted, so that the user terminal device 100 and the news service device 120 are direct (eg, P2P) It means that a part or all of the components constituting the news service device 120 may be integrated into a network device (eg, a wireless switching device, etc.) of the communication network 110 to perform communication or to be configured. In order to help a sufficient understanding of the invention, it will be described as including everything.

The user terminal device 100 may operate as a fraudulent news search apparatus according to an embodiment of the present invention, and may include a desktop computer, a laptop computer, a tablet PC, a smart phone, a smart TV, and the like. Of course, it may also include a wearable device such as a domestic S company's Galaxy *. The user terminal device 100 according to an embodiment of the present invention may use various devices as long as it is possible to search for fraudulent news, and any device may be used as long as it can execute a program for searching for fraudulent news.

The user terminal device 100 typically detects fraudulent news in media articles such as news in order to prepare for a national disaster situation, etc. Acquire the points pointed out as content and use them to reflect national disaster preparedness, response, or prevention policies. Therefore, the user terminal device 100 according to the embodiment of the present invention may preferably be a computer used by a specific researcher, such as a researcher of a national government institution.

For example, the user terminal device 100 may pre-store a program for performing an operation according to an embodiment of the present invention. Of course, the program may be installed in various forms, such as being installed through a storage medium or downloaded by accessing a server of a program developer through the communication network 110 . In addition, the user terminal device 100 may access the news service device 120 through the communication network 110 , collect media articles related to national disasters, etc., create a folder and store it therein. Of course, the number of news articles stored in the folder may be less than 100 or more as news articles related to national disasters. Of course, the collected news articles may be saved in a file format such as Excel in addition to being saved in a folder. Since a news article according to an embodiment of the present invention may be processed in various forms, it will not be particularly limited to any one form. However, as long as the title of the news article and the content of the main text are available, any form is irrelevant. For example, when URL address information of the relevant media article is used, it may be possible to collect titles and articles by accessing the address information.

Specifically, when the user terminal device 100 executes the program according to the embodiment of the present invention, analysis may be performed by dragging a folder on the monitor screen or by dragging a collected news article file. Accordingly, a negative index is calculated, and an item or headline for each news article and the calculated negative index may be displayed on one side of the area displaying the analysis result on the monitor screen. For example, in the embodiment of the present invention, since there is an interest in news articles having a negative index of 5.0 or higher, only news articles having a negative index of 5.0 or greater are collected on one side and displayed on the screen. Since various forms are possible in relation to UX/UI, it will not be particularly limited to any one form.

When a negative index is displayed through analysis according to an embodiment of the present invention on a computer used by a researcher of a national research institute, the researcher selects a news article based on the negative index, checks the text again, and the corresponding news article Through this, it will be possible to check the pointed out matters related to national disasters, etc. and reflect them in policies.

Of course, the embodiment of the present invention will not be particularly limited to computers used by researchers, etc., and will not be particularly limited to national disasters. In other words, the method of searching for fraudulent news according to an embodiment of the present invention can be freely used in users' smartphones in the form of an application (hereinafter, an app), and furthermore, it can be used to screen for fake news. . For example, if it is produced in the form of an app, it may be possible for the general public to download and use the app through the App Store.

In addition, when the operation according to the embodiment of the present invention is performed in the news service device 120 , the user terminal device 100 accesses the corresponding news service device 120 and lists or files the news articles collected by the user terminal device 100 . It is also possible to use the same service as above (eg, a service to obtain a negative index for a news article) by providing Therefore, the embodiment of the present invention will not be particularly limited to any one form. In other words, in the embodiment of the present invention, the user terminal device 100 collects interesting news articles related to national disasters, analyzes the collected news articles using a pre-stored program, and displays the analysis result in the form of a negative index on the screen. However, it is possible to perform the analysis in the news service device 120 or to perform the operation according to the embodiment of the present invention in conjunction with the news service device 120 . Of course, it may be possible to construct a separate device (eg, a server) that performs a fraudulent news search operation separately from the user terminal device 100 or the news service device 120 .

The communication network 110 includes both wired and wireless communication networks. For example, a wired/wireless Internet network may be used or interlocked as the communication network 110 . Here, the wired network includes an Internet network such as a cable network or a public telephone network (PSTN), and the wireless communication network includes CDMA, WCDMA, GSM, Evolved Packet Core (EPC), Long Term Evolution (LTE), and Wibro networks. meaning to include Of course, the communication network 110 according to the embodiment of the present invention is not limited thereto, and may be used as an access network of a next-generation mobile communication system, for example, a cloud computing network under a cloud computing environment, a 5G network, and the like. For example, when the communication network 110 is a wired communication network, the access point in the communication network 110 can connect to a switching center of a telephone company, etc., but in the case of a wireless communication network, it connects to a SGSN or GGSN (Gateway GPRS Support Node) operated by a communication company to transmit data. Data can be processed by accessing various repeaters such as a BTS (Base Transceiver Station), NodeB, and e-NodeB.

The communication network 110 may include an access point. Here, the access point includes a small base station, such as a femto or pico base station, which is often installed in a building. Here, the femto or pico base station is classified according to the maximum number of video editing devices 130, etc. that can be connected to the small base station. Of course, the access point may include a short-range communication module for performing short-distance communication such as Zigbee and Wi-Fi with the user terminal device 100 . The access point may use TCP/IP or Real-Time Streaming Protocol (RTSP) for wireless communication. Here, short-range communication may be performed in various standards such as Bluetooth, Zigbee, infrared, radio frequency (RF) such as ultra high frequency (UHF) and very high frequency (VHF), and ultra-wideband communication (UWB) in addition to Wi-Fi. Accordingly, the access point may extract the location of the data packet, designate the best communication path for the extracted location, and forward the data packet to the next device, for example, the news service device 120 along the designated communication path. The access point may share several lines in a general network environment, and includes, for example, a router, a repeater, and a repeater.

The news service device 120 may include a media company that provides news, a server of a portal operator such as Yahoo, Never, Da* Daum, and the like. In addition, the news service device 120 may include various devices that provide news. For example, in the fraudulent news search process according to an embodiment of the present invention, it may be possible to search for news articles in the form of videos as well as news articles in text format. For example, the voice of a moving image may be recognized and converted into text, and the textualized content may be used in an embodiment of the present invention. Accordingly, the news service device 120 according to the embodiment of the present invention may include both a news article in a text format and a news article in a video format.

As mentioned above, the news service device 120 may perform an operation of calculating a negative index based on a news article or news article-related information (eg, URL information, etc.) provided by the user terminal device 100, , in this case, such analysis may be possible in the news service device 120, but it may also be constructed to operate as a fraudulent news search device constituting a separate device. However, in the embodiment of the present invention, the present invention is performed in the user terminal device 100 such as a computer used by the researcher as much as it is intended to search for fraudulent news related to national disasters, and only the government-affiliated National Disaster and Safety Research Institute, etc. Since it is preferable to perform the operation according to the example, further description is omitted.

For example, the user of the user terminal device 100 can check the search list related to news articles on the screen by accessing the portal service, that is, accessing the news service device 120 and inputting a keyword of interest to the user in the search box. have. In addition, the news articles displayed on the screen can be collected one by one, but if there is a function to collectively select and collect news articles retrieved from the news service device 120 (eg, select all, etc.), 100 You can collect or more news articles and save them in a folder or collect them in the form of an Excel file. Of course, even in the case of the Excel file, it is preferable to include the title and body of the news article here.

FIG. 2 is a block diagram illustrating a detailed structure of the user terminal device of FIG. 1 as a fraudulent news search apparatus according to an embodiment of the present invention.

As shown in FIG. 2 , the user terminal device 100 according to an embodiment of the present invention is a computer such as a desktop computer, a laptop computer, and furthermore a tablet PC, and includes a communication interface unit 200 , a control unit 210 , and negative news. Some or all of the search unit 220 and the storage unit 230 are included, and may further include a user interface unit for receiving a user command, a display unit for displaying a UI screen, and the like.

Here, “including some or all” means that some components such as the storage unit 230 are omitted, or some components such as the negative news search unit 220 are other components such as the control unit 210 . It is described as including all in order to help a sufficient understanding of the invention as meaning that it can be configured by being integrated in the present invention.

The communication interface unit 200 connects to the server of the program developer via the communication network 110 of FIG. 1 to download and store the program according to the embodiment of the present invention, or to search the illegal news search unit 220 for the program in the form of EEPROM. ), or it can be operated to store a program through a storage medium such as CD or USB.

In addition, the communication interface unit 200 may access the news service device 120 according to a request of a user using the user terminal device 100 to receive, for example, a press article related to a national disaster and provide it to the control unit 210 . have. Of course, when there is an image in the process of collecting news articles, the communication interface unit 200 may perform an operation such as decoding to process it. In more detail, the communication interface unit 200 may perform various operations such as modulation/demodulation, muxing/demuxing, encoding/decoding, and scaling for converting a resolution in order to communicate with the news service device 120 , which are those skilled in the art. Since it is self-explanatory, further explanation is omitted.

The control unit 210 is in charge of overall control operations of the communication interface unit 200, the fraudulent news search unit 220, and the storage unit 230 constituting the user terminal device 100 of FIG. For example, the control unit 210 may store a program for searching for fraudulent news according to an embodiment of the present invention in the fraudulent news search unit 220 . And, when there is a user's request, the program may be executed by controlling the fraudulent news search unit 220 .

In addition, when the news articles related to national disasters collected by the user are provided by the communication interface unit 200, the control unit 210 stores them in the storage unit 230 and then calls them and provides them to the fraudulent news search unit 220 to provide the negative index etc. can be produced. In addition, the control unit 210 may perform various operations with the fraudulent news search unit 220 in order to search for fraudulent news of a news article according to an embodiment of the present invention.

The fraudulent news search unit 220 performs an overall operation related to the search for fraudulent news according to the request of the control unit 210 . For example, the fraudulent news search unit 220 calculates a negative index for each news article with respect to the news articles collected by the user in relation to a specific topic (eg, a national disaster, etc.) make it possible to check To this end, the negative news search unit 220 generates a list of negative words or positive words related to news by part-of-speech, and stores information of negative words and positive words, that is, data by part-of-speech. And, when content related to the collected news article is received, all words not in the list (or dictionary) are deleted using pre-stored negative and positive words. In an embodiment of the present invention, this may be called filtering.

Of course, in the embodiment of the present invention, the original text of the news article is classified by part-of-speech through morpheme analysis, and the words are selected based on the part-of-speech information of the words stored in the list, and for this purpose, the part-of-speech information of the list can be used. , since a UI screen for separately inputting the part-of-speech information on the screen may be provided and the part-of-speech information input through this may be used, the embodiment of the present invention will not be particularly limited to any one form. The negative news search unit 220 analyzes how important selected words, for example, negative words or positive words, in one news article through a model such as TF-IDF, and also words having a low value. can be deleted. Here, TF-IDF is a weight used in information retrieval and text mining, and is a statistical number indicating how important a word is in a specific document when there is a document group consisting of several documents (eg, reporter, news, etc.). The reason for using TF-IDF is to increase the accuracy when calculating the negative index in the filtered news article. For example, the TF-IDF value is expected to be low because words such as 'reporter' or 'news' are included in all articles and do not act as important keywords when searching for negative news. In this case, these words are excluded. Therefore, this step is a step that needs to be done before calculating the negative index.

The negative news search unit 220 performs an operation for calculating a negative index through cosine similarity measurement with respect to the selected, that is, filtered negative or positive words. For example, assuming that the number of selected negative words is n, the cosine similarity of each negative word with negative words stored in advance for each negative word is measured, and the word with the highest value is used as a representative value. In this way, the sum of representative values repeated n times can be calculated as the negative index of news. For example, if the number of negative noun words stored in the dictionary is 1000 and the number of selected noun negative words is 50, each of the 50 negative words is compared with words corresponding to the same part-of-speech in the dictionary, and the cue sign similarity is measured to obtain a representative value. In this process, the representative values are summed to calculate the negative index for one news article. More details will be dealt with later.

The negative index is calculated by measuring the cosine similarity of the remaining words except for the words with the low TF-IDF value by comparing them with the negative words in the list of parts of speech, and measuring the cosine similarity of each negative word to determine the highest value The sum of the representative values using the word as the representative value is calculated as a negative index. Of course, in the case of a positive word, it can be made in the form of calculating a positive index by comparing it with the positive word in the list of positive words by part-of-speech.

In addition, the negative news search unit 220 may reflect the length of the body of the news article in order to increase the accuracy of the calculated negative index. In other words, the longer the text length, the more words there are, so it is necessary to normalize it. Accordingly, the negative news search unit 220 normalizes the negative index by dividing it by the body length. Thereafter, the negative news search unit 220 subtracts the positive index calculated in the same manner as above from the negative index, that is, calculates the calculated negative index and positive index, respectively, to calculate the final negative index. For example, since a negative index is calculated for each news article, when 100 news articles are collected and analyzed, each negative index for 100 articles may be calculated.

The storage unit 230 may store content for news articles collected by the user, that is, the title or body of the article, and thus may be stored in the form of a list. In addition, negative words and positive words may be stored in the storage unit 230 before the execution of the program, but the data for these negative words and positive words are stored in the program registry of the negative news search unit 220 . Since it may be, the embodiment of the present invention will not be particularly limited to any one form.

On the other hand, as another embodiment of the present invention, the control unit 210 may include a CPU and a memory, and may be formed as a single chip. The CPU includes a control circuit, an arithmetic unit (ALU), an instruction interpreter and a registry, and the memory may include a RAM. The control circuit may perform a control operation, the operation unit may perform an operation operation of binary bit information, and the instruction interpretation unit may perform an operation of converting a high-level language into machine language and/or converting a machine language into a high-level language, including an interpreter or compiler. , the registry may be involved in software data storage. According to the above configuration, for example, at the beginning of the operation of the user terminal device 100 of FIG. 1 , the program stored in the fraudulent news search unit 220 is copied, loaded into a memory, that is, a RAM, and then data is calculated by executing it. The processing speed can be increased quickly.

3 is a block diagram illustrating another detailed structure of the fraudulent news search unit of FIG. 2 .

As shown in FIG. 3 , the fraudulent news search unit 220 of FIG. 2 may have the structure of the fraudulent news search unit 220 ′ as in FIG. 3 , the word list unit 300 and the original news reduction unit 310, including some or all of the index calculating unit 320 and the final index determining unit 330, where "includes some or all" has the same meaning as above.

3, the word list unit 300, the news original text reduction unit 310, the index calculation unit 320, and the final index determination unit 330 are hardware (H/W), software (S/W), or a combination thereof. Since it may be configured by , it will not be particularly limited to any one form in the embodiment of the present invention.

The word list unit 300 classifies and stores negative or positive words by parts of speech. For example, a total of about 2,000 words can be used for negative words, such as makmak, laziness, and back book. It can be classified into 100 nouns (NN), 200 roots (XR), and 200 general adverbs (MAG). In addition, as the affirmative word, words such as provision, promotion, reinforcement, etc. may be used.

The news original text reduction unit 310 tags one piece of news through morphological analysis of the title and text, and deletes unrelated words while leaving only the part-of-speech in the negative word (or positive word) list. For this purpose, for example, part-of-speech information stored in a DB or memory may be used, or part-of-speech information input through a separate UI screen may be used. Also, words with low importance are deleted through TF-IDF. TF-IDF is a numerical value indicating how important a specific word selected by part-of-speech is considered in one news article. As described above, the news original text reduction unit 310 reduces the text amount of the original news text by deleting or filtering unwanted words from all one news article.

In addition, the index calculator 320 may calculate the positive index as well as the negative index. For example, it is assumed that there are n words (w) remaining through the news original text reduction unit 310 . Therefore, the cosine similarity between one remaining word and a negative word among n is measured, and the highest value becomes the representative value of the remaining one above. However, it measures the cosine similarity between negative words on the list, such as the part-of-speech of one remaining word.

The index calculation unit 320 further sets the sum of the values repeated n times as the negative index of the news. The calculation of negative indices for n pieces can be expressed as <Equation 1>.

For example, it is assumed that there are 100 words selected by the news original text reduction unit 310 . At this time, there are 70 nouns, 20 roots, and 10 general adverbs. In this case, the largest value is obtained by measuring the cosine similarity between one noun and 1000 nouns in the negative word, and repeat for 70 negative words. The similarity is measured for each of the 70 nouns. This is because there are 70 nouns among the remaining words above. Also, measure the cosine similarity between one root and 200 roots in the negative word to find the largest value, and repeat 20 times. This is because there are 20 roots of the remaining words. Furthermore, measure the cosine similarity between one general adverb and 200 general adverbs of the negative word to find the largest value, and repeat 10 times. This is because there are 10 general adverbs among the remaining words. The sum of 100 representative values obtained in this way becomes the negative index of the news.

In addition, the final index determination unit (or the final index calculation unit) 330 reflects the text length for the negative index calculated by the index calculation unit 320, and reflects the index of the positive word. In other words, the longer the text length, the more words there are, so normalization is necessary. Therefore, the negative exponent output from the exponent calculation unit 320 is divided by the length of the text and normalized. In addition, the final negative index can be calculated by applying the positive index obtained by calculating the positive index by performing the same procedure as the above negative index calculation process with positive words (eg, preparation, promotion, reinforcement, etc.). That is, the final negative index can be obtained by calculating, for example, subtracting the positive index from the negative index calculated by the above index calculation unit 320 .

In the embodiment of the present invention, the final negative index is calculated for each news article in the same manner as above, and the finally calculated negative index of each news article is graphically processed in various forms so that the user can display the screen of the user terminal device 100 . can be created for viewing. Accordingly, for example, as a fraudulent news search apparatus according to an embodiment of the present invention, the user terminal device 100 of FIG. 1 may further include a graphic processing unit for displaying the finally calculated negative index on one side of each news article.

4 is a flowchart illustrating a driving process of a fraudulent news search apparatus according to an embodiment of the present invention.

For convenience of explanation, referring to FIG. 4 together with FIG. 1, at least one of the user terminal device 100 and the news service device 120 of FIG. As the search apparatuses 100 and 120, or separately, a negative news search apparatus may be constructed, but a list of parts of speech for negative words and positive words is stored (S400). It can also be seen that a dictionary of words for each part of speech is created.

In addition, the fraudulent news search apparatus 100, 120 according to an embodiment of the present invention filters the collected news articles by using the part-of-speech information related to the part-of-speech in the stored list, and through this, for example, the amount of text in the original text. A negative index and a positive index are respectively calculated using the negative and positive words of the reduced and filtered news article, and the calculated negative index and the calculated positive index are calculated to calculate the final negative index (S410).

For example, the content of the original article, ie, words, is reduced by filtering one news article, ie, selecting it based on pre-stored part-of-speech information. The number of words is reduced. In this case, the fraudulent news search apparatuses 100 and 120 may additionally select words in consideration of the importance of each word in the news article. In addition, the negative news search apparatuses 100 and 120 measure the cosine similarity of selected words (eg, negative words, positive words) for each part of speech. The objects to be compared to measure the similarity are the above pre-stored words. For example, assuming that 1000 words are stored for a noun and 70 selected words, each representative value for 70 words can be obtained by measuring the similarity between each word among 70 and 1000 words, , to calculate the negative index of the news article by the sum of them. However, the comparison target of cosine similarity must be in the same part-of-speech category.

In addition, when the negative index is calculated, the negative news search apparatuses 100 and 120 finally calculate the negative index in consideration of the length of the body of the news article and the positive index. The final negative index for a specific news article can be calculated by dividing the calculated negative index by the text length, normalizing it, and subtracting the positive index.

In addition to the above, the fraudulent news search apparatuses 100 and 120 according to an embodiment of the present invention can perform various operations, and since other detailed information has been sufficiently described above, it will be replaced with the contents.

On the other hand, even though it has been described that all components constituting the embodiment of the present invention are combined or operated in combination, the present invention is not necessarily limited to this embodiment. That is, within the scope of the object of the present invention, all the components may operate by selectively combining one or more. In addition, all of the components may be implemented as one independent hardware, but some or all of the components are selectively combined to perform some or all of the functions of one or a plurality of pieces of hardware. It may be implemented as a computer program having Codes and code segments constituting the computer program can be easily deduced by those skilled in the art of the present invention. Such a computer program is stored in a computer-readable non-transitory computer readable media, read and executed by the computer, thereby implementing an embodiment of the present invention.

Here, the non-transitory readable recording medium refers to a medium that stores data semi-permanently and can be read by a device, not a medium that stores data for a short moment, such as a register, cache, memory, etc. . Specifically, the above-described programs may be provided by being stored in a non-transitory readable recording medium such as a CD, DVD, hard disk, Blu-ray disk, USB, memory card, ROM, and the like.

In the above, preferred embodiments of the present invention have been illustrated and described, but the present invention is not limited to the specific embodiments described above, and it is common in the technical field to which the present invention pertains without departing from the gist of the present invention as claimed in the claims. Various modifications may be made by those having the knowledge of, of course, and these modifications should not be individually understood from the technical spirit or perspective of the present invention.

100, 120: fraudulent news search device 100: user terminal device
110: communication network 120: news service device
200: communication interface unit 210: control unit
220, 220': negative news search unit 230: storage unit
300: word list section 310: news original text reduction section
320: index calculation unit 330: final index determination unit

Claims (9)

a storage unit for storing a list by classifying negative and positive words by part-of-speech; and
Filtering news articles collected based on the part-of-speech information related to the part-of-speech of the stored list, and using the words of the filtered news article and the negative and positive words of the stored list of parts-of-speech, negative index and A control unit that calculates each positive index and calculates the calculated negative index and the calculated positive index to calculate the final negative index;
Including fraudulent news search device. According to claim 1,
The control unit calculates the negative index by further filtering in consideration of the importance in the news article with respect to the words of the filtered news article, and to determine the importance in the news article, TF-IDF (Term Frequency - Inverse) A device for detecting fraudulent news using the Document Frequency) model. 3. The method of claim 2,
The control unit calculates the negative index by measuring cosine similarity by comparing the words obtained by the additional filtering with negative and positive words in the list of parts of speech, and measuring the cosine similarity for each negative word. A fraudulent news search device for calculating the sum of representative values with the negative word having the highest value as the representative value as the negative index. 4. The method of claim 3,
The control unit normalizes the calculated negative index by dividing the length of the body of the news article, and calculates the final negative index by calculating the calculated positive index on the normalized negative index. storing, by the storage unit, a list of negative words and positive words by part-of-speech; and
The control unit filters the news articles collected based on the part-of-speech information related to the part-of-speech in the stored list, and uses the words of the filtered news article and negative and positive words in the stored list of parts-of-speech. Calculating a negative index and a positive index, respectively, calculating the final negative index by calculating the calculated negative index and the calculated positive index;
A driving method of a fraudulent news search device, including. 6. The method of claim 5,
The calculating step is
Further comprising; further filtering the words of the filtered news article in consideration of the importance in the news article to calculate the negative index by further filtering;
A driving method of a fraudulent news search apparatus using a TF-IDF model to determine the importance in the news article. 7. The method of claim 6,
The calculating step is
calculating the negative index by measuring the cosine similarity by comparing the words obtained by the additional filtering with negative and positive words in the list for each part-of-speech; and
Measuring the cosine similarity for each negative word, calculating the sum of representative values using the negative word having the highest value as the representative value as the negative index; 8. The method of claim 7,
The calculating step is
normalizing the calculated negative index by dividing it by the length of the body of the news article; and
calculating the final negative index by calculating the calculated positive index on the normalized negative index;
Driving method of the fraudulent news search device further comprising. As a computer-readable recording medium including a program for executing a driving method of a fraudulent news search apparatus,
The driving method of the fraudulent news search device,
storing a list by classifying negative words and positive words by part-of-speech; and
Filtering news articles collected based on the part-of-speech information related to the part-of-speech in the stored list, and using the words of the filtered news article and the negative and positive words of the stored list of parts-of-speech, negative index and Calculating each positive index, calculating the final negative index by calculating the calculated negative index and the calculated positive index;
A computer-readable recording medium that executes.

Images