KR20200072724A - An apparatus for detecting spam news with spam phrases, a method thereof and computer recordable medium storing program to perform the method - Google Patents

Abstract

The present invention relates to an apparatus for detecting spam news having a spam phrase, a method therefor, and a computer-readable recording medium in which a program for performing the method is recorded. The present invention provides an apparatus for detecting spam news, a method therefor, and a computer-readable recording medium in which a program for performing the method is recorded, wherein the apparatus for detecting spam news includes: a category classification unit for deriving a probability at which a plurality of paragraphs included in the news belong to a plurality of categories; a correlation calculation unit for calculating a correlation between each paragraph and another paragraph through correlation analysis based on the probability of belonging to the plurality of categories; and a spam news discrimination unit for specifying a sentence having the lowest correlation among the plurality of paragraphs as a suspected spam paragraph and determining whether the news is genuine or fake according to the correlation of the suspected spam paragraph.

Description

An apparatus for detecting spam news with spam phrases, a method thereof and computer recordable medium storing program in which a method for detecting spam news containing spam phrases, a method therefor and a program for performing the method are recorded to perform the method}

The present invention relates to a technology for detecting spam news, and more particularly, to an apparatus for detecting spam news containing spam phrases, a method therefor, and a computer-readable recording medium in which a program performing the method is recorded.

There are a number of news on the Internet that insert sentences or paragraphs that are irrelevant to the content of the text to make you read spam of a specific content.

[Advanced technical literature]

[Patent Document] Registered Korean Registered Patent No. 1864439 on June 01, 2018 (Name: Fake News Discrimination System with Graphical User Interface Screen Window for Posts that Can Be Discriminated by Fake News)

The present invention provides an apparatus for detecting spam news in which spam phrases irrespective of the text content of an article are inserted, a method therefor, and a computer-readable recording medium in which a program for performing the method is recorded.

An apparatus for detecting spam news according to a preferred embodiment of the present invention for achieving the above object includes a category classification unit for deriving a probability that a plurality of paragraphs included in the news belong to a plurality of categories, and the plurality of categories Correlation calculation unit that calculates the correlation between each paragraph and other paragraphs through correlation analysis based on the probability of belonging to, and the sentence with the lowest value of the correlation among the plurality of paragraphs is a suspected spam paragraph And a spam news discrimination unit for determining whether the news is real or fake according to the correlation degree of the spam interrogation panel.

The spam news discrimination unit derives a fake news probability model that derives a probability of being a fake news based on a distribution of correlation between fake news and a real news, and a probability model of a real news that derives a probability of being a real news, and interrogates the spam suspect By substituting the correlations of the fake news probability model and the real news probability model to calculate the probability that the news is real news and the probability that the news is fake news, according to the probability that it is real news and the probability that it is fake news It is characterized by determining whether the news is real or fake.

The spam news discrimination unit is characterized in that if the correlation of the suspected spam interceptor is less than a predetermined value, the news is faked.

The correlation calculating unit is characterized in that it calculates a correlation of an average of the probability of each category of the paragraphs other than the paragraph with respect to the probability of each category of each paragraph.

The category classification unit is characterized in that when a paragraph is inputted, learning to output a probability that the inputted paragraph corresponds to each of a plurality of categories is numerically.

Method for detecting spam news according to a preferred embodiment of the present invention for achieving the above object is a category classification unit deriving the probability that a plurality of paragraphs included in the news belong to a plurality of categories, the correlation calculation Calculating a correlation between each paragraph and other paragraphs through correlation analysis based on a probability that the unit belongs to the plurality of categories, and the spam news discrimination unit has the highest value of the correlation among the plurality of paragraphs. And specifying a low sentence as a suspected spam questionnaire and determining whether the news is real or fake according to the correlation degree of the suspected spam questionnaire.

The step of determining whether the news is real or fake is based on the probability model of the spam news and the probability model of the fake news that derives the probability of the fake news based on the distribution of the correlation between the fake news and the real news. Deriving a real news probability model to derive, and the spam news discrimination unit substitutes the correlation of the spam interrogation panel into the fake news probability model and the real news probability model to determine whether the news is real news and the news And calculating a probability of being fake news, and determining whether the news is real or fake according to the probability that the spam news discrimination unit is the real news and the probability that the fake news is false.

The step of determining whether the news is real or fake is characterized in that if the correlation between the spam suspects and the spam interrogation panel is less than a predetermined value, the news is fake.

The step of calculating the correlation is characterized by calculating a correlation of an average of the probability of each category of other paragraphs other than the paragraph with respect to the probability of each category of each paragraph.

The step of deriving the probability, if the paragraph is input, characterized in that it further comprises the step of learning the category classifier to output the probability that the input paragraph corresponds to each of a plurality of categories as a number.

According to another aspect of the present invention, it is possible to provide a computer-readable recording medium in which a program for performing a method for detecting spam news according to a preferred embodiment of the present invention for achieving the above object.

According to the present invention, spam news can be detected in advance, thereby saving time wasted reading spam news. This can provide a new user experience (UX) to the user.

1 is a block diagram illustrating the configuration of an apparatus for detecting spam news according to an embodiment of the present invention.
2 is a block diagram for explaining the configuration of a control unit of a spam detection device according to an embodiment of the present invention.
3 is a flowchart illustrating a method for detecting spam news according to an embodiment of the present invention.
4 to 6 are diagrams for explaining a method for detecting spam news according to an embodiment of the present invention.

Prior to the detailed description of the present invention, the terms or words used in the present specification and claims described below should not be interpreted as being limited to the ordinary or lexical meanings, and the inventors of their own inventions in the best way. In order to explain, it should be interpreted as meanings and concepts consistent with the technical spirit of the present invention based on the principle that it can be properly defined as a concept of terms. Therefore, the embodiments shown in the embodiments and the drawings described in this specification are only the most preferred embodiments of the present invention, and do not represent all of the technical spirit of the present invention, and various equivalents can be substituted at the time of application. It should be understood that there may be water and variations.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. At this time, it should be noted that the same components are denoted by the same reference numerals in the accompanying drawings. In addition, detailed descriptions of well-known functions and configurations that may obscure the subject matter of the present invention will be omitted. For the same reason, in the accompanying drawings, some components are exaggerated, omitted, or schematically illustrated, and the size of each component does not entirely reflect the actual size.

First, a configuration of a device for detecting spam news according to an embodiment of the present invention will be described. 1 is a block diagram illustrating the configuration of an apparatus for detecting spam news according to an embodiment of the present invention. 2 is a block diagram for explaining the configuration of a control unit of a spam detection device according to an embodiment of the present invention.

First, referring to FIG. 1, a device for detecting spam news (hereinafter abbreviated as'spam detection device') includes a communication unit 110, an input unit 120, a display unit 130, a storage unit 140, and It includes a control unit 200.

The communication unit 110 is for communicating with other devices through a network. The communication unit 110 may include a radio frequency (RF) transmitter (Tx) for up-converting and amplifying the frequency of the transmitted signal, and an RF receiver (Rx) for amplifying the received signal with low-noise and down-converting the frequency. In addition, the communication unit 110 may include a modem for modulating the transmitted signal and demodulating the received signal. For example, the communication unit 110 may download news by accessing a server (not shown) that provides Internet news under the control of the control unit 200.

The input unit 120 receives a key operation of a user for controlling the spam detection device 100, generates an input signal, and transmits the input signal to the control unit 200. The input unit 120 may include various types of keys for controlling the spam detection device 100. When the display unit 130 is made of a touch screen, the input unit 120 may perform functions of various types of keys on the display unit 130, and if all functions can be performed only with the touch screen, the input unit 120 may be omitted. It might be.

The display unit 130 visually provides a menu of the spam detection device 100, input data, function setting information, and various other information to the user. The display 130 performs a function of outputting a screen of a boot screen, a standby screen, a menu screen, and the like of the spam detection device 100. The display unit 130 may be formed of a liquid crystal display (LCD), organic light emitting diodes (OLED), active matrix organic light emitting diodes (AMOLED), or the like. Meanwhile, the display unit 130 may be implemented as a touch screen. In this case, the display unit 130 includes a touch sensor. The touch sensor detects the user's touch input. The touch sensor may include a touch sensing sensor such as a capacitive overlay, a pressure type, a resistive overlay, an infrared beam, or a pressure sensor. . In addition to the sensors, all kinds of sensor devices capable of sensing contact or pressure of an object can be used as the touch sensor of the present invention. The touch sensor detects a user's touch input, generates a detection signal, and transmits it to the controller 200. In particular, when the display unit 130 is made of a touch screen, some or all of the functions of the input unit 120 may be performed through the display unit 130.

The storage unit 140 stores programs and data necessary for the operation of the spam detection device 100. In particular, the storage unit 140 stores a synonym and antonym dictionary, a news database including a plurality of Internet news, and the like. Various types of data stored in the storage 140 may be deleted, changed, or added according to a user's manipulation.

The control unit 200 may control the overall operation of the spam detection device 100 and the signal flow between the internal blocks 110 to 140 of the spam detection device 100, and may perform a data processing function for processing data. The control unit 200 may be a central processing unit (CPU), a graphic processing unit (GPU), a digital signal processor (DSP), or the like.

Referring to FIG. 2, the control unit 200 includes a category classification unit 210, a correlation calculation unit 220, and a spam news discrimination unit 230.

The category classification unit 210 is for deriving a probability that a plurality of paragraphs included in the news belong to a plurality of categories. When the paragraph is input to the artificial neural network, the category classification unit 210 is learned to output the probability that the input paragraph corresponds to each of a plurality of categories as a numerical value. For example, if a paragraph such as "The KOSPI index has fallen again a day and has given 2400 lines" is entered, the probability that the entered paragraph belongs to each of the categories {social, political, economic, international, entertainment, sports} It is trained to derive {0.42, 0.12, 0.97, 0.24, 0.01, 0.01}.

The correlation calculation unit 220 is for calculating correlation between each paragraph and another paragraph through correlation analysis based on a probability of belonging to a plurality of categories derived by the category classification unit 210.

The spam news discrimination unit 230 identifies a sentence having the lowest correlation number among a plurality of paragraphs as a suspected spam questionnaire, and determines whether the news is genuine or fake according to the correlation of the suspected spam questionnaire.

The operation of the control unit 200 including the category classification unit 210, the correlation calculation unit 220, and the spam news discrimination unit 230 will be described in more detail below.

Then, a method for detecting spam news according to an embodiment of the present invention will be described. 3 is a flowchart illustrating a method for detecting spam news according to an embodiment of the present invention. 4 to 6 are diagrams for explaining a method for detecting spam news according to an embodiment of the present invention.

Prior to the process of detecting the spam news, the category classification unit 210 is trained to output, as a numerical value, a probability that the input paragraph corresponds to each of a plurality of categories when a paragraph is input through a deep learning technique. Is assumed. For example, when the category classification unit 210 inputs a paragraph such as "The KOSPI index fell again and again in a day and gave 2400 lines," the input paragraph is a category {social, political, economic, international, entertainment , Sports} It is learned to derive the probability {0.42, 0.12, 0.97, 0.24, 0.01, 0.01} to belong to each.

Referring to FIG. 3, the control unit 200 of the spam detection apparatus 100 accesses a web server operating a web page providing a news article through the communication unit 110 in step S110 to access news contained in the web page. You can download it.

The category classification unit 210 of the control unit 200 derives a probability that a plurality of paragraphs included in the news belong to the plurality of categories in step S120. For example, as illustrated in FIG. 4, assuming that four paragraphs 41 exist in the news, the category classification unit 210 may include four paragraphs, for example, social, political, and economic, as shown in Table 1 below. The probability 42 is derived from multiple categories such as international, entertainment, and sports.

[Table 1]

Next, the correlation calculating unit 220 of the control unit 200 determines correlation between each paragraph and other paragraphs through correlation analysis based on the probability that a plurality of paragraphs belong to a plurality of categories in step S130. Calculate. At this time, the correlation calculating unit 220 calculates a correlation of an average of the probability of each category of the other paragraphs other than the paragraph with respect to the probability of each category.

For example, the correlation calculator 220 may calculate a correlation between paragraph 1 and another paragraph, that is, paragraphs 2, 3, and 4 through correlation analysis based on the probability values in Table 1. Specifically, the correlation diagram calculating unit 220 obtains the average of the probability for each category of paragraphs 2, 3, and 4. For example, the average of the probability for each category of paragraphs 2, 3, and 4 is {0.30, 0.41, 0.04, 0.65, 0.32, 0.01}. And the correlation calculator 220 averages the probability of each category in paragraph 1 {0.56, 0.64, 0.21, 0.98, 0.02, 0.01} and the probability of each category in paragraphs 2, 3, and 4, {0.30, 0.41, 0.04, 0.65, The correlation of 0.32, 0.01} is calculated.

Accordingly, the correlation diagram calculating unit 220 may calculate the correlation diagram 43 as shown in Table 2 below.

[Table 2]

In step S140, the spam news discrimination unit 230 identifies the sentence having the lowest correlation number 42 among the plurality of paragraphs as a suspected spam questionnaire. For example, according to Table 2, the spam news discrimination unit 230 specifies paragraph 3 as a suspected spam paragraph because paragraph 3 has the lowest correlation.

Subsequently, in step S150, the spam news discrimination unit 230 determines whether the news is real or fake according to the correlation degree of the spam interrogation panel.

According to an embodiment, as shown in FIGS. 5 and 6, the spam news discrimination unit 230 is based on a correlation distribution (A1) for pre-stored fake news and a correlation distribution (B1) for real news. We derive a fake news probability model (A2) that derives the probability of fake news, and a real news probability model (B2) that derives the probability of true news. And the spam news discrimination unit 230 substitutes the correlations of the suspected spam suspects into the fake news probability model (A2) and the real news probability model (B2) to calculate the probability that the news is real news and the probability that the news is fake news. . Subsequently, the spam news discrimination unit 230 determines whether the news is real or fake according to a case where a value of a greater probability value between a probability of being genuine news and a probability of fake news is valued.

According to another embodiment, the spam news discrimination unit 230 determines that the news is fake if the correlation of the suspected spam interrogation group is less than a predetermined value.

For example, it is assumed that the news includes paragraphs 1 to 4, and the correlation calculator 220 calculates correlations with paragraphs 2, 3, and 4, as shown in Table 3 below. In addition, it is assumed that a predetermined number of correlations, which is a criterion for judging spam suspects, is "-0.17". For example, as shown in FIG. 6, "-0.17" may be set as a reference value for determining the suspected group of spam based on the fake news probability model A2 and the real news probability model B2.

[Table 3]

In this case, the spam news discrimination unit 230 determines that the news containing the paragraph 3 is fake news because the correlation of the paragraph 3 is “-0.2” smaller than “-0.17”.

On the other hand, the method according to the embodiment of the present invention described above is implemented in a program readable form through various computer means may be recorded on a computer-readable recording medium. Here, the recording medium may include program instructions, data files, data structures, or the like alone or in combination. The program instructions recorded on the recording medium may be specially designed and configured for the present invention or may be known and usable by those skilled in computer software. For example, the recording medium includes magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs, DVDs, and magnetic-optical media such as floptical disks ( magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions may include high-level language wires that can be executed by a computer using an interpreter, as well as machine language wires such as those produced by a compiler. Such a hardware device can be configured to operate as one or more software modules to perform the operation of the present invention, and vice versa.

Although the present invention has been described using some preferred embodiments, these embodiments are illustrative and not limiting. As described above, those skilled in the art to which the present invention pertains will understand that various changes and modifications can be made according to the theory of equality without departing from the spirit of the present invention and the scope of the rights set forth in the appended claims.

100: Spam detection device
110: communication department
120: input unit
130: display unit
140: storage
200: control unit
210: Category classification
220: Correlation calculation unit
230: Spam News Division

Claims (11)

In the device for detecting spam news,
A category classification unit deriving a probability that a plurality of paragraphs included in the news belong to a plurality of categories;
A correlation calculator calculating a correlation between each paragraph and another paragraph through correlation analysis based on a probability of belonging to the plurality of categories; And
Includes a spam news discrimination unit that identifies the sentence having the lowest correlation value among the plurality of paragraphs as a suspected spam questionnaire and determines whether the news is real or fake according to the correlation degree of the suspected spam questionnaire. Characterized by
Device for detecting spam news. According to claim 1,
The spam news discrimination department
Based on the distribution of correlation between fake news and real news, we derive a fake news probability model that derives the probability of fake news, and a real news probability model that derives the probability of true news,
The probability of the news is real news and the probability that the news is fake news are calculated by substituting the correlations of the spam interrogation groups into the fake news probability model and the real news probability model,
And determining whether the news is real or fake based on the probability of the real news and the probability of the fake news.
Device for detecting spam news. According to claim 1,
The spam news discrimination department
Based on the distribution of correlation between fake news and real news, we derive a fake news probability model that derives the probability of fake news, and a real news probability model that derives the probability of true news,
If the correlation of the suspected spam questionnaire is less than a reference value set based on the fake news probability model and the real news probability model, the news is judged as fake.
Device for detecting spam news. According to claim 1,
The correlation calculation unit
And calculating a correlation of an average probability of each category of the paragraphs other than the above paragraphs with respect to the probability of each paragraph by category.
Device for detecting spam news. According to claim 1,
The category sorter
When the paragraph is input, characterized in that learning to output the probability that the input paragraph corresponds to each of a plurality of categories as a number
Device for detecting spam news. In the method for detecting spam news,
Deriving a probability that the plurality of paragraphs included in the category classification unit belong to the plurality of categories;
Calculating a correlation between each paragraph and another paragraph through correlation analysis based on a probability that the correlation diagram calculation unit belongs to the plurality of categories; And
The spam news discrimination unit identifies a sentence having the lowest correlation number among the plurality of paragraphs as a suspected spam questionnaire, and determines whether the news is real or fake according to the correlational degree of the suspected spam questionnaire; Characterized by including
Method for detecting spam news. The method of claim 6,
The step of determining whether the news is real or fake is
The spam news discrimination unit deriving a fake news probability model for deriving a probability of a fake news based on a distribution of correlation between fake news and a real news, and a real news probability model for deriving a probability of real news;
Calculating a probability that the news is real news and a probability that the news is fake news by substituting the correlation of the spam suspects into the fake news probability model and the real news probability model; And
And determining whether the news is real or fake according to the probability that the spam news discrimination unit is the real news and the probability that it is the fake news.
Method for detecting spam news. The method of claim 6,
The step of determining whether the news is real or fake,
The spam news discrimination unit derives a fake news probability model that derives a probability of being fake news based on a distribution of a correlation between fake news and real news, and a probability model of a real news that derives probability of being true news, and interrogates the spam suspect If the correlation of the less than the reference value set based on the fake news probability model and the real news probability model, characterized in that to determine the news as a fake
Method for detecting spam news. The method of claim 6,
The step of calculating the correlation (correlation)
And calculating a correlation of an average probability of each category of the paragraphs other than the above paragraphs with respect to the probability of each paragraph by category.
Device for detecting spam news. The method of claim 6,
Before the step of deriving the probability,
And when the paragraph is input, training the category classifier to output the probability that the input paragraph corresponds to each of a plurality of categories as a numerical value. A computer-readable recording medium in which a program for performing a method for detecting spam news according to any one of claims 6 to 10 is recorded.

Images