KR102147496B1 - Method and system for blocking continuous input of similar comments - Google Patents

Abstract

A method and system for blocking continuous input of like comments are disclosed. The abuzzer detection method includes: performing embedding for expressing each sentence input as a comment as a vector for a plurality of comments that are continuously input or input within a predetermined interval for a unit time; Calculating a similarity between sentences inputted as comments using the embedded vector; And determining whether the plurality of comments are an abuser by using the similarity between the sentences.

Description

Method and system for blocking continuous input of similar comments {METHOD AND SYSTEM FOR BLOCKING CONTINUOUS INPUT OF SIMILAR COMMENTS}

The description below relates to a technique for blocking comments of similar content that are continuously input.

With the advancement of Internet technology, as Internet use becomes more popular, the age range of Internet use expands from children to the elderly, and not only can easily obtain desired information, but also freely express his or her thoughts and opinions.

The frequency of users' opinions or feedback appearing through comments on specific objects (persons, products, events, contents, etc.) on the Internet is increasing, and their influence is also increasing.

By exploiting anonymity in which information about the comment author is disclosed only by the user identifier or not at all, some netizens are prevalent in posting malicious comments such as criticizing, insulting, or slandering specific targets.

Recently, a method of blocking registration of malicious comments has been developed and provided. For example, Korean Patent Laid-Open Publication No. 10-2009-0001435 (published on January 09, 2009) includes not only malicious comments in the form of general words, but also emoticons. A technology that can block the registration of a malicious comment is disclosed.

Provides a method and system to block public opinion manipulation actions or macros that continuously input comments of similar content.

A method and system capable of detecting and blocking an abusing behavior using an embedding technique that expresses a sentence as a numerical vector is provided.

What is claimed is: 1. A method for detecting an abuser performed in a computer system, the method comprising: performing embedding for expressing each sentence input as a comment as a vector for a plurality of comments that are continuously input or input within a predetermined interval for a unit time; Calculating a similarity between sentences inputted as comments using the embedded vector; And determining whether the plurality of comments are an abuser using the similarity between the sentences.

According to an aspect, the performing may include converting a sentence input as a comment into a numeric vector in a multidimensional virtual space.

According to another aspect, the calculating may include calculating a similarity between the sentences by calculating a distance between the embedded vectors.

According to still another aspect, the determining may include predicting whether or not an abuser for the plurality of comments is performed using a deep neural network (DNN) learning model.

According to another aspect, the DNN learning model may use a comment input by the same person as training data, but as the training data, comments belonging to an abuser and comments that do not belong to the abuser may be learned together.

According to another aspect, the DNN learning model uses comments input by the same person as training data, but the training data may be selected using similarity between sentences input as comments.

According to still another aspect, the determining may include determining whether to be an abuser using an average of similarity between sentences for the plurality of comments.

According to another aspect, the determining may include calculating a similarity between sentences corresponding to comments in the plurality of comments, and then calculating a similarity average for all of the plurality of comments; And determining with an abuser if the average of the similarity is equal to or greater than a predetermined threshold.

It provides a non-transitory computer-readable recording medium in which a program for executing the method for detecting an abuzzer in a computer is recorded.

A computer system, comprising: a memory; And at least one processor connected to the memory and configured to execute computer-readable instructions included in the memory, wherein the at least one processor is a plurality of comments that are continuously input or input within a predetermined interval for a unit time An embedding execution unit for performing embedding for expressing each sentence input as a comment as a vector; A similarity calculation unit for calculating a similarity between sentences input as comments using the embedded vector; And it provides a computer system including an abuser determination unit for determining whether or not the plurality of comments is an abuser by using the similarity between the sentences.

According to embodiments of the present invention, comments of similar content such as public opinion manipulation actions or macros are continuously input by comparing the similarity of the subject of the sentence rather than the identity of the sentence itself using an embedding technology that expresses a sentence as a numeric vector. Abusing behavior can be effectively detected.

According to embodiments of the present invention, it is possible to improve comment evaluation performance by detecting a continuous input pattern of similar comments by comparing the similarity of the subject of a sentence, not the identity of the sentence itself, and by blocking continuous input of similar comments in advance. The spread of information or illegal information can be minimized.

1 is a diagram showing an example of a network environment according to an embodiment of the present invention.
2 is a block diagram illustrating an internal configuration of an electronic device and a server according to an embodiment of the present invention.
3 is a diagram illustrating an example of components that may be included in a processor of a server according to an embodiment of the present invention.
4 is a diagram illustrating an example of a method that can be performed by a processor of a server according to an embodiment of the present invention.
5 is an exemplary view for explaining a process of converting a comment sentence into an embedding vector in an embodiment of the present invention.
6 to 7 show examples of comments that can be used as training data of a recurrent neural network (RNN) learning model in an embodiment of the present invention.
FIG. 8 is an exemplary diagram for explaining a process of detecting whether a comment that is continuously input is an abuser using an RNN learning model according to an embodiment of the present invention.
9 to 10 are exemplary diagrams for explaining a process of detecting whether an abuser is present by using an average similarity of comments that are continuously input in an embodiment of the present invention.

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

Embodiments of the present invention relate to a technology for blocking malicious comments, and more particularly, to a method and system capable of blocking public opinion manipulation actions or macros that continuously input comments of similar content.

Embodiments including those specifically disclosed in this specification can detect and block public opinion manipulation actions or macros of patterns that continuously input comments of similar content, and through this, in terms of efficiency, accuracy, comment evaluation performance, etc. Significant advantages can be achieved.

1 is a diagram showing an example of a network environment according to an embodiment of the present invention. The network environment of FIG. 1 shows an example including a plurality of electronic devices 110, 120, 130, and 140, a plurality of servers 150 and 160, and a network 170. 1 is an example for explaining the present invention, and the number of electronic devices or servers is not limited as in FIG. 1.

The plurality of electronic devices 110, 120, 130, and 140 may be a fixed terminal implemented as a computer device or a mobile terminal. Examples of the plurality of electronic devices 110, 120, 130, 140 include smart phones, mobile phones, navigation systems, computers, notebook computers, digital broadcasting terminals, personal digital assistants (PDAs), portable multimedia players (PMPs). ), tablet PC, game console, wearable device, internet of things (IoT) device, virtual reality (VR) device, and augmented reality (AR) device. As an example, in FIG. 1, the shape of a smartphone is shown as an example of the electronic device 110, but in the embodiments of the present invention, the electronic device 110 is substantially different through the network 170 It may mean one of various physical computer devices capable of communicating with the electronic devices 120, 130, and 140 and/or the servers 150 and 160.

The communication method is not limited, and not only a communication method using a communication network (for example, a mobile communication network, wired Internet, wireless Internet, broadcasting network, satellite network, etc.) that the network 170 may include, but also short-range wireless communication between devices is included. I can. For example, the network 170 includes a personal area network (PAN), a local area network (LAN), a campus area network (CAN), a metropolitan area network (MAN), a wide area network (WAN), and a broadband network (BBN). , Internet, and the like. In addition, the network 170 may include any one or more of a network topology including a bus network, a star network, a ring network, a mesh network, a star-bus network, a tree or a hierarchical network, etc. Not limited.

Each of the servers 150 and 160 is a computer device or a plurality of computers that communicates with a plurality of electronic devices 110, 120, 130, and 140 through a network 170 to provide commands, codes, files, contents, services, etc. It can be implemented with devices. For example, the server 150 may be a system that provides a first service to a plurality of electronic devices 110, 120, 130, 140 accessed through the network 170, and the server 160 is also a network ( It may be a system that provides a second service to a plurality of electronic devices 110, 120, 130, and 140 accessed through 170). As a more specific example, the server 150 provides a service (for example, a comment service, etc.) targeted by the application through an application as a computer program installed and driven on a plurality of electronic devices 110, 120, 130, and 140. As a first service, it may be provided to a plurality of electronic devices 110, 120, 130, and 140. As another example, the server 160 may provide a service for distributing the above-described application installation and running file to the plurality of electronic devices 110, 120, 130, and 140 as the second service.

2 is a block diagram illustrating an internal configuration of an electronic device and a server according to an embodiment of the present invention. In FIG. 2, internal configurations of the electronic device 110 and the server 150 are described as examples of the electronic device. In addition, other electronic devices 120, 130, 140 or server 160 may also have the same or similar internal configuration as the electronic device 110 or server 150 described above.

The electronic device 110 and the server 150 may include memories 211 and 221, processors 212 and 222, communication modules 213 and 223, and input/output interfaces 214 and 224. The memories 211 and 221 are non-transitory computer-readable recording media, such as random access memory (RAM), read only memory (ROM), disk drive, solid state drive (SSD), flash memory, and the like. It may include a permanent mass storage device. Here, a non-destructive mass storage device such as a ROM, SSD, flash memory, or disk drive may be included in the electronic device 110 or the server 150 as a separate permanent storage device that is separate from the memories 211 and 221. In addition, the memories 211 and 221 include an operating system and at least one program code (for example, a browser installed and driven in the electronic device 110 or a code for an application installed in the electronic device 110 to provide a specific service) Can be saved. These software components may be loaded from a computer-readable recording medium separate from the memories 211 and 221. Such a separate computer-readable recording medium may include a computer-readable recording medium such as a floppy drive, disk, tape, DVD/CD-ROM drive, and memory card. In another embodiment, software components may be loaded into the memories 211 and 221 through communication modules 213 and 223 other than a computer-readable recording medium. For example, at least one program is a computer program installed by files provided by a file distribution system (for example, the server 160 described above) that distributes the installation files of developers or applications through the network 170 It may be loaded into the memories 211 and 221 based on (for example, the above-described application).

The processors 212 and 222 may be configured to process instructions of a computer program by performing basic arithmetic, logic, and input/output operations. The instructions may be provided to the processors 212 and 222 by the memories 211 and 221 or the communication modules 213 and 223. For example, the processors 212 and 222 may be configured to execute a command received according to a program code stored in a recording device such as the memories 211 and 221.

The communication modules 213 and 223 may provide a function for the electronic device 110 and the server 150 to communicate with each other through the network 170, and the electronic device 110 and/or the server 150 A function for communicating with an electronic device (for example, the electronic device 120) or another server (for example, the server 160) may be provided. As an example, a request generated by the processor 212 of the electronic device 110 according to a program code stored in a recording device such as the memory 211 is sent to the server 150 through the network 170 under the control of the communication module 213. ) Can be delivered. Conversely, control signals, commands, contents, files, etc. provided under the control of the processor 222 of the server 150 are transmitted via the communication module 223 and the network 170 to the communication module 213 of the electronic device 110. ) Through the electronic device 110. For example, control signals, commands, contents, files, etc. of the server 150 received through the communication module 213 may be transmitted to the processor 212 or the memory 211, and the contents or files may be transmitted to an electronic device ( 110) may be stored in a storage medium (permanent storage device described above) that may further include.

The input/output interface 214 may be a means for an interface with the input/output device 215. For example, the input device may include a device such as a keyboard, a mouse, a microphone, and a camera, and the output device may include a device such as a display, a speaker, and a haptic feedback device. As another example, the input/output interface 214 may be a means for interfacing with a device in which input and output functions are integrated into one, such as a touch screen. The input/output device 215 may be configured with the electronic device 110 and one device. In addition, the input/output interface 224 of the server 150 may be a means for an interface with a device (not shown) for input or output that is connected to the server 150 or that the server 150 may include. As a more specific example, when the processor 212 of the electronic device 110 processes a command of a computer program loaded in the memory 211, a service configured by using data provided by the server 150 or the electronic device 120 A screen or content may be displayed on the display through the input/output interface 214.

In addition, in other embodiments, the electronic device 110 and the server 150 may include more components than those of FIG. 2. However, there is no need to clearly show most of the prior art components. For example, the electronic device 110 may be implemented to include at least some of the input/output devices 215 described above, or other components such as a transceiver, a global positioning system (GPS) module, a camera, various sensors, and a database. May include more. As a more specific example, when the electronic device 110 is a smartphone, an acceleration sensor or a gyro sensor, a camera module, various physical buttons, a button using a touch panel, an input/output port, a vibrator for vibration generally included in the smartphone Various components such as may be implemented to be further included in the electronic device 110.

Hereinafter, a specific embodiment of a method and system for blocking continuous input of similar comments will be described.

One of the malicious comment blocking technologies is a method that blocks the continuous input of comments with the same content using the copy-paste function, which uses the similarity of the sentence itself to repeat the same comment. It blocks the act of writing.

The malicious comment blocking technology described above is vulnerable to the manipulation of public opinion by continuously entering macros or similar comments in recent years. This is because a macro or public opinion manipulation attempter enters a comment with similar content by changing the content of the comment little by little to bypass the existing method for blocking malicious comments.

Another method of blocking malicious comments is a method of limiting the number of comments that an individual can input during a unit time (eg, 1 day). However, the method of limiting the number of comments can damage good users with sound purposes or opinions, and it is possible to bypass any number of ways, such as using illegal accounts.

An act of manipulating a specific opinion into a number of opinions by using comments to achieve a specific purpose can be defined as an abuser, and the comments of these abusers are generally similar to the subject according to a specific purpose and specific opinion. It has a pattern that is input continuously.

In the present specification, comments that are continuously input are a plurality of comments input by the same user, and may include comments inputted within a predetermined interval for a unit time as well as comments input consecutively.

The present embodiments provide a technique for detecting and blocking an abuzzer in which comments of similar subject matter are continuously input.

3 is a block diagram showing an example of components that may be included in a processor of a server according to an embodiment of the present invention, and FIG. 4 is an example of a method performed by a server according to an embodiment of the present invention. Is a flow chart showing.

The server 150 according to the present embodiment may provide a function of filtering comments input to a platform related to the server 150. In other words, the server 150 may automatically filter comments input from a plurality of electronic devices 110, 120, 130, and 140 that are clients, and in particular, comments of similar content are continuously inputted. It can detect and block the abuser.

The processor 222 of the server 150 is a component for performing the method of detecting an abuser according to FIG. 4, and as shown in FIG. 3, an embedding execution unit 310, a similarity calculation unit 320, and an A determination unit 330 may be included. Depending on the embodiment, components of the processor 222 may be selectively included or excluded from the processor 222. Further, according to an embodiment, components of the processor 222 may be separated or merged to express the function of the processor 222.

The components of the processor 222 and the processor 222 may control the server 150 to perform the steps (S410 to S440) included in the method of detecting the abuzzer of FIG. 4. For example, the processor 222 and the components of the processor 222 may be implemented to execute an instruction according to the code of the operating system included in the memory 221 and the code of at least one program.

Here, the components of the processor 222 may be expressions of different functions of the processor 222 performed by the processor 222 according to an instruction provided by the program code stored in the server 150. . For example, the embedding execution unit 310 may be used as a functional expression of the processor 222 that controls the server 150 according to the above-described command so that the server 150 performs embedding representing a sentence as a vector. .

In step S410, the processor 222 may read a required command from the memory 221 loaded with a command related to control of the server 150. In this case, the read command may include a command for controlling the processor 222 to execute steps S420 to S440 to be described later.

In step S420, the embedding execution unit 310 may perform embedding in which each sentence input as a comment is expressed as a vector. The embedding execution unit 310 may perform embedding for a plurality of comments continuously input by the same user, and at this time, the plurality of comments are scheduled for a unit time (e.g., 1 day) as well as comments entered consecutively. Comments entered within the interval may be included. For example, the embedding execution unit 310 may convert comments inputted consecutively in the most recent order among comments input during a unit time into a vector.

For example, the embedding execution unit 310 may express a sentence input as a comment in a numerical multidimensional vector form. Referring to FIG. 5, the embedding execution unit 310 converts a sentence input as a comment into a number vector by using doc2vec, an embedding technology that converts a sentence into a number vector 501 that is a real value vector in a multidimensional virtual space 500. I can express it. The embedding execution unit 310 may remove meaningless text information for each sentence input as a comment using a language preprocessor such as a morpheme analyzer or an index word extractor. For example, after removing unnecessary parts of speech words or special symbols (e.g., !, ?, /, etc.), such as investigation and adjectives, from the sentence, extracting the rest of the sentence, that is, words corresponding to body language or root, You can convert words into numeric vectors.

In the present embodiment, an embedding technique for expressing a sentence as a numeric vector using doc2vec is applied, but it is not limited thereto, and it is also possible to use a known embedding technique such as word2vec and sense2vec.

In FIG. 4 again, in step S430, the similarity calculation unit 320 may calculate the similarity between sentences by using the embedding vector of each sentence input as a comment. In this case, the similarity calculation unit 320 may calculate a similarity score between sentences using a vector set representing words constituting a sentence. The similarity calculation unit 320 may calculate the similarity between sentences using doc2vec, which expresses the sentences as numeric vectors. For example, the similarity calculation unit 320 may calculate the similarity between sentences by calculating the Euclidean distance between doc2vec.

Doc2vec of sentence p is [p ₁ , p ₂ ,… , p _n ], and doc2vec of sentence q is [q ₁ , q ₂ ,… , q _n ], the Euclidean distance representing the similarity between two sentences p and q may be defined as in Equation 1.

[Equation 1]

The similarity calculation unit 320 may determine the similarity between sentences from the Euclidean distance between embedding vectors for comments that are continuously input for a unit time, and in this case, the closer the Euclidean distance is, the higher the sentence similarity is. The similarity for determining the similarity between sentences is not limited to the Euclidean distance, and various measurement methods can be applied in consideration of sentence characteristics such as cosine similarity and Hamming distance.

In step S440, the abuser determination unit 330 may determine whether a comment is a commenter by using the similarity between sentences with respect to comments continuously input for a unit time period, and filter and block comments determined as the abuser. Since the distance between embedding vectors indicates structural similarity between words as well as semantic similarity between words, calculating the similarity between sentences using doc2vec allows you to compare the content contained in the sentence, that is, the similarity of the subject, not simply sentence identity. It is possible to detect and defend against the comment abuser, which is a malicious act of continuously entering comments of comments.

The following two methods can be used to block the abuser of the pattern that continuously inputs comments with similar content.

For example, the abuzzer determination unit 330 may detect and block the abuzzer pattern with artificial intelligence using a learning model based on a recurrent neural network (RNN), which is one of deep neural networks (DNN). RNN is an artificial neural network that learns a nonlinear relationship including multiple hidden layers between the input layer and the output layer. By doing so, it is possible to predict whether or not the comment will be an abuser.

In order to build an RNN learning model, n comments inputted by one person in succession can be used as a training data set. Referring to FIGS. 6 and 7, as the learning data, an abuser sentence 600 classified as a malicious comment according to an abuser and a normal sentence 700 classified as a normal reply may be used. The training data set can be selected using the Euclidean distance between sentences. When comparing the Euclidean distance between the two sentences for the abuzzer sentence 600 and the normal sentence 700, in the case of the abuser sentence 600 in which similar contents are repeated, the Euclidean distance between sentences is the normal sentence 700. It can be seen that it is significantly lower than that. Comments that one person continuously inputs are used as learning data, but continuous input comments whose Euclidean distance between sentences is less than or equal to a certain value (d1) are used as the Abuser sentence 600, and the Euclidean distance between sentences is a certain value. (d2, d2>d1) or more consecutive input comments may be used as the normal sentence 700.

As the correct answer data set for the training data, whether or not an abuser is present may be input, in other words, the correct answer data is'the abuser' for the abuser sentence 600, and'is not an abuser' for the normal sentence 700. Is entered.

The abuser determination unit 330 may predict whether or not a comment that is input for determining the abuser is an abuser through an RNN learning model in which comments belonging to the abuser and comments that do not belong to the abuser are learned. Referring to FIG. 8, the abuser determination unit 330 is an RNN learning model 800 that uses a set of comments 801 that are subject to the abuser determination, that is, n comments continuously inputted by one person during a unit time as input data. ) Can be used to predict whether the corresponding comment set 801 is an abuser. The abuzzer determination unit 330 is a discrimination value indicating whether an abuzzer is the output data of the last neural network of the RNN learning model 800 (for example, a neural network that receives the most recent comments among n comments 801). It is possible to block input of comments when it is determined as abusing by determining whether it is an abusing comment that continuously inputs comments of similar content.

Accordingly, the abuzzer determination unit 330 may detect and defend a comment of a corresponding abuzzing pattern by using an RNN learning model in which comments of an abuzzer pattern that continuously input similar content are learned.

As another example, the abuser determination unit 330 may detect and block a comment abuser by using an average of the similarity between sentences input as comments for comments that are continuously input. Referring to FIG. 9, the abusor determination unit 330 sequentially obtains the similarity (for example, Euclidean distance) between sentences corresponding to consecutive comments from a set of n comments 901 continuously input by one person. After that, the similarity average for the entire comment set 901 is calculated, and if the similarity average is greater than or equal to a predetermined threshold, the comment input may be blocked by judging by the abuser.

The threshold value is a criterion value for filtering comments, and may be set according to the comment management policy, and when the threshold value is set lower, filtering is strengthened and strict comment management can be performed.

Referring to FIG. 10, a macro or public opinion manipulation attempter continuously inputs comments (1010) of contents blaming a specific target for malicious purposes. At this time, in order to avoid pattern detection, it is correlated with the subject of interest. A subject leaving comment 1011 is entered. If the average of the similarity for n consecutively inputted comments set 1010 is greater than or equal to the threshold value, that is, if the consecutively inputted comment sentences are similar to the average more than the threshold, it is judged by an abuser to leave the subject comment (1011). ) Can be included to detect and defend against an abuser.

Accordingly, the abuser determination unit 330 can determine whether a person is an abuser by using the average similarity of n comments continuously inputted by a person, and thus detects even if a comment that bypasses the continuous input pattern is included. In other words, in this embodiment, not simply limiting the number of comments that can be entered by an individual, but limiting the continuous input of comments with similar content within a certain period of time, the efficiency of macro or public opinion manipulation may be reduced.

As described above, according to embodiments of the present invention, comments of similar contents such as public opinion manipulation actions or macros are compared by comparing the similarity of the sentence subject rather than the identity of the sentence itself using an embedding technology that expresses the sentence input as a comment as a numeric vector. It is possible to effectively detect abusing behavior that continuously inputs.

In addition, according to embodiments of the present invention, it is possible to improve comment evaluation performance by comparing the similarity of the sentence subject, not the same as the sentence itself, to detect a continuous input pattern of similar comments, and to block continuous input of similar comments in advance. Thus, the spread of false or illegal information can be minimized.

The apparatus described above may be implemented as a hardware component, a software component, and/or a combination of a hardware component and a software component. For example, the devices and components described in the embodiments include a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable gate array (PLU). It may be implemented using one or more general purpose computers or special purpose computers, such as a logic unit, a microprocessor, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and one or more software applications executed on the operating system. In addition, the processing device may access, store, manipulate, process, and generate data in response to the execution of software. For the convenience of understanding, although it is sometimes described that one processing device is used, one of ordinary skill in the art, the processing device is a plurality of processing elements and/or a plurality of types of processing elements. It can be seen that it may include. For example, the processing device may include a plurality of processors or one processor and one controller. In addition, other processing configurations are possible, such as a parallel processor.

The software may include a computer program, code, instructions, or a combination of one or more of these, configuring the processing unit to behave as desired or processed independently or collectively. You can command the device. Software and/or data may be embodyed in any type of machine, component, physical device, computer storage medium or device to be interpreted by the processing device or to provide instructions or data to the processing device. have. The software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. In this case, the medium may be one that continuously stores a program executable by a computer, or temporarily stores a program for execution or download. In addition, the medium may be a variety of recording means or storage means in a form in which a single or several pieces of hardware are combined, but is not limited to a medium directly connected to a computer system, but may be distributed on a network. Examples of media include magnetic media such as hard disks, floppy disks and magnetic tapes, optical recording media such as CD-ROMs and DVDs, magnetic-optical media such as floptical disks, and And a ROM, RAM, flash memory, and the like, and may be configured to store program instructions. In addition, examples of other media include an app store that distributes applications, a site that supplies or distributes various software, and a recording medium or a storage medium managed by a server.

As described above, although the embodiments have been described by the limited embodiments and drawings, various modifications and variations are possible from the above description by those of ordinary skill in the art. For example, the described techniques are performed in a different order from the described method, and/or components such as a system, structure, device, circuit, etc. described are combined or combined in a form different from the described method, or other components Alternatively, even if substituted or substituted by an equivalent, an appropriate result can be achieved.

Therefore, other implementations, other embodiments, and claims and equivalents fall within the scope of the claims to be described later.

Claims (17)

In the abuzzer detection method performed in a computer system,
Performing embedding in which each sentence input as a comment is expressed as a vector by inputting a plurality of comments that are continuously input or input within a predetermined interval for a unit time;
Calculating a similarity between sentences inputted as comments using the embedded vector; And
Determining whether the plurality of comments are an abuser by using the similarity between the sentences
Including,
The determining step,
Based on the similarity between the sentences, the plurality of comments are determined as comments of an abusing pattern in which similar content is continuously input,
Calculating a similarity between sentences corresponding to the comments in the plurality of comments and then calculating an average of the similarity of all the plurality of comments;
Determining by an abuser if the average of the similarity is equal to or greater than a predetermined threshold as a reference value for filtering comments; And
Filtering and blocking comments judged by the abuser
Abuser detection method comprising a. The method of claim 1,
The performing step,
Converting a sentence entered as a comment into a numeric vector in a multidimensional virtual space
Abuser detection method comprising a. The method of claim 1,
The calculating step,
Calculating the similarity between the sentences by calculating the distance between the embedded vectors
Abuser detection method comprising a. The method of claim 1,
The determining step,
Predicting whether an abuser for the plurality of comments is made using a deep neural network (DNN) learning model
Abuser detection method comprising a. The method of claim 4,
The DNN learning model uses comments input by the same person as training data, but as the training data, comments belonging to the averaging and comments not belonging to the averaging are learned together.
Abuser detection method, characterized in that. The method of claim 4,
The DNN learning model uses comments entered by the same person as training data, but the training data is selected using similarity between sentences entered as comments.
Abuser detection method, characterized in that. delete delete A non-transitory computer-readable recording medium in which a program for executing the method of detecting an abuser according to any one of claims 1 to 6 is recorded on a computer. In a computer system,
Memory; And
At least one processor connected to the memory and configured to execute computer readable instructions contained in the memory
Including,
The at least one processor,
An embedding execution unit that performs embedding for expressing each sentence input as a comment as a vector by inputting a plurality of comments that are continuously input or input within a predetermined interval for a unit time;
A similarity calculation unit for calculating a similarity between sentences input as comments using the embedded vector; And
An abuser determination unit that determines whether or not the plurality of comments are an abuser using the similarity between the sentences
Including,
The abuser determination unit,
Based on the similarity between the sentences, the plurality of comments are determined as comments of an abusing pattern in which similar content is continuously input,
After calculating the similarity between sentences corresponding to the comments in the plurality of comments, an average of the similarity of all the plurality of comments is calculated,
If the average of the similarity is greater than or equal to a predetermined threshold as a reference value for filtering comments,
Filtering and blocking comments judged as the above
Computer system, characterized in that. The method of claim 10,
The embedding execution unit,
Converting a sentence entered as a comment into a numeric vector in a multidimensional virtual space
Computer system, characterized in that. The method of claim 10,
The similarity calculation unit,
Calculating the similarity between the sentences by calculating the distance between the embedded vectors
Computer system, characterized in that. The method of claim 10,
The abuser determination unit,
Predicting whether the plurality of comments are an abuser using a DNN learning model
Computer system, characterized in that. The method of claim 13,
The DNN learning model uses comments input by the same person as training data, but as the training data, comments belonging to the averaging and comments not belonging to the averaging are learned together.
Computer system, characterized in that. The method of claim 13,
The DNN learning model uses comments entered by the same person as training data, but the training data is selected using similarity between sentences entered as comments.
Computer system, characterized in that. delete delete

Images