KR20220000679A - Method and apparatus for self-regulating typed comment visualization - Google Patents

Abstract

Provided is a method and apparatus for visualizing a self-regulation type comment. The method is a method for visualizing a self-regulation type comment of a computing apparatus operated by at least one processor, and the method includes: a step of inputting a comment for a post input from a user terminal into an emotion classification model trained based on machine learning, and calculating a positive index, which is a positive emotion probability of the input comment, and a negative index, which is a negative emotion probability, from the emotion classification model; a step of setting a plurality of comments as predetermined figures, respectively, and differentially applying a hue, saturation, and blinking effect determined according to the positive and negative indexes of each comment to the figures set for each comment; and a step of listing the figures to which the hue, saturation, and blinking effect are differentially applied on one page, and displaying the figures together with the post. The present invention can reduce an act of posting malicious comments.

Description

Self-regulating comment visualization method and device {METHOD AND APPARATUS FOR SELF-REGULATING TYPED COMMENT VISUALIZATION}

The present invention relates to a self-regulating comment visualization method and apparatus.

With the development of the Internet and online news media, a commenting culture has emerged in which many people actively exchange opinions online on various interests or issues through comments.

A comment is a short reply to the original text that someone posted on the Internet. Comments have a positive function in that they form a forum for free discussion. However, it also has a negative function in that anonymity can be used to spread slander, hate, ridicule, or false information to others.

In the 2000s, as damage caused by malicious comments such as suicide cases of famous celebrities occurred, interest in the seriousness of malicious comments, the negative side of the comment culture, began to rise, and the demand for improvement of the comment culture is increasing.

Malicious comments refer to the act of writing posts online that may cause psychological harm to others. Malicious comments are activated due to the characteristics of Internet comments that are anonymous, non-face-to-face, and collective.

Anonymity of not exposing one's identity online makes swearing, slander, and groundless rumors spread more freely on the Internet. For those who own it, it can increase their tendency to use more violence on the Internet.

Also, if more people write similar malicious content, it makes them feel less responsible for hiding in the group and posting malicious comments.

Despite various attempts to reduce malicious comments, the problem of malicious comments is still unresolved and is recognized as one of the important social problems to be solved in the expanding online society.

Compared to the research on visualizing internet posts, there are relatively few studies on analysis and visualization of comments. As mentioned above, the purpose of various studies that visualized comments and blog posts is focusing on the overall situation by allowing a large number of information to be viewed at a glance. However, the conventional visualization does not have the purpose of reducing malicious comments.

In addition, the forced method of filtering and automatically blocking malicious comments is not yet perfect, so it is difficult to solve the problem of malicious comments by relying on it alone. Sometimes it blocks non-malicious comments, and sometimes it fails to block malicious comments. This is creating a lot of victims.

The task to be solved is to use the emote classification model to determine the positive and negative index of comments, and to provide a visual effect to warn you when a comment that is entered in real time is determined to be a malicious comment based on the negative index, and to provide a positive index of the comment And by displaying the figures differentially applied with hue, saturation, and blinking effect according to the negative index, and providing these figures on one page, malicious comments are created both in the process of writing comments and in the process of reading other people's comments It is to provide a self-regulating comment visualization method and device that induces overexposure to decrease naturally.

According to one aspect of the present invention, the self-regulating comment visualization method is a self-regulating comment visualization method of a computing device operated by at least one processor, and is generated by pre-learning a comment input by a user terminal with respect to a post. Input to the emotion classification model and calculating the positive index, which is the positive emotion probability, and the negative index, which is the negative emotion probability, of the input comment from the emotion classification model and differentially applying the hue, saturation, and blinking effects determined according to the negative index to the figures set for each comment, and listing the figures to which the hue, saturation, and blinking effects are differentially applied on one page to the post Including the step of displaying with

In the self-regulating comment visualization method, if the negative index of a comment inputted in real time from the user terminal with respect to the post is greater than or equal to a threshold index, the real-time inputted comment is determined as a malicious comment, and a visual effect of warning the malicious comment is displayed close to the comment input window. It may further include the step of outputting to a location.

In the outputting step, if as many morpheme units constituting the input comment are collected enough to calculate the negative index, continuously calculating the negative index based on the collected morpheme unit; If it is equal to or greater than a threshold index, outputting a warning icon image to the adjacent position, and if the negative index is greater than or equal to a second threshold index, applying an animation effect to the warning icon image.

After the calculating step, the empathy index indicating the number of sympathy input from the user terminals with respect to the comment and the non-sympathy index indicating the number of non-sympathy are collected, and the brightness of the figure is applied according to the empathy index and the non-sympathy index It may include further steps.

After the step of displaying, detecting a mouseover in one of the figures, and setting a delay timer when the positive index of the figure in which the mouseover is detected is less than or equal to the threshold index or the negative index exceeds the threshold index The method may further include popping up the comment content mapped to the figure when the delay timer expires.

Prior to the displaying, the method further includes collecting a sympathy index indicating the number of sympathy input from user terminals with respect to the comment and a non-sympathy index indicating the number of non-sympathy, wherein the pop-up includes: If the non-sympathy index of the comment for the detected figure is greater than the sympathy index by more than a threshold index, the sensed figure may be displayed by lowering the saturation.

After the calculating step, calculating the average of the positive index and negative index of each of the comments for the specific post as the comment index of the specific post, displaying the specific post as a figure, and the color tone of the figure according to the comment index and determining the saturation, and displaying the posts displayed as figures on one page.

After the displaying, the method may further include collecting the number of comments registered for the specific post in real time, and applying a blinking effect to the figure when the number of comments is collected more than a threshold number.

According to another feature of the present invention, the self-regulating comment visualization apparatus includes a communication device, a memory storing a program for implementing self-regulating comment visualization, and at least one processor executing the program, the at least one processor classifies the reply data received from the user terminal into units of morphemes, inputs the classified morpheme units into an emotion classification model learned based on machine learning, and calculates a negative index, which is the probability of negative emotions in units of morphemes, from the emotion classification model And, if the negative index is greater than or equal to a first threshold, a warning icon is output around the comment input window into which the comment data is input, and if the negative index is greater than or equal to a second threshold, an animation effect is applied to the warning icon, and The output of the warning icon and the animation effect are continuously applied based on the comment data input in real time.

The at least one processor sets each of a plurality of comments registered with respect to an arbitrary post to a predetermined figure, and applies hue, saturation, and blinking effects determined according to the positive and negative indexes of each comment to the figure set for each comment. It is applied differentially, and the figures to which the hue, saturation, and blinking effect are differentially applied can be arranged on one page and displayed together with the arbitrary post.

When the at least one processor detects a mouseover in one of the figures, if a positive index that is a positive emotion probability of the figure in which the mouseover is detected is less than or equal to a threshold index or a negative index exceeds a threshold index, After delaying for a predetermined time, the content of the comment mapped to the figure may be popped up.

The at least one processor calculates, for each post, an average of a positive index and a negative index, which is a probability of positive emotion of each of the comments on each post, as the comment index of each post, and displays each post as a figure to represent each post. The color tone and saturation of the figure are determined according to the comment index, the posts displayed in the figure are displayed on one page, and when the number of comments registered for each post is collected over a threshold number, a blinking effect is applied to the figure can do.

According to the embodiment, by using machine learning technology to provide a visual effect to warn users when a comment that is input in real time is determined to be a malicious comment, it helps users who leave a comment to become aware of the malicious comment autonomously to create malicious comments. can reduce

In addition, by displaying the comments on one page as figures to which hue, saturation, and blinking effects are differentially applied according to the positive and negative indices of the comments, it is possible to induce both the creation and exposure of malicious comments to decrease naturally.

1 shows the configuration of a comment visualization system for filtering and blocking malicious comments on posts according to an embodiment of the present invention.
2 is a flowchart illustrating a method for setting a reply feedback effect according to an embodiment of the present invention.
3 is an example of a real-time comment feedback effect according to an embodiment of the present invention.
4 is an example of a real-time comment feedback effect according to another embodiment of the present invention.
5 is a flowchart illustrating a method for setting a reply feedback effect according to another embodiment of the present invention.
6 shows an example of a feedback effect of all comments on a post according to another embodiment of the present invention.
7 is a flowchart illustrating a method for setting a reply feedback effect according to another embodiment of the present invention.
8 is an example of a real-time comment feedback effect according to an embodiment of the present invention.
9 is an example of a comment feedback effect according to another embodiment of the present invention.
10 is a hardware configuration diagram of a computing device according to an embodiment.

Hereinafter, with reference to the accompanying drawings, the embodiments of the present invention will be described in detail so that those of ordinary skill in the art to which the present invention pertains can easily implement them. However, the present invention may be embodied in several different forms and is not limited to the embodiments described herein. And in order to clearly explain the present invention in the drawings, parts irrelevant to the description are omitted, and similar reference numerals are attached to similar parts throughout the specification.

Throughout the specification, when a part "includes" a certain element, it means that other elements may be further included, rather than excluding other elements, unless otherwise stated.

In addition, terms such as “…unit”, “…group”, “…module”, etc. described in the specification mean a unit that processes at least one function or operation, which may be implemented by hardware or software or a combination of hardware and software. can

The devices described in the present invention are composed of hardware including at least one processor, a memory device, a communication device, and the like, and a program to be executed in combination with the hardware is stored in a designated place. The hardware has the configuration and capability to implement the method of the present invention. The program includes instructions for implementing the method of operation of the present invention described with reference to the drawings, and is combined with hardware such as a processor and a memory device to execute the present invention.

As used herein, "transmission or provision" may include not only direct transmission or provision, but also transmission or provision indirectly through another device or using a detour path.

In this specification, expressions described in the singular may be construed in the singular or plural unless an explicit expression such as “a” or “single” is used.

In this specification, regardless of the drawings, the same reference numbers refer to the same components, and "and/or" includes each and every combination of one or more of the referenced components.

In this specification, terms including an ordinal number such as first, second, etc. may be used to describe various elements, but the elements are not limited by the terms. The above terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present disclosure, a first component may be referred to as a second component, and similarly, a second component may also be referred to as a first component.

In the embodiments described in this specification with reference to the drawings, an arbitrary embodiment may be implemented alone, various embodiments may be combined or divided, and a specific operation may not be performed in each embodiment.

1 shows the configuration of a self-regulating comment visualization system according to an embodiment of the present invention.

Referring to FIG. 1 , the comment visualization system includes a comment visualization control device 100 , a web/app server 200 , a network 300 , and a plurality of user terminals 400 .

The comment visualization control device 100 recognizes the emotional expression of the comment attached to the post, and visualizes and displays how positive or negative the comment is based on this, and the comments of the negative expression are exposed after a certain time delay and , provides post visualization reflecting the emotional expression index of comments.

The comment visualization control device 100 collects and accumulates comments from the web/app server 200 in real time, and outputs the visualization instruction information determined according to the emotional expression recognition result of the comments to the web/app server 200 .

The web/app server 200 transmits the comments received from the user terminal 400 to the comment visualization control device 100 with respect to a post provided through a web or mobile application. The web/app server 200 provides a visual feedback effect according to the visualization instruction information received from the comment visualization control device 100 to a web or mobile application.

The plurality of user terminals 400 are connected to the web/app server 200 through the network 300 to receive a post, input a comment on the post, and transmit it to the web/app server 200, and for the post Receives registered comments and visual feedback effects according to the comments.

The network 300 includes a wired network or a wireless network, and includes various networks that are already known and will be developed in the future.

According to an embodiment, the comment visualization control device 100 includes a comment collecting unit 110 , a comment emotion determining unit 120 , a comment emotion index calculation unit 130 , a reply reaction collecting unit 140 , and a comment feedback effect setting. It includes a unit 150 and a feedback effect rule DB (Database) 160 .

The comment collecting unit 110 receives comments input from the plurality of user terminals 400 from the web/app server 200 and outputs them to the comment emotion determining unit 120 . In this case, the comments include incomplete sentences, that is, sentences being written in the comment input window of the web/app server 200 .

The comment emotion determining unit 120 analyzes the comment received from the comment collecting unit 110 to determine the emotional expression of the comment. In an embodiment of the present invention, the determination of the emotional expression of a comment includes determination of whether it is a positive expression or a negative expression.

The comment emotion determining unit 120 uses a machine learning model to determine the emotional expression of a comment, and a Keras sequential model may be used as the machine learning model. The comment emotion determining unit 120 determines whether the comment is a positive expression or a negative expression using a Keras sequential model learned in advance for emotional expression classification. Here, the Keras sequential model trained for classification of emotional expressions will be collectively referred to as "emotional expression classification model".

The comment emotion determining unit 120 determines whether the comment is a positive expression or a negative expression using the emotional expression classification model. A specific method of determining the comment emotion may use known techniques, but briefly described as follows.

The comment emotion determining unit 120 classifies the comment in units of morphemes, and inputs the divided morphemes as an emotion expression model. A morpheme may include a noun, a proposition, a verb, an adjective, a special character, an alphabetic case, an emoticon, and the like.

The emotional expression classification model can numerically output the probability of emotions corresponding to the input morphemes. In the embodiment of the present invention, the probability for positive emotion and the probability for negative emotion can be calculated and output. . For example, it is possible to output a positive emotion probability of 72% and a negative emotion probability of 25%.

Hereinafter, in the present specification, the positive emotion probability is referred to as a positive index (unit: %), and the negative emotion probability is referred to as a negative index (unit: %).

The comment emotion determination unit 120 outputs positive and negative indexes for the received comment to the comment feedback effect setting unit 140 .

The reply response collecting unit 130 collects user reactions input from the web/app server 200 to comments, that is, reply reactions. The comment response includes an empathy index and a non-sympathy index for the comment, and the empathy index and non-sympathy index may be calculated as a ratio of the number of sympathy selections to the number of non-sympathy selections in the overall index. The reply response collecting unit 130 outputs the sympathy index (unit: %) and the non-sympathy index (unit: %) for each comment to the reply feedback effect setting unit 140 for each comment.

The comment feedback effect setting unit 140 extracts the feedback effect corresponding to the positive index and negative index for the comment, the empathy index and the non-sympathy index for the comment from the feedback effect rule DB 150, and implements the extracted feedback effect Visualization instruction information instructing to do is output to the web/app server 200 .

According to an embodiment of the present invention, the comment feedback effect setting unit 140 may provide a real-time reply feedback effect. The real-time comment feedback effect is a visual effect displayed in an area adjacent to the comment input window, and may be a visual effect to warn that a comment being written may be a malicious comment. For example, the feedback effect may display a siren icon above the comment input window. This will be described as shown in FIGS. 2 to 4 .

2 is a flowchart illustrating a method for setting a reply feedback effect according to an embodiment of the present invention, FIG. 3 is an example of a real-time reply feedback effect according to an embodiment of the present invention, and FIG. 4 is another embodiment of the present invention This is an example of the real-time comment feedback effect.

Referring to FIG. 2 , the comment emotion determining unit 120 determines an emotion for a comment inputted in real time from the web/app server 200 ( S101 ). At this time, when the comment emotion determination unit 120 is a real-time comment target, when input words are collected enough to classify the emotional expression, the words collected so far are inputted to the emotional expression model, and affirmed from the emotional expression model Acquire exponents and negative exponents. Here, the words refer to data used as input to the emotional expression classification model, and include not only nouns, surveys, adjectives, alphabets, but also emoticons.

The comment feedback effect setting unit 140 generates visualization instruction information instructing to implement the visual effect defined in the size of the negative index determined in step S101 (S103).

The comment feedback effect setting unit 140 transmits the visualization instruction information generated in step S103 to the web/app server 200 (S105). Then, the web/app server 200 implements a visual effect according to the visualization instruction information on the screen of the user terminal 400 .

3 and 4 , the comment feedback effect setting unit 140 may generate visualization indication information instructing to display a siren icon image in a position close to the comment input window.

The siren icon may be expressed in a direction to reinforce the strength of the warning according to the magnitude of the negative index. For example, if the negative index is 50% or less or the positive index is 50% or more, the comment is determined as an overall positive expression, and a feedback effect is not provided. That is, no icons are displayed around the comment input window.

On the other hand, if the negative index exceeds 50%, the comment feedback effect setting unit 140 displays the siren icon image P21 on the comment input window P10 as shown in FIG. 3 to visually indicate that the currently written comment is a malicious comment. This may be the primary warning (alert).

If the negative feedback effect setting unit 140 exceeds 70%, as shown in FIG. 4 , the comment feedback effect setting unit 140 may output a moving siren icon image P23. That is, while the siren icon animation P23 displayed on the comment input window P10 is activated, a stronger visual warning is provided by movement.

This comment feedback effect is implemented for real-time input comments, not for completed comments. That is, when only as many words as can be determined emotions are input, the feedback feedback effect setting unit 140 may determine emotions each time and provide a feedback feedback effect. That is, by judging and visualizing the positive index and negative index of the comment being written in real time, it is possible to warn the comment author that the comment currently being written may be a malicious comment.

Of course, the visual effect is not limited to the siren icon, but the visual meaning of the siren, which is generally recognized by the public, is a warning. It may be optimal to give them a midnight chance to think for themselves.

In this way, when a user writes a comment to express an opinion on a post such as a news article, it is possible to determine in real time the degree of positive/negative of the comment entered by the user to provide immediate feedback.

According to another embodiment, the comment feedback effect setting unit 140 may provide a comment feedback effect that allows access to all comments within one screen by visualizing the positive and negative indices of all comments on one post. This will be described as shown in FIGS. 5 and 6 .

5 is a flowchart illustrating a method for setting a comment feedback effect according to another embodiment of the present invention, and FIG. 6 shows an example of a feedback effect of all comments on a post according to another embodiment of the present invention.

Referring to FIG. 5 , the comment feedback effect setting unit 140 collects all comments on an arbitrary post, that is, one post ( S201 ).

The comment feedback effect setting unit 140 determines the emotion (positive/negative) index for each of the collected comments (S203).

The comment feedback effect setting unit 140 displays one comment as a figure, and determines the hue and saturation of the figure mapped to the emotion (positive/negative) index of the comment (S205). The comment feedback effect setting unit 140 configures the figures to which the hue and saturation determined in step S205 are applied into one page and all comment visualization pages and displays them together with the post (S207).

The comment feedback effect setting unit 140 may determine the size of the figure of the comment and the size of the entire comment visualization page according to the number of comments registered for an arbitrary post. The color of the figure is visualized according to the positive and negative indexes of comments.

Referring to FIG. 6 , in the entire comment visualization page, the hue and saturation of figures are expressed differently according to the emotional (positive/negative) index of the comments.

Hue and saturation according to the emotion (positive/negative) index may be determined using Table 1 below.

negative exponent
100 → 0 positive quotient
0 → 100 Saturation, Hue
(0-359) 0 (red) → 100 (green) 100(green)→200(blue) Hue, Saturation
(0-100%) 100%→50% 50%→100%

Referring to Table 1, the most negative comments with the highest negative scores are visualized as saturated (100%) red (hue=0), and the most positive comments with the highest positive scores are visualized with saturated (100%) blue. It is visualized as (hue=200). Neutral comments that are neither positive nor negative are visualized as green with a saturation of 50% (hue=100), and in all other cases, the hue and saturation are determined by linear interpolation.

The comment feedback effect setting unit 140 collects the empathy index and the non-sympathy index for the comments (S209).

The comment feedback effect setting unit 140 determines the brightness and blinking of the figure based on the empathy index and the non-sympathy index (S211), and applies the determined brightness and blinking to the figure and displays it (S213).

Brightness is determined by the net empathy index minus the empathy index minus the empathy index. Blinking is determined by the total number of responses that are the sum of empathy and non-empathy. For example, when a specific comment receives a lot of responses from people, the shape has a blinking effect, allowing readers to intuitively search which comment is receiving the most attention.

In this way, the entire comment visualization page is usually ACOP (All Comments on One Page), which allows you to check all comments on one page at a glance, breaking away from the method of displaying comments by listing and scrolling on multiple pages. way to provide comments. The ACOP method makes it possible to understand the overall reaction and tendency of thoughts of readers to a post, and accordingly broadens the range of thoughts of readers, thereby preventing the case of being mentally harmed by some localized specific malicious comments.

In addition, the ACOP method enables intuitive search. That is, you can check the positive or negative trend of all comments on a post at a glance. Through the hue and brightness of the comment, you can tell whether the comment is a positive comment or a negative comment, and whether it is receiving a lot of sympathy or dislike. can be seen intuitively.

However, in the case of a comment subject to a time delay, which will be described later in FIG. 7 , that is, a malicious comment with a high negative index and non-sympathy index is excluded from the blinking effect in order not to receive the reader's attention.

In addition, the ACOP method can encourage the participation of readers to express their opinions. Unlike the list method, in which most comments are hidden and difficult to access if lost in the ranking, all comments are displayed on one page, so commenters can recognize that their comments are exposed along with all other comments and feel a sense of accomplishment. It can encourage more participation.

In addition, the ACOP method may provide a comment search function. According to one embodiment, the comment search function may be a method of filtering to display only the comments corresponding to the hue and saturation based on the hue and saturation input from the user terminal 400 as search criteria. According to another embodiment, the reply search function may be a method of confirming the contents of an actual comment corresponding to the figure when a specific figure is clicked among figures displayed on the entire comment visualization page. That is, without scrolling the screen, while viewing all comments at a glance, the actual content can be selected as a specific figure selected by the reader.

According to another embodiment, the comment feedback effect setting unit 140 may provide a comment feedback effect for filtering malicious comments.

All comments on an arbitrary post are displayed as the All Comments visualization page described in FIGS. 5 and 6 . In this case, the comment feedback effect setting unit 140 exposes the content of a comment written by another user through filtering through two steps, rather than being directly exposed. In other words, malicious comments can be filtered based on the hue/saturation of the figure and the time delay. This will be described with reference to FIGS. 7 and 8 .

7 is a flowchart illustrating a method for setting a reply feedback effect according to another embodiment of the present invention, and FIG. 8 is an example of a real-time reply feedback effect according to an embodiment of the present invention.

Referring to FIG. 7 , the comment feedback effect setting unit 140 displays the entire comment visualization page in which figures to which hue and saturation are applied according to the emotion (positive/negative) index are listed together with the posts with the comments (S301). That is, the comment feedback effect setting unit 140 displays only colored figures without directly exposing the content of the comment.

The comment feedback effect setting unit 140 detects a mouseover on an arbitrary figure among figures listed on the entire comment visualization page (S303). Of course, it is not limited to a mouseover, and may be various user input operations such as clicking a figure.

The comment feedback effect setting unit 140 determines whether the emotion (positive/negative) index mapped to the figure in which the mouseover is detected in step S303 satisfies a threshold condition (S305).

The threshold condition may include whether the positive index is greater than or equal to the threshold index, and whether the negative index is less than or equal to the threshold index. For example, if the positive threshold index is 50% and the negative threshold index is 70%, it may be determined whether the positive index mapped to the figure is 50% or more and the negative threshold index is 70% or less in step S305.

If it is determined that the comment feedback effect setting unit 140 satisfies the threshold condition in step S305, the comment content mapped to the figure is popped up (S307). That is, comments with a high positive index and a low negative index are immediately exposed in a pop-up form when the user places the mouse pointer on the shape of the comment.

If it is determined that the comment feedback effect setting unit 140 does not satisfy the threshold condition in step S305, it compares the empathy index and the non-sympathy index of the comment (S309).

The comment feedback effect setting unit 140 determines whether the non-sympathy index is greater than the empathy index (S311).

If the non-sympathy index is not greater than the empathy index, the comment feedback effect setting unit 140 performs step S307.

If the non-sympathy index is greater than the empathy index, the comment feedback effect setting unit 140 adjusts the saturation of the figure to be low and displays it (S313). And the comment feedback effect setting unit 140 drives the delay timer (S315), and when the delay timer expires (S317), pops up the comment content mapped to the figure (S319).

As such, when the comment feedback effect setting unit 140 receives more dislikes than likes from other users among the comments with a high negative index, the corresponding figure is expressed with low brightness (50%), Even if the mouse pointer is placed on the corresponding figure, the content is not displayed immediately, and the exposure can be further limited and displayed by giving a time delay (3 seconds). In other words, once a comment is written, the degree of exposure is determined autonomously by the participation of the public. Therefore, the more public participation and the higher the dislike index, the less exposed the comments.

As shown in FIG. 8(a), the comment feedback effect setting unit 140 lowers only the saturation of a specific shape even when a mouseover is detected on a specific shape, and after delaying a predetermined time, a position close to the shape as shown in FIG. 8(b) Pop up a comment on

This crowdsourcing method can reduce errors such as incorrect blocking or detection failure compared to methods that automatically detect and block malicious comments with a program. In addition, compared to the method of blocking by the site administrator according to the report of people, it is possible to reduce the burden on the administrator who has to manage numerous comments and remove the variable of the administrator's personal judgment.

According to another embodiment, the comment feedback effect setting unit 140 may provide a comment feedback effect for visualization of posts centered on comments. This will be described as shown in FIG. 9 .

9 is an example of a comment feedback effect according to another embodiment of the present invention.

Referring to FIG. 9 , the comment feedback effect setting unit 140 calculates the average of the positive index and the negative index of the comments for each post as the comment index of the post for each of the plurality of posts ( S401 ).

The comment feedback effect setting unit 140 determines the hue and saturation mapped to the comment index calculated in step S401 (S403). Then, a blinking effect is determined according to the number of comments on each post (S405).

The comment feedback effect setting unit 140 displays each post as a figure having the same shape, and differentially applies and displays the hue, saturation, and blinking determined in step S403 for each figure (S405).

As such, the comment feedback effect setting unit 140 may show all posts for a specific period (eg, the past 3 days) on one page, breaking away from the post enumeration method used in a general online site. If the post is news, this method may be referred to as an ANOP (All News on One Page) method.

The comment feedback effect setting unit 140 displays all posts on one page as figures to which hue and saturation are applied according to the comment index of the post, and applies a blinking effect to figures of posts with the number of comments equal to or greater than a threshold.

For example, if the color of most of the figures is red, it means that readers read the posts and left a lot of negative comments, suggesting that more negative events are now becoming news. Therefore, it is possible to intuitively grasp the tendency of readers to react to the recent posts.

In addition, in the case of a post with a lot of comments, a blinking effect is given so that readers can intuitively see the post that people are responding to a lot.

In this way, the comment feedback effect setting unit 140, when a mouseover is detected on any figure representing each post, exposes a summary of the post mapped to the figure in a pop-up method, and when the figure is clicked, the entire Posts can be exposed.

In this way, since posts are sorted and listed on one screen in the same way as comments visualization from positive to negative, readers can select and read articles based on that criteria. In other words, when some readers want to read mainly positive articles, this is possible.

Also, in general, in the case of a news home, only a small number of articles, such as articles that people have read a lot, latest articles, or headline news articles posted by a site are exposed to readers. However, in an embodiment of the present invention, since all posts during a specific period, for example, news articles have the same screen size and are exposed, it is possible to prevent a specific article from being buried for various reasons. Therefore, it can provide readers with an opportunity to broaden their thinking by increasing access to more news.

Next, the results of testing the Examples of the present invention will be described.

[Experimental example]

The experimenter collected comment data by selecting one news item from each of the political, social, and entertainment categories of Naver News. The fact that a comment was posted serves as a kind of signal that makes it visible that other people have paid attention to the news as well. Therefore, the experimenter selected the news article with the highest number of views counted during a specific period in the political and social sectors, and selected the news article that ranked first among the most viewed daily news among TV entertainment news in the entertainment sector. After collecting the content of the comments, the number of likes and dislikes for each selected news article, the experimenter calculates the net sympathy (sympathy - dissympathy) and the total number of responses (sympathy + dissympathy) of each comment and adds them A new data set was created.

The experimenter used the rating and review dataset of Naver Movies to classify the positive and negative degrees of the collected comment data. The rating and review dataset consists of a total of 200,000 reviews by collecting 100 reviews per movie. 10 points are negatively sampled at the same rate. Based on these rating and review data sets, a model that can predict the positive, negative, and his probability of a comment (i.e., classification of emotional expressions) using Keras, one of the deep learning libraries implemented in Python, is used. model) was created.

In order to find out whether the embodiment of the present invention is effective in reducing malicious comments and improving the comment culture, a survey was conducted after conducting a comparative experiment with the Naver news site for a total of 100 people in their 10s and 30s recruited online. The experimental group was selected from those who read Naver news articles at least once a week.

For 20 minutes, the experimenters read articles in three fields of politics, society, and entertainment, respectively, on the general Naver news site (comparison system 1) and the system to which the embodiment of the present invention was applied (comparison system 2), check the comments on the articles, and make comments. I was instructed to write it myself. In order to minimize the effect of the system experience sequence, the experiment was conducted in a counterbalancing method. In other words, after dividing 100 people into two groups, A and B, 50 people each, group A experienced comparison system 1 first and then used comparison system 2, and group B experienced comparison system 2 first and then used comparison system 1. did. As a result of the homogeneity test for the age and gender distribution between the two groups, there was no statistically significant difference, so the two groups were considered as homogeneous groups. Table 2 shows the homogeneity verification results for age and gender for groups A and B for the counterbalancing experiment.

Group A (N=50) Group B (N=50) p-value Age 24.34 25.12 .560 Gender Male 6 (12.0) 6 (12.0) One Female 44 (88.0) 44 (88.0)

Experimental participants responded to the questionnaire after experiencing Comparative System 1 and Comparative System 2, and all questionnaires were measured on a 7-point Likert Scale. Participants answered 5 questionnaires for the comparative evaluation of user experience between the two systems, and additionally answered 5 questions for the proposed system.

The survey question for comparison between the two systems was "A-1. Were many readers' thoughts about the article content easy to check through the comment system? A-2. Overall readers' positive/ Was it easy to detect negative reaction trends? A-5. Did you have easy access to many news articles through the news home system?”, and participants reported on scores ranging from “strongly disagree” (1) to “strongly agree” (7).

Additional questionnaire items related only to comparison system 2 were "B-1. It was helpful to search for comments on one page by three criteria (in order of total empathy (blink), order of net empathy (brightness), order of positivity (color))" B-2. In the news home, what kind of articles people are interested in at a specific time, a blinking visualization linked to the total number of comments on each article, as well as a color visualization of the average degree of positive or negative of those comments in color It was helpful to search B-3. Through the news home system, it was easy to understand the tendency of readers to respond positively or negatively to the entire article B-4. This comment system seems to be helpful in improving the commenting culture .B-5. I would like to use this news system again.

Corresponding sample t-test was performed to find out whether there was a statistically significant difference between the comparison system 1 and the comparison system 2 on the questionnaire result data. Table 3 shows this.

Questions Comparison system1 Comparison System 2 t(t-test value) p-value dF (degree of freedom) A-1 4.92±1.28 5.82±1.01 -6.25 <0.001 ^*** 99 A-2 4.88±1.62 5.82±1.30 -5.10 <0.001 ^*** 99 A-3 5.32±1.57 5.92±1.09 -4.24 <0.001 ^*** 99 A-4 5.89±1.12 5.49±1.54 2.28 .025 ^* 99 A-5 5.52±1.11 5.66±1.15 -1.04 .302 99

Referring to Table 3, in the items of A-1, A-2, and A-3, the comparison system 2 showed a statistically significant difference and a higher score (p<0.001) than the comparison system 1. In other words, it was found that the participants could easily grasp the general reader's thoughts along with the positive and negative trend information in the comparison system 2, and thought more about the influence that they could have on others when writing comments.

In questionnaire B, which was additionally asked only for comparative system 2, participants reported an average score of 5 or higher in all five items (B-1 (M=5.5), B-2 (M=5.8), B-3 ( M=6.1), B-4 (M=5.5), B-5 (M=5.5)), it was found that they were generally satisfied with the comparison system 2, had a re-use intention, and thought that it would be helpful in improving the comment culture.

Meanwhile, FIG. 10 is a structural diagram of a computing device according to another embodiment of the present invention.

The comment visualization control apparatus 100 including the reply collecting unit 110, the reply emotion determining unit 120, the reply reaction collecting unit 130 and the reply feedback effect setting unit 140 described in FIGS. 1 to 9 is a computer It may be implemented in a system, for example, a computer-readable medium.

Referring to FIG. 10 , in the computing device 500 operated by at least one processor, a program including instructions described to execute the operation of the present invention described with reference to FIGS. 1 to 9 may be executed. .

The computing device 500 includes a processor 510 , a memory 520 , an input interface device 530 , an output interface device 540 , a network interface 550 , and a storage device 560 , which include a bus 570 . can communicate through

The processor 510 may be configured to implement the methods described with reference to FIGS. 1 to 9 above. The processor 510 may be a central processing unit (CPU) or a semiconductor device that executes instructions stored in the memory 520 or the storage device 560 .

The memory 520 is connected to the processor 510 and stores various information related to the operation of the processor 510 . The memory 520 may store instructions to be executed by the processor 510 or may temporarily store instructions by loading the instructions from the storage device 560 . The processor 510 may execute instructions stored in or loaded from the memory 520 . The memory may include a read only memory (ROM) 521 and a random access memory (RAM) 522 .

In an embodiment of the present invention, the memory 520 may be located inside or outside the processor 510 , and the memory 520 may be connected to the processor 510 through various known means.

The network interface device 550 is connected to a network and configured to transmit and receive signals. In particular, the network interface device 550 is configured to provide data received through the network 200 to the processor 510 .

It should be understood that the systems, apparatuses and methods disclosed in the embodiments of the present invention may be implemented in other ways. For example, the disclosed device embodiments are by way of example only. For example, the division of processing units in a system is merely a logical division of functions, and may be other divisions during actual implementation. For example, multiple devices or components may be combined or integrated into other systems, or some functions may be ignored or not performed.

In addition, the devices configured to perform the method according to the embodiment of the present invention may be integrated into one device, each of the devices may exist alone physically, or two or more devices may be integrated into one device. An apparatus configured to perform the method described in the above embodiments may be implemented in the form of hardware or may be implemented in the form of a software functional unit.

The embodiment of the present invention described above is not implemented only through the apparatus and method, and may be implemented through a program for realizing a function corresponding to the configuration of the embodiment of the present invention or a recording medium in which the program is recorded.

Although the embodiments of the present invention have been described in detail above, the scope of the present invention is not limited thereto. is within the scope of the right.

Claims (12)

A self-regulating comment visualization method of a computing device operated by at least one processor, comprising:
Entering the comment input from the user terminal with respect to the post into an emotion classification model generated through pre-learning, and calculating a positive index that is a positive emotion probability of the input comment and a negative index that is a negative emotion probability from the emotion classification model;
setting each of the plurality of comments as a predetermined figure, and differentially applying the hue, saturation, and blinking effects determined according to the positive and negative indexes of each comment to the figures set for each comment; and
Listing the figures to which the hue, saturation, and blinking effects are differentially applied on one page and displaying them together with the post
Including, self-regulating comment visualization method. In claim 1,
When the negative index of a comment inputted in real time from the user terminal with respect to the post is greater than or equal to a threshold index, determining the real-time inputted comment as a malicious comment and outputting a visual effect of warning the malicious comment to a position close to the comment input window;
Further comprising, a self-regulating comment visualization method. In claim 2,
The output step is
continuously calculating the negative index based on the collected morpheme units when the number of morpheme units constituting the input comment is collected enough to calculate the negative index;
If the negative index is greater than or equal to a first threshold index, outputting a warning icon image to the adjacent position, and
If the negative index is greater than or equal to a second threshold index, applying an animation effect to the warning icon image
Including, self-regulating comment visualization method. In claim 1,
After the calculating step,
Collecting the empathy index indicating the number of sympathy input from the user terminals with respect to the comment and the non-sympathy index indicating the number of non-sympathy, and applying the brightness of the figure according to the empathy index and the non-sympathy index
Further comprising, a self-regulating comment visualization method. In claim 1,
After the step of displaying,
detecting a mouseover in one of the figures; and
When the positive index of the figure in which the mouseover is detected is less than or equal to the threshold index or the negative index exceeds the threshold index, driving a delay timer, and popping up the reply content mapped to the figure when the delay timer expires
Further comprising, a self-regulating comment visualization method. In claim 5,
Before the step of displaying,
Further comprising the step of collecting the empathy index indicating the number of sympathy input from the user terminals with respect to the comment and the non-sympathy index indicating the number of non-sympathy,
The step of pop-up is,
If the non-sympathy index of the comment for the figure in which the mouseover is detected has a greater difference than the empathy index by more than a threshold index, the self-regulating comment visualization method for displaying by lowering the saturation of the detected figure. In claim 1,
After the calculating step,
Calculating the average of the positive index and negative index of each of the comments on the specific post as the comment index of the specific post;
displaying the specific post as a figure, and determining the hue and saturation of the figure according to the comment index; and
Steps to display posts displayed in shapes on one page
Further comprising, a self-regulating comment visualization method. In claim 7,
After the step of displaying,
Collecting the number of comments registered for the specific post in real time, and applying a blinking effect to the figure when the number of comments is collected more than a threshold number
Further comprising, a self-regulating comment visualization method. communication device,
a memory in which a program for implementing the self-regulating comment visualization is stored; and
at least one processor executing the program;
the at least one processor,
Classifying the reply data received from the user terminal into morpheme units, inputting the classified morpheme units into an emotion classification model learned based on machine learning, and calculating a negative index, which is the probability of negative emotions of the morpheme units, from the emotion classification model,
If the negative index is greater than or equal to the first threshold, a warning icon is output around the comment input window where the comment data is input, and if the negative index is greater than or equal to the second threshold, an animation effect is applied to the warning icon,
The output of the warning icon and the animation effect are,
A self-regulating comment visualization device that is continuously applied based on real-time inputted comment data. In claim 9,
the at least one processor,
Each of the plurality of comments registered for a random post is set as a predetermined figure, and the color tone, saturation and blinking effects determined according to the positive and negative index of each comment are differentially applied to the figure set for each comment,
A self-regulating comment visualization device that lists the figures to which the hue, saturation, and blinking effect are differentially applied on one page and displays them together with the arbitrary post. In claim 10,
the at least one processor,
When a mouseover is detected in one of the figures, if the positive index, which is the probability of a positive emotion of the figure in which the mouseover is detected, is less than or equal to the threshold index, or if the negative index exceeds the threshold index, after delaying for a predetermined time, the A self-regulating comment visualization device that pops up the comment content mapped to the figure. In claim 9,
the at least one processor,
For each post, the average of the positive index and negative index, which is the probability of positive emotion of each of the comments for each post, is calculated as the comment index of each post,
Each post is displayed as a figure to determine the hue and saturation of the figure according to each comment index, and the posts displayed as figures are displayed on one page,
When the number of comments registered for each post is collected more than a threshold number, a blinking effect is applied to the figure, a self-regulating comment visualization device.

Images