Automatic detection of toxic expressions considering coined/hidden words and contexts for preventing cyber bullying among children
| Project/Area Number |
17K13254
|
|---|
| Research Category |
Grant-in-Aid for Young Scientists (B)
|
| Allocation Type | Multi-year Fund |
|---|
| Research Field |
Childhood science (childhood environment science)
|
|---|
| Research Institution | Ritsumeikan University |
|---|
Principal Investigator |
Nishihara Yoko 立命館大学, 情報理工学部, 准教授 (70512101)
|
|---|
| Project Period (FY) |
2017-04-01 – 2021-03-31
|
| Project Status |
Completed (Fiscal Year 2020)
|
| Budget Amount *help |
¥2,600,000 (Direct Cost: ¥2,000,000、Indirect Cost: ¥600,000)
Fiscal Year 2019: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Fiscal Year 2018: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000)
Fiscal Year 2017: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000)
|
|---|
| Keywords | ネットいじめの防止 / ネットいじめ / 不適切表現 / 言い換えの獲得 / 時系列深層学習 / 文脈の学習 / 情報工学 |
|---|
| Outline of Final Research Achievements |
In order to prevent cyber bullying among children, this study investigated a method to automatically detect toxic expressions used in text messages and bulletin boards on the Web. I assumed that there are two types of toxic expressions: direct and indirect expressions. Direct expressions are those that anyone can see as toxic, such as "stupid" or "idiot. Indirect expressions are those that can be toxic depending on the context, such as hidden words and coined words. I proposed a dictionary-based method for detecting direct expressions. As a method to detect indirect expressions, I proposed a method to represent the context and detect indirect harmful expressions by using time-series deep learning. Using the obtained automatic detection method, I also proposed a method to encourage children to take down messages containing toxic expressions.
|
| Academic Significance and Societal Importance of the Research Achievements |
不適切な表現の判定は情報フィルタリングの分野で研究が進められてきた。既存研究にも直接的な不適切表現をフィルタリングする手法は提案されているが、隠語や造語を用いることでフィルタリングを回避することは可能であった。本研究では文脈によって不適切な表現となりうる隠語や造語の判定手法を実現した点に意義がある。
不適切な表現が含まれるかどうかを判定することはできるが、結局のところ投稿をするしないはユーザの判断に任されており、不適切な表現の判定だけではネットいじめの防止は難しい。本研究では不適切な表現が含まれる時に取り下げを促す手法を提案し、有用性を確認した点に意義がある。
|
Report
(5 results)
Research Products
(5 results)
