An important part of bullding  hate speech classifier is access to labelled corpora for training

 - No benchmark data

 - authors usually collect and label their own data.

 - annotating hate speech is an extremely time consuming effort.

Because of less harmful comments compared to harmless comments the  dataset tends to be imbalanced.


San Jose State University

The rise of social media created a platform for the general audience to express and communicate their perspective towards societal issues and so a need to automatically detect hate speech increased over the time. The basic word filtering methods used by NLP are not sufficient to handle this issue so a specific system has to be designed to handle this hate speech.

Hate Speech Detection using Natural Language Processing

A Survey on Hate Speech Detection using Natural Language Processing by Anna Schmidt  and Michael Wiegand

https://aclanthology.org/W17-1101.pdf

A Survey on Hate Speech Detection using Natural Language Processing

Hate is expressed differently by different people. 
We can say if anyone belittle any other person/persons based on race ,color, ethnicity, gender, sexual orientation, nationality, religion, or other characteristic then its a hate speech.

Analysis of how hate is expressed or communicated for NLP 

 - Domain of utterance
 - Discourse context along with images ,videos, audio 
 - Time of posting and word event at the time of posting 
 - Identity of Author
 - Targeted audience


NLP's Perspective for Hate Speech's Origin and Influence

 - Go fucking kill yourself and die already useless ugly pile of shit scumbag. 
 - The Jew Faggot Behind The Financial Collapse 
 - Hope one of those bitches falls over and breaks her leg


Examples of Hate Speech

Features from the data used for a classification model always have a greater impact on how the model works. Supervised learning is used in hate speech classification models.
The various features extracted from the text are

 - Simple Surface Features
 - Word Generalization
 - Sentiment Analysis
 - Lexical Resources
 - Linguistic Features
 - Knowledge-Based Features
 - Meta-Information
 - Multimodal Information




Textual Features extracted for Hate Speech Detection using NLP

Another interesting research work is to assign name tags for people involved directly and indirectly in a hate speech. The entire bullying event or(bullying trace )i assessed and has come up with names such as 
 - bully, 
 - victim, 
 - assistant,
 - defender, 
 - bystander, 
 - reinforcer, 
 - reporter 
 - accuser


Persons Involved in Bullying Episodes and Their Roles in NLP

Another area of researches are to detect hate speech detection within a short span of time inorder to predict unexpected and unfortunate events like racial violence ,terrorist attacks before they occur. 

Anticipating Alarming Societal Changes using NLP

 - Supervised learning methods
 - Support vector machines 
 - Recurrent Neural Network Language Models
 - One strp classification approaches
 - Mutl step classification approaches used in sentiment analysis
 - Semi supervised approaches for instance bootstrapping


Classification Methods employed for Hate Speech Detection

Data and Annotation for Hate Speech Detection using NLP

 - Need for bench mark data 

 - Multilingual perspective to hate speech may be worthwhile

 - Very less work on classification approaches using knowledge-based features , metainformation or multi-modal information.

- Authors are worried that the hate speech problem is similar to sarcasm where only limited subtypes can be detected using NLP.


Learn Before

Related