Download | - View final version: On cross-dataset generalization in automatic detection of online abuse (PDF, 679 KiB)
- View accepted manuscript: On cross-dataset generalization in automatic detection of online abuse (PDF, 641 KiB)
|
---|
Author | Search for: Nejadgholi, Isar1; Search for: Kiritchenko, Svetlana1 |
---|
Affiliation | - National Research Council of Canada. Digital Technologies
|
---|
Format | Text, Article |
---|
Conference | WOAH 2020 - Fourth Workshop on Online Abuse and Harms, November 20th, 2020 - [Held Online] |
---|
Abstract | NLP research has attained high performances in abusive language detection as a supervised classification task. While in research settings, training and test datasets are usually obtained from similar data samples, in practice systems are often applied on data that are different from the training set in topic and class distributions. Also, the ambiguity in class definitions inherited in this task aggravates the discrepancies between source and target datasets. We explore the topic bias and the task formulation bias in cross-dataset generalization. We show that the benign examples in the Wikipedia Detox dataset are biased towards platformspecific topics. We identify these examples using unsupervised topic modeling and manual inspection of topics’ keywords. Removing these topics increases cross-dataset generalization, without reducing in-domain classification performance. For a robust dataset design, we suggest applying inexpensive unsupervised methods to inspect the collected data and downsize the non-generalizable content before manually annotating for class labels. |
---|
Publication date | 2020-11-20 |
---|
Date created | 2020-11-30 |
---|
Publisher | Association for Computational Linguistics |
---|
Licence | |
---|
In | |
---|
Language | English |
---|
Peer reviewed | Yes |
---|
Export citation | Export as RIS |
---|
Report a correction | Report a correction (opens in a new tab) |
---|
Record identifier | 846aa815-b9d6-4b34-9501-9163df950d7a |
---|
Record created | 2020-11-30 |
---|
Record modified | 2020-12-02 |
---|