Feature engineering for abusive language detection for the Ukr language
summary
Describes creating an abusive-language detector for Ukrainian tweets by engineering linguistic and surface features tailored to the language. Covers which feature types (e.g., lexical, morphological, orthographic) were extracted and used with standard machine-learning classifiers to distinguish abusive vs. non-abusive content. Discusses practical issues specific to Ukrainian on Twitter such as spelling variants and limited labeled data.
AI-written summary · about
chapters (5)
related talks
37mDetecting offensive messages using Deep Learning: A micro-service based approach
23mTowards Author Profiling - A Methodology for Gender Expression Classification
46mPractical Machine Learning and Rails
27mSentiment analysis of tweets in Polish language using deep...
30mOutwitting Deep Learning Models
21mMachine Learning and Cyber Security - Detecting malicious URLs in the haystack
38mQuality Assurance in Projects with Machine Learning
39mFeature Engineering: Unlocking Potential for Predictive Models