ANNOTATED ARABIC RESOURCES
We developed a manually annotated data corpus for Arabic author profiling comprising of 15 regions. All regions have 198 users each with at least 100 arabic tweets and all users are equally with respect to age (3 age groups), gender and dialect.
We developed manually annotated Arabic corpora for irony and deception detection, based on data retrieved from Twitter.
ARABIC AUTHOR PROFILING TOOLS
Based on our corpora, we developed tools to automatically infer an author profile from anonymous Arabic text or Twitter handle using linguiistic features and machine learning techniques.
CYBER-SECURITY & MARKETING
The research is relevant for government agencies that fight against cyber-crimes for example in their investigation to narrow down the set of potential authors of a threat message.
We also addressed other application scenarios in marketing such as profiling followers and prospects on social media to undersatnd customer segmentation.
Copyright © 2017-2020 Arabic Author Profiling (ARAP) Research Sponsored by Qatar National Research Fund, NPRP Cycle 9 grant 9-175-1-033