Institutional Repository of Vilnius University
Multi-Feature content moderation system with ai-based detection /
Abstract
dc:descriptionThis thesis addresses the problem of content moderation in Telegram group chats and proposes an automated moderation system solution. The work evaluates the necessity of such a system, analyses the content moderation domain and existing tools, and defines functional and non‑functional requirements for a new information system. Artificial intelligence and natural language processing methods are investigated, focusing on the use of Google Gemini models for hate speech, toxic content and spam detection, as well as rule‑based spam filters. Based on this analysis, a system design is prepared and a working multi‑feature moderation system called "Speech Sentinel" is implemented, providing real‑time message analysis, violation severity classification, automated warning and penalty mechanisms, appeal management and a web‑based administrative interface. The system is validated through functional, negative and security testing, and its accuracy and reliability are evaluated in realistic usage scenarios.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Ishchenko, Daniil,
Subjects
dc:subject × 1Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- eng
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD277690340&prefLang=en_US