Work

Curated Source Network and Credibility Dataset for Khoj: An AI-Driven Real-Time Fact-Verification Framework in Bengali

Year: 2026

Type: dataset

Abstract: This dataset contains the curated list of 384 authoritative sources used by Khoj, an AI-driven composite scoring framework for real-time fact verification in Bengali. To mitigate misinformation and ground LLM reasoning, the sources have been rigorously selected across national newspapers, international reliable media, regional reporting portals, government gazettes, accredited fact-checkers (IFCN/EFCSN certified), and peer-reviewed scientific/reference resources. Dataset Details:- Total Sources: 384- Columns: 1. ID: Unique integer identifier (1–384) 2. Name: Publication or platform name 3. URL: Canonical landing URL 4. Category: Domain classification (e.g., National Newspapers, Regional Portals, International Media, Worldwide Fact-Checking, Science & Reference Resources)- Format: CSV (UTF-8 encoded) This resource serves as the foundational knowledge-retrieval layer for the priority parallel crawler and credibility-gated retrieval pipeline in Khoj.

Source: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.24066629

Authors: Tusher, Mahatir Ahmed, Dey, Sagar Chandra, Ferdushie, Zannatul

Institutions: VIT-AP University, University of Frontier Technology, Bangladesh

Language: Bengali

Topic: Misinformation and Its Impacts

Subfield: Sociology and Political Science

Field: Social Sciences

Domain: Social Sciences

Open Access status: green

OpenAlex ID: W7225614138

Cites:

Cited by:

Related to: 10