
Benazir T. A, Parvesh S · Zenodo (CERN European Organization for Nuclear Research) 2026 · 2026
DOI: 10.5281/zenodo.24238866
Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).
Competitive-examination candidates must monitor information published across multiple news sources and identify facts relevant to a particular examination and geographic region. This paper presents FlashFacts, an implemented current-affairs intelligence system designed to collect configured news items, normalize them into structured event records, classify them by domain and geography, and prioritize content according to examination relevance. The system supports natural-language queries that combine geographic and examination constraints, retrieves relevant stored events, and uses retrieved information to support answers and multiple-choice question generation. Tamil Nadu Public Service Commission (TNPSC) Group 4 is used as a representative use case, with emphasis on Tamil Nadu and India. The backend is implemented with FastAPI and Python, PostgreSQL is used for structured event storage, Redis provides supporting services, and Docker is used to package the application environment. The supplied project record reports a functional test run of 85 passing tests and one warning; this is reported as a project test result rather than a measure of educational effectiveness or model accuracy. A dashboard snapshot included in the manuscript displays 1,365 indexed events, 15 domain categories, and six permitted sources on 3 October 2026. These interface values are descriptive snapshot data and are not presented as an independently audited benchmark. The paper discusses the system workflow, implementation, observed functional evidence, and current limitations.
No comments yet — start the discussion below.