Provjera istinitosti činjenica iz teksta korištenjem metoda strojnog učenja

master's thesis

• Title (english): Fact-Checking from Text using Machine Learning Methods
• Author: Roman Kerčmar
• Mentor: Jan Šnajder (mentor)
• Committee member: Jan Šnajder (predsjednik povjerenstva)
• Committee member: Alan Jović (član povjerenstva)
• Committee member: Domagoj Jakobović (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2020-07-10, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Provjerom istinitosti činjenica iz teksta želi se validirati točnost određene tekstne tvrdnje na temelju čistog teksta i neke baze znanja, u slučaju ovog rada, Wikipedije. Korištene su metode strojnog i dubokog učenja. Korišten je okvir zadatka s radionice FEVER. Isprobani su referentni model s radionice FEVER i alternativne komponente za odabir dokumenata i rečenica. Kao najbolji model po FEVER-metrici na jednostavnijim primjerima pokazao se referentni s izmijenjenom komponentom za odabir... More dokumenata (0.31) dok se na složenijim primjerima kao najbolji pokazao referentni model (0.27). Less

Abstract (english)

Fact Checking from natural language aims to validate the correctness of the claim using raw text of that claim and a given knowledge base, in our case Wikipedia articles. We used machine learning and deep learning. We also used FEVER Workshop task as a problem framework. We tried FEVER baseline model and modified it with alternative document and sentence selection components. On simpler examples document selection modified model gave the best results by FEVER metric (0.31) and on more... More complex examples no model outperformed the baseline (0.27). Less

• Keywords: Obrada prirodnog jezika; strojno učenje; duboko učenje; provjera činjenica; dvosmjerna povratna neuronska mreža; ćelija s dugoročnom memorijom; engleski jezik
• Keywords (english): Natural language processing; machine learning; deep learning; fact checking; convolutional neural network; bidirectional neural network; long short-term memory unit; English language
• Language: croatian
• URN:NBN: urn:nbn:hr:168:593295
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-13 10:09:24