Identifikacija sudionika u transkriptu razgovora između više osoba uporabom analize teksta

master's thesis

• Title (english): Identifying Participants in Conversation Transcripts Involving Multiple People Using Text Analysis
• Author: Karlo Brajdić
• Mentor: Damir Pintar (mentor)
• Committee member: Damir Pintar (predsjednik povjerenstva)
• Committee member: Zoran Skočir (član povjerenstva)
• Committee member: Boris Vrdoljak (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2019-07-09, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Identifikacija sudionika u transkriptu razgovora je problem koji je predstavljen kao uvod u veći problem, a to je rudarenje karaktera. Problem se rješava rezolucijom koreferentnosti, ali prijašnjim modelima je tekst u obliku dijaloga bio mana. Model rezolucije koreferentnosti zasnovan na parovima spominjanja se koristi u ovom radu, koji uz pomoć više tipova značajki grupira spominjanja u koreferentne lance. Sustav nakon toga povezuje entitete s tim lancima pomoću algoritma zasnovanog na glasanju. Ugniježđena unakrsna provjera i nekolicina modela su korištena pri evaluaciji sustava.

Abstract (english)

Character identification on multiparty conversation is a problem introduced as an intro to a bigger problem called character mining. Usually, the problem is solved using coreference resolution, but coreference resolution models at the time were struggling with data as dialogue. Mention-pair model is used for coreference resolution in this thesis, which links mentions in coreference chains. After that, the system links entities with chains using a voting scheme. Nested cross-validation and a number of models were used during the evaluation of the system.

• Keywords: analiza teksta; identifikacija sudionika; strojno učenje; procesiranje prirodnog jezika; unakrsna provjera; rudarenje karaktera; rezolucija koreferentnosti; povezivanje entiteta
• Keywords (english): text analysis; character identification; machine learning; natural language processing; cross validation; character mining; coreference resolution; entity linking
• Language: croatian
• URN:NBN: urn:nbn:hr:168:995361
• 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: 2020-01-07 16:59:12