Model koherentnosti tekstova na hrvatskome jeziku temeljen na analizi entiteta

undergraduate thesis

• Title (english): Entity-Based Coherence Model for Croatian Texts
• Author: Jura Šlosel
• Mentor: Jan Šnajder (mentor)
• Committee member: Jan Šnajder (predsjednik povjerenstva)
• Committee member: Goran Glavaš (član povjerenstva)
• Committee member: Mladen Karan (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2015-09-16, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Koherentnost je značajka prirodnog testka koja direktno utječe na razumljivost teksta, a opisuje povezanost djelova teksta u cjelinu. U radu je prezentiran model lokalne koherentnosti temeljen na rešetci entiteta za tekstove na hrvatskom jeziku. Algoritam iz tekstova vadi entitete i njihove gramatičke uloge te apstrahira tekst u rešetku prijelaza uloga entiteta te na sintetički označenom korpusu uči raspodjelu prijelaza karakterističnu za koherentne tekstove. Rezultantni model rangira ulazne tekstove po koherentnosti. Model postiže visoku točnost na zadatku određivanja poretka rečenica, 87%-93% na kratkim novinskim člancima.

Abstract (english)

Coherence is a feature of natural texts that directly influences its understandability and corresponds to the level of connectedness of parts of texts into a logical whole. This thesis presents an entity grid based model of local coherence for Croatian texts. The algorithm extracts entites and their grammatical roles from texts and abstracts the text into a grid of entity role transitions. The model learns distributions of transitions characteristic for coherent texts on synthetically annotated corpora. The resulting model rangs input texts by coherence. A high performance is achieved by the model on the sentence ordering task, 87%-93% on short newspaper articles.

• Keywords: obrada prirodnog jezika; hrvatski jezik; diskurs; rešetka entiteta; lokalna koherentnost; entitet
• Keywords (english): natural language processing; Croatian language; discourse; entity grid; local coherence; entity-based
• Language: croatian
• URN:NBN: urn:nbn:hr:168:845198
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2018-02-06 12:20:51