Optimizacija parametara ovisnosnog parsera za hrvatski jezik

master's thesis

• Title (english): Optimizing Dependency Parser Parameters for Croatian Language
• Author: Petra Bevandić
• Mentor: Jan Šnajder (mentor)
• Committee member: Jan Šnajder (predsjednik povjerenstva)
• Committee member: Stjepan Groš (član povjerenstva)
• Committee member: Nikola Bogunović (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2014-07-07, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Strojna sintaktička analiza rečenice ili parsanje preduvjet je za više razine strojne analize teksta. Postavlja se pitanje kako uspješno provesti parsanje morfosintaktički složenih jezika poput hrvatskog jezika.

U sklopu diplomskog rada provedeno je istraživanje utjecaja morfološke analize riječi u rečenici na rad postojećih ovisnosnih parsera te je implementiran sustav koji omogućava optimizaciju morfološkog skupa značajki. Implementirani je sustav primjenjen na hrvatski jezik kod kojeg se... More pokazalo da je moguće značajno smanjiti veličinu skupa morfoloških značajki bez narušavanja kvalitete rada parsera.

Razvijeni model ispitan je na hrvatskom srodnim jezicima slovenskom i češkom koji pokazuju slično ponašanje. Less

Abstract (english)

Parsing is the neccessary step in any kind of higher-level text analysis. However, it is still unclear how to improve parsing of morphologically rich languages such as Croatian.

This works presents the analysis of the impact different morphological features have on the success of parsing and describes the system implemented in order to optimize the set of morphological features which improves the quality of parsing. The implemented system was then used to optimize the morphological feature... More set for parsing of Croatian language. This works shows that it is possible to significantlly reduce the size of the morphological feature set while still maintaining the quality of parsing.

The model was then tested on Czech and Slovene language, and these test show similar results. Less

• Keywords: ovisnosno parsanje; optimizacija parametara; ovisnosne gramatike; morfološke značajke
• Keywords (english): dependency parsing; dependency grammar; featura optimization; morphological features
• Language: croatian
• URN:NBN: urn:nbn:hr:168:081468
• 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: 2018-02-06 07:52:58