Plitko semantičko parsanje tekstova na hrvatskome jeziku

master's thesis

• Title (english): Shalow Semantic Parsing of Croatian Texts
• Author: Krešimir Baksa
• Mentor: Jan Šnajder (mentor)
• Committee member: Jan Šnajder (predsjednik povjerenstva)
• Committee member: Marko Čupić (član povjerenstva)
• Committee member: Slobodan Ribarić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2015-07-13, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U ovom radu definiran je skup semantičkih okvira za manji broj glagola koji su bili najzastupljeniji u dostupnom hrvatskom korpusu. Na temelju tih okvira označen je relativno malen skup rečenica iz korpusa te je izvršeno vrednovanje modela temeljenih na nadziranom strojnom učenju na zadatku plitkog semantičkog parsanja. Pokazano da na tako malom skupu primjera automatski modeli temeljeni na strojnom učenju postižu dobre rezultate F1=58.54. Na reduciranom skupu semantičkih uloga, modeli temeljeni na strojnom učenju nadmašuju rezultate osnovnog modela te u tom slučaju najbolju F1-mjeru postiže stroj potpornih vektora (F1=69.20).

Abstract (english)

In this thesis a set of semantic frames for smaller number of most common verbs in available Croatian corpus was defined. Based on those semantic frames relatively small set of sentences was labeled with semantic roles and, based on those sentences, the evaluation of supervised machine learning models was performed on a task of shallow semantic parsing. It is shown that even on relatively small training set models based on supervised machine learning can perform well and achive good results with F1=58.54. On a reduced set of sematic roles, machine learning models outperform baseline model and in that case the best F1-score is achived with Support Vector Machine (F1=69.20).

• Keywords: obrada prirodnog jezika; hrvatski jezik; plitko semantičko parsanje; nadzirano strojno učenje; logistička regresija; stroj potpornih vektora
• Keywords (english): natural laguage processing; Croatian language; shallow semantic parsing; supervised machine learning; logistic regression; support vector machine
• Language: croatian
• URN:NBN: urn:nbn:hr:168:239652
• 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 07:42:08