prikaz prve stranice dokumenta Modeli dubokog učenja za automatsko sažimanje dokumenta
No public access
master's thesis
Modeli dubokog učenja za automatsko sažimanje dokumenta
Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing, 2019. urn:nbn:hr:168:167130
Kukovačec, Marin
University of Zagreb
Faculty of Electrical Engineering and Computing

Institutional repository: FER Repository

Cite this document

Kukovačec, M. (2019). Modeli dubokog učenja za automatsko sažimanje dokumenta (Master's thesis). Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing. Retrieved from https://urn.nsk.hr/urn:nbn:hr:168:167130

Kukovačec, Marin. "Modeli dubokog učenja za automatsko sažimanje dokumenta." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2019. https://urn.nsk.hr/urn:nbn:hr:168:167130

Kukovačec, Marin. "Modeli dubokog učenja za automatsko sažimanje dokumenta." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2019. https://urn.nsk.hr/urn:nbn:hr:168:167130

Kukovačec, M. (2019). 'Modeli dubokog učenja za automatsko sažimanje dokumenta', Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, accessed 27 June 2024, https://urn.nsk.hr/urn:nbn:hr:168:167130

Kukovačec M. Modeli dubokog učenja za automatsko sažimanje dokumenta [Master's thesis]. Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing; 2019 [cited 2024 June 27] Available at: https://urn.nsk.hr/urn:nbn:hr:168:167130

M. Kukovačec, "Modeli dubokog učenja za automatsko sažimanje dokumenta", Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, 2019. Available at: https://urn.nsk.hr/urn:nbn:hr:168:167130

Metadata
TitleModeli dubokog učenja za automatsko sažimanje dokumenta
Title (english)Deep Learning Models for Automated Document Summarization
AuthorMarin Kukovačec
MentorJan Šnajder (mentor)
Committee memberJan Šnajder (predsjednik povjerenstva)
Committee memberVlado Glavinić (član povjerenstva)
Committee memberSiniša Šegvić (član povjerenstva)
GranterUniversity of Zagreb
Faculty of Electrical Engineering and Computing
Zagreb
Defense date and country2019-07-10, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Abstract
Automatsko sažimanje dokumenta jedan je od važnijih zadataka u obradi prirodnoga jezika. Dobar postupak sažimanja u obzir uzima dva aspekta teksta: (1) njegovu informacijsku strukturu, tako da zadržava informacijski relevantne i ispušta manje relevantne dijelove, te (2) koherentnost odnosno kohezivnost teksta. Dva glavna pristupa automatskom sažimanju su ekstraktivno sažimanje, kod kojega se koriste dijelovi izvornoga teksta, te apstraktivno sažimanje, kod kojega se generira novi tekst na temelju izvornoga teksta. Razvojem dubokog učenja došlo je do značajnog napretka u području sažimanja dokumenata, posebice u apstraktivnim pristupima.Tema rada jesu modeli za automatsko sažimanje dokumenta temeljeni na dubokom učenju. Napraviti pregled postojećih metoda ekstraktivnog i apstraktivnog sažimanja, s naglaskom na postupke temeljene na dubokom učenju. Razviti i implementirati modele dubokog učenja za ekstraktivno i apstraktivno sažimanje za engleski jezik. Za razvoj i vrednovanje modela koristiti javno dostupne skupove podataka, kao što su DUC i CNN/Daily Mail. Provesti eksperimentalno vrednovanje modela, usporedbu s postojećim modelima i statističku analizu rezultata. Radu priložiti izvorni i izvršni kod razvijenog sustava, skupove podataka i programsku dokumentaciju te citirati korištenu literaturu.
Abstract (english)
Automated document summarization is one of the more prominent natural language processing tasks. A good summarization method will consider two aspects of text: (1) it's informational structure, so that it retains the informationally relevant and leaves out less informationally relevant parts, and (2) text coherence and cohesion. Two main approaches to automated summarization are extractive summarization, which reuses parts of the original text, and abstractive summarization, which generates new text based on the existing text. Recent developments in deep learning have brought improvements to document summarization, especially for abstractive approaches. The topic of the thesis are models for automated document summarization based on deep learning. Do a literature survey on extractive and abstractive summarization, with an emphasis on deep learning methods. Devise and implement a deep learning model for extractive and abstractive summarization for the English language. Model development and evaluation should rely on freely available datasets, such as DUC and CNN/Daily Mail. Carry out an experimental evaluation of the model, a comparison against the existing models, and a statistical analysis of the results. All references must be cited, and all source code, documentation, executables, and datasets must be provided with the thesis.
Keywords
Keywords (english)
Languageenglish
URN:NBNurn:nbn:hr:168:167130
Study programmeTitle: Computing
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
Type of resourceText
File originBorn digital
Access conditionsClosed access
Terms of useLicence In copyright icon
Public note
Created on2020-01-07 19:32:18