Model za analizu sentimenta u tvitovima na hrvatskome jeziku

master's thesis

• Title (english): Sentiment Analysis in Tweets in Croatian Language
• Author: Dino Dolović
• Mentor: Jan Šnajder (mentor)
• Committee member: Jan Šnajder (predsjednik povjerenstva)
• Committee member: Slobodan Ribarić (član povjerenstva)
• Committee member: Vlado Glavinić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2015-07-14, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Blogovi, recenzije proizvoda, korisnički generirane poruke poput tvitova i ostali pisani tekstovi ogroman su izvor informacija i znanja. Takvi tekstovi i poruke u velikoj većini slučajeva iznose mišljenja i stavove autora o nečemu ili nekome. Agregacijom takvih mišljenja te njihovom automatskom obradom, odnosno analizom, uvelike se može doprinijeti osobama od interesa, najčešće poslovnim entitetima. Analiza sentimenta ili mišljenja zadatak je iz područja obrade prirodnog jezika te je predmet ovog diplomskog rada.
U okviru ovog diplomskog rada proučeni su postupci za analizu sentimenta u mikroblogovima, s naglaskom na metode temeljene na nadziranom strojnom učenju. Provedeno je označavanje određenog skupa podataka, kojeg čine tvitovi koji se odnose na domenu dvaju pjevačkih emisija: The Voice te X Factor Adria. Nadalje, razvijen je model za analizu sentimenta u tvitovima na hrvatskome jeziku temeljen na nadziranom strojnom učenju koji omogućuje klasifikaciju tvita na razini cijele poruke. Provedeno je iscrpno vrednovanje na odgovarajućem skupu podataka, uključujući i analizu značajki. Dobiveni rezultati usporedivi su s rezultatima dobivenim na natjecanju SemEval-2013, uz postignutu F-micro mjeru od 87%, odnosno F-macro od 71%.

Abstract (english)

Blogs, product reviews, user generated content such as tweet messages and other written texts are huge source of information and knowledge. Such texts and messages are full of users' opinions and their attitudes towards something or someone. One can significantly contribute to the entities that were mentioned, which are mostly in business related domains, by aggregating and automatically analysing such opinions and attitudes. Sentiment analysis is a task in natural language processing and is studied in this thesis.
In this research, certain approaches have been studied that consider sentiment analysis in microblogs, focusing on supervised machine learning techniques. A sample of data was annotated, which consisted of tweets in Croatian about two popular music shows: The Voice and X Factor Adria. A system based on the supervised machine learning technique was implemented in order to classify the tweet into one possible sentiment class considering the whole tweet message. The implemented system was then evaluated and yielded results that could be measured with those achieved at SemEval-2013 competition, producing the F-micro of 87% and the F-macro of 71%.

• Keywords: Analiza sentimenta; analiza mišljenja; mikroblogovi; Twitter; tvit; nadzirano strojno učenje; hrvatski jezik
• Keywords (english): Sentiment analysis; opinion mining; microblogs; Twitter; tweet; supervised machine learning; Croatian
• Language: croatian
• URN:NBN: urn:nbn:hr:168:817186
• 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 08:26:37