Naslov
Rule-Based Sentiment Analysis in Narrow Domain: Detecting Sentiment in Daily Horoscopes Using Sentiscope

Autori
Agić, Željko ; Merkler, Danijela

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
Proceedings of the 2nd Workshop on Sentiment Analysis where AI meets Psychology / Bandyopadhyay, Sivaji ; Okumura, Manabu - Mumbai : The COLING 2012 Organizing Committee, 2012, 115-124

Skup
2nd Workshop on Sentiment Analysis where AI meets Psychology (SAAIP 2012)

Mjesto i datum
Mumbai, Indija, 15.12.2012

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
sentiment analysis; narrow domain; rule-based system

Sažetak
We present a prototype system --- named Sentiscope --- for collecting daily horoscopes from online news portals written in Croatian, detecting polarity phrases and overall sentiment conveyed by these texts and providing sentiment-analysis-based visualizations in a graphical user interface on the web. The system was evaluated using a dataset of daily horoscopes which was manually annotated for (positive and negative) polarity phrases and (positive, negative and neutral) overall sentiment. Linearly weighted kappa coefficient of 0.593 has indicated moderate inter-annotator agreement on overall sentiment annotation. The system achieved an F1-score of 0.566 on overall sentiment and 0.402 on phrase detection. An overview of implementation is provided --- with special emphasis on the polarity phrase detection module implemented in NooJ linguistic IDE --- and the system is made available to users on the web.

Izvorni jezik
Engleski

Znanstvena područja
Informacijske i komunikacijske znanosti

POVEZANOST RADA