Nguyen Thi Thu An
Graduation School of Science and Technology, Keio University, Yokohama, Japan
Masafumi Hagiwara
Graduation School of Science and Technology, Keio University, Yokohama, Japan
Download article
Published in: KEER2014. Proceedings of the 5th Kanesi Engineering and Emotion Research; International Conference; Linköping; Sweden; June 11-13
Linköping Electronic Conference Proceedings 100:102, p. 1219-1234
Published: 2014-06-11
ISBN: 978-91-7519-276-5
ISSN: 1650-3686 (print), 1650-3740 (online)
This paper proposes a new method to estimate impression of short sentences considering adjectives. In the proposed system; first; an input sentence is analyzed and preprocessed to obtain keywords. Next; adjectives are taken out from the data which is queried from Google N-gram corpus using keywords-based templates. The semantic similarity scores between the keywords and adjectives are then computed by combining several computational measurements such as Jaccard coefficient; Dice coefficient; Overlap coefficient; and Pointwise mutual information. In the next step; the library sentiment of patterns.en - natural language processing toolkit is utilized to check the sentiment polarity (positive or negative) of adjectives and sentences. Finally; adjectives are ranked and top na adjectives (in this paper na is 5) are chosen according to the estimated values. We carried out subjective experiments and obtained fairly good results. For example; when the input sentence is “It is snowy”; selected adjectives and their scores are: white (0.70); light (0.49); cold (0.43); solid (0.38) and scenic (0.37).
Impression; polarity; relatedness; semantic similarity.
