Fuzzy rules grading and sentiment compound-based dictionary for estimating the degree of sentiment polarity
2025, vol.17 , no.3, pp. 49-60
Article [2025-03-05]
This paper proposes a fuzzy rule-based sentiment analysis framework for airline reviews. A domain-specific sentiment compound dictionary is developed by incorporating intensifiers, negators, and sentence position indicators. Sentences are classified using decision trees constructed from Parts of Speech (PoS) tagged components, resulting in eight distinct sentiment classes. Proposed to handle ambiguity in sentiment class boundaries, Fuzzy membership functions are applied, enabling graded polarity scoring. The model is evaluated on Air India and Twitter US Airline Sentiment datasets and achieved 88% and 91% as accuracy respectively. Comparative results with VADER, GINS and BERT based models confirm the proposed method’s performance is encouraging.
sentiment analysis, sentiment dictionary, airline review, airline tweets, fuzzy functions
https://doi.org/10.59035/TGCH4813
Lijimol George, P. Sumathy, A. Vadivel. Fuzzy rules grading and sentiment compound-based dictionary for estimating the degree of sentiment polarity. International Journal on Information Technologies and Security, vol.17 , no.3, 2025, pp. 49-60. https://doi.org/10.59035/TGCH4813