حسين اله وردي ميگوني

In the modern era, sentiment analysis has emerged as one of the most significant subfields of Natural Language Processing and text mining, gaining increasing importance due to its wide range of applications. The main objective of sentiment analysis is to extract emotional and subjective information from textual data, which is valuable in various domains such as customer feedback analysis, social media monitoring, public opinion tracking, and market trend prediction. Despite remarkable advancements in this field, numerous challenges still hinder the enhancement of model accuracy and reliability. One of the major challenges is the inherent ambiguity and uncertainty in natural language, which affects the performance of sentiment classification systems, particularly when dealing with vague or context-dependent expressions. To address this problem, the present study introduces a novel approach based on the fuzzification of WordNet, a well-known lexical-semantic ontology that organizes words into synonym sets (synsets) and captures semantic relations between them. The proposed method aims to improve the effectiveness of sentiment analysis by modeling the semantic relationships within WordNet using fuzzy logic. In this context, fuzzy membership functions are defined for synsets, enabling a more flexible and gradual representation of meaning, which better handles ambiguity and semantic overlap between words.The research begins with an overview of the foundations of fuzzy logic and the process of applying fuzzification to WordNet. Then, appropriate fuzzy membership functions are designed and integrated into a fuzzy inference system for sentiment classification. The semantic relations within WordNet are encoded as fuzzy rules, and a Mamdani-style fuzzy reasoning mechanism is utilized to infer the final sentiment of a given text. The proposed fuzzy sentiment analysis system is implemented and eva‎luated using benchmark datasets. Comparative experiments with traditional, non-fuzzy sentiment analysis models demonstrate that the fuzzified WordNet approach yields a 2.3% improvement in classification accuracy, particularly in scenarios with high levels of linguistic uncertainty and context-dependence. These results confirm the potential of semantic fuzzification as a promising direction for enhancing sentiment analysis in real-world, noisy text environments.

وردنت توابع عضويت فازي , عدم قطعيت مدل‌هاي تحليل احساسات , قطبيت احساسات , تحليل بر اساس جنبه

WordNet , Fuzzy Membership Functions , Uncertainty Sentiment Analysis Models , Fuzzification , Textual Data Ambiguity , NLP Systems

Hossien Allahverdi

Dr.Behrouz Minaei Bidgoli