The impact of semantics on aspect level opinion mining

Area of Science:

• Natural Language Processing
• Computational Linguistics
• Information Retrieval

Background:

• Online shopping generates vast amounts of customer feedback.
• Analyzing product reviews is crucial for both consumers and sellers.
• Existing opinion mining methods struggle with the complexities of online comments, such as negation and diverse product aspects.

Purpose of the Study:

• To propose a novel semantic-based aspect level opinion mining (SALOM) model.
• To enhance the accuracy of sentiment analysis in online product reviews.
• To address challenges posed by negation words and varied product aspect representations.

Main Methods:

• Developed a SALOM model leveraging semantic similarity for aspect extraction.
• Incorporated handling of negation words within the sentiment classification process.
• Accounted for synonyms, hyponyms, and hypernyms to represent product aspects comprehensively.

Main Results:

• Evaluated the SALOM model on three distinct datasets.
• Achieved high performance metrics: 94.8% precision, 93% recall, and 92.6% F-measure.
• Demonstrated the model's effectiveness in classifying sentiment accurately.

Conclusions:

• The proposed SALOM model significantly improves opinion mining performance.
• Semantic analysis and comprehensive aspect handling are key to accurate sentiment classification.
• The SALOM model offers a robust solution for analyzing online customer feedback.