Machine Learning Based Aspect Category Detection for Hindi Data Using TF-IDF and Count Vectorization

Sentiment is related to emotions. The emergence of e-commerce has initiated a novel trend in the realm of natural language processing, mainly through sentiment analysis applied to user-generated reviews. Hindi, as an Indian language, is utilized by many users. With the remarkable surge in online product reviews in Hindi, there has also been a rapid increase in post-web Hindi reviews. Aspect-based sentiment analysis comprises aspect category detection (ACD) and addresses the challenge of assigning a specific review sentence to a predefined set of aspect categories. In this task, it's important that a single review sentence can potentially be associated with one or more of these predefined categories. In this paper, machine learning-based transformation approaches for aspect category detection are examined on a well-accepted Hindi data set. This is based on a multi-label classification paradigm. Term-frequency / Inverse-Document Frequency (TF-IDF) and count vectorization algorithms are used for feature extraction. A range of evaluation parameters, including as precision, recall, F1-score, and Hamming Loss, are included in the results. In every case, the logistic regression (LR) classifier consistently generated the most pertinent and relevant results.