Word Segmentation and Sentiment Word Categorization Using Feature Extraction – a Novel Framework

EasyChair Preprint 6042

Abstract

Sentiment analysis (SA) is essential for classifying people's thoughts about whatever they submit as reviews online. Because the content on these media is unstructured, the segmentation of the sentiment word, which is critical for detecting attitudes, must be done properly to overcome the problem of missing data, which can lead to erroneous criticism classifications and render the SA approach useless. This study provides a novel approach for automatically segmenting the sentiment word in order to categorise the sentiment of "reviews." This framework contains a pre-processing technique, feature extraction with characteristics such as Terms of presence and frequency (TPF), Parts of speech (POS), Opinion words and phrases (OWP), and Negations, and word segmentation using the RBDT algorithm. Experiments show that our proposed techniques are successful and efficient in segmenting the words necessary for sentiment classification without incurring data loss, with 92% accuracy and a time complexity of 0.0008 ms. Furthermore, with a time complexity of only 0.0006ms, the classification of sentiment words obtained excellent accuracy of 94%

Keyphrases: Negations, Opinion words and phrases(OWP), Parts of Speech (POS), RBDT algorithm, Segmentation, Sentiment Analysis (SA), Terms of presence and frequency(TPF), data pre-processing, feature extraction

@booklet{EasyChair:6042,
  author    = {S. Ashika Parvin and M. Sumathi},
  title     = {Word Segmentation and Sentiment Word Categorization Using Feature Extraction – a Novel Framework},
  howpublished = {EasyChair Preprint 6042},
  year      = {EasyChair, 2021}}