Aspect Extraction of E-Commerce and Marketplace Applications Using Word2vec and Wordnet Path

Abstract

Aspect extraction is an essential element in Aspect-Based Sentiment Analysis (ABSA). Errors in determining aspects of ABSA will result in errors in determining the sentiment of an opinion and the accuracy value of ABSA. This study aims to obtain elements of opinion sentences on using e-commerce applications and marketplaces in Indonesia. Corrections of the statement were sourced from social media Twitter with the keywords "e-commerce" and "marketplace" from August 2020 to January 2022, and a total of 54,244 comments were obtained. Determination of the words that are candidate aspects is selected using POS Tagging for classes of noun singular (NN), noun plural (NNS), proper noun singular (NNP), and proper noun plural (NNPS).