Aspect Based Sentiments Analysis of Locations using Deep Learning Based Natural Language Processing

Abstract

With the growth of web based social media networks, lots of reviews are written by people about different place, facilities, and business etc. Such reviews or Information about the location and its aspects can be systematically derived from the text written about the locations. The sentiments can be analyzed for each aspect to determine good or bad about the locations. Aspect based sentiment analysis using Deep Learning based Natural Language Processing (NLP) technique was used in this research to effectively extracts aspects from the reviews data and analyze the sentiments attached with each aspect. Since the reviews written about different aspects of a location are subjects to individuals’ likings, personality, and several other associated factors. This research integrated user profiling, aspects-based sentiments, and preferences using Analytic Hierarchy Process (AHP) to achieve more relevant, and preference- based results. AHP provided preference-based aspects scores i.e., positive, negative, and net sentiment scores based on the characteristics of locations for different preferences. AHP provided preference-based aspect scores such as positive, negative, and net sentiment scores are assigned scores, negative score, and. The reviews about the location depend and vary based on different demographics, liking and disliking. Thus, the reviewer’s segmentation was done based on gender, ethnicity, and nationality to demonstrate that segmentation-based analysis and outputs are more relevant and appropriate. The output varied when priority of the input criteria was changed. Aspect extraction and sentiment analysis using deep learning-based NLP produced promising results and appears as a reliable way to extract useful information about the locations. Reviewer segmentation-based analysis produced more relevant results due to the different traits and sentiments of each profile.