EAPRAST: Extensive Approach for Product Ranking in Aspect-Based Sentiment Analysis using TRIE

Abstract

To assist prospective consumers make educated purchasing choices, we are analyzing and mining data from product reviews on online shopping websites. Two methods exist for extracting aspects. Rule-based and Highest Adjective Count (HAC) algorithms. The aspect ranking will use MAX opinion score method and enhanced SentiWordNet opinion score. SentiWordNet uses a hash map structure to turn keys into tiny values that may be used to index data. Hashing can search, insert, and remove in O(L) time. The disadvantage is that if two keys give the same hashCode value, the hashMap's speed suffers. When HashMap buckets are full, they need to be resized. We replaced it with TRIE, which can insert and locate strings in O(L) time, where L is the word length. TRIE is quicker than Hashing because of its implementation. Here hash function and collision handling is not required (like we do in open addressing and separate chaining). TRIE also allows us to print all words in alphabetical order, which is not feasible using hashing. TRIE can effectively search for prefixes. On the other hand, we offer a method that ranks items based on their similarity in terms of features and price. The Suggested method is applied to three conventional databases such as Amazon, Yelp, and IMDB and the solution provides a more effective and dependable online buying experience.