Aspect-based sentiment analysis (ABSA) is a text analysis technique that categorizes data by aspect and identifies the sentiment attributed to each one. Aspect-based sentiment analysis identifies the aspects of a given target entity and the sentiment expressed toward each aspect. Aspect categories are (e.g., food, Sound, Quality, Storage, Performance etc.) For Example: Aspect-based sentiment analysis would analyze the comment: “Its performance is reliable,” as a positive opinion about the feature/Aspect, “Performance.” ** Existing Systems ** The Existing systems do recommendations on star base which means we don’t have any idea on which bases customer gives specific stars to product So, In star rating we can't be able to find out the reason behind the bad rating or good rating but in aspect based sentiment analysis, we can have enough information about the rating of a product. In star based rating seller do not know which specification of product is like or dislike by customer.

** Problem Statement ** In a world, when the popularity of e-commerce is increasing rapidly, it is very difficult for the customers to find out a best product according to their needs. So, in this scenario they need to visit the website to know about the reviews about the product, they are interested in because the consumers have become increasingly keen to express their views and emotions on the Internet. But the problem is, it is very difficult to read thousands of reviews one by one because it is very time consuming and energy consuming. Another drawback in star base rating is sometimes by accidental touch, an unwanted rating is done and is saved in database. That rating cannot be undone or change.

** Goals and Objectives!** Our Main Goal is to propose comparative study of aspect based sentiment analysis to do product recommendation. To utilize various preprocessing steps available for text preprocessing. To use Feature Extraction method TF-IDF and utilize existing libraries to dataset for supervised approach. To use many classification methods to classify the reviews to aspects and sentiments. To avoid waste of time and energy in reading and analyzing different comments. To promote business efficiency and online shopping. To avoid fraud and scams.

** Result **

** Conclusions **

A comparative study was done with all existing techniques and methods in text processing field. Feature extraction method TF-IDF is used and performed best in dataset. Four classification methods were product which will be helpful for business to compared Naïve bayes and SVM performed the best with average accuracies 80%. Sentiment analysis of aspects was displayed with the have insights and improve and also best helpful for customer before buying something. It will display the percentage of positive, negative and neutral reviews.

Thank You.