Aspect-based sentiment Analysis (ABSA) is a technique to extract fine-grained opinions from product reviews, but not enough support has been given to a live recommendation system. In this dissertation, an end-to-end framework is presented to use ABSA for product recommendations in the Cell Phones and Accessories category of the 2018 Amazon Reviews dataset. In this dissertation, a hybrid aspect extraction approach was used to implement the process where a subset of reviews was manually labeled and used to train the machine learning model. Aggregation of the resulting feature‑level sentiment scores of the products was performed to rank the products with overall easy-to-follow recommendations. The lightweight backend for the entire system responds instantly; the web interface is simple so that users can choose the features they need. This research fills the gap between advanced sentiment analysis and personalization of e‑commerce and further propagates it to the space of user-friendly real‑time suggestions.