| Summary: | The consumption and production of online reviews have become an integral part of our modern lives as the Internet takes over as the main source of information. Unfortunately the vast volume of information cannot simply be digested by traditional means – reading, and additional tools will be required to select better ones that are worth reading. This solution should also help in processing highly unstructured texts with subjective opinions.
The development of Natural Language Processing has made it possible to for machines to process through large amounts of textual data and find out important points. This project makes use of text mining and machine learning methods to categorize texts into structured information and then derive a new rating by evaluating the sentiment of the author. This standardized rating also aims to reduce human bias inherent within written reviews.
This project focuses on food review data within a local setting, Singapore, which was scraped from a website. Aside from constructing a new rating system, we are interested in finding out local preferences from their reviews.
Due to a limited length of word lists fed into the system, it was not able to sufficiently recognize and evaluate one particular aspect. Aside from that aspect, the system demonstrated that it performed well in narrowing down customer choices and produced normally distributed ratings.
|
|---|