The project focuses on utilizing NLP techniques to analyze customer reviews of video game.
This project will look at what people say in their reviews about video games on Amazon. I'll use NLP methods to understand their opinions. My tasks include:
Getting the Amazon reviews dataset. Making my own dataset from these reviews. Deciding if people like or dislike the games and giving each review a score between -1 and 1. Checking how well my score matches the review ratings. Trying different ways to analyze sentiments (positive, negative, or neutral). Sharing my findings with the team, listing what gamers like and dislike about video games.
Here's a quick overview of what will be done and the techniques that will be used:
Balancing Data: I'll handle uneven datasets using the imbalanced-learn package.
Sentiment Analysis: I'll figure out the sentiment (like or dislike) of the reviews using NLTK, a toolkit for natural language processing in Python.
Assessing Accuracy: I'll check how well our method is working by evaluating the data with scikit-learn in Python.
Deep Learning: I'll dig deep into reviews using an advanced technique called DistilBERT. To make this work, we'll use Pytorch, transformers, and simpletransformers packages.
Model Evaluation: I'll assess our model and create useful statistics with the scikit-learn library in Python before sharing the results with our boss.
Visualization: Lastly, I'll make the preferred and non-preferred words related to video games more understandable by visualizing them using Altair.