Sentiment Analysis Using Natural Language Processing
Sentiment Analysis Using Natural Language Processing
We will be developing a sentiment analysis project. Assume that we have an application that receives customer reviews at regular intervals. In this project, we’ll build a pipeline to read the reviews and extract the sentiment of the reviews. We will be using a sample dataset from the IMDB Review dataset.
Disclaimer: Educative is using this dataset as it originally exists in IMDB. We are not altering or changing any information within it.
The dataset contains two columns: review and sentiment. Sentiment can be positive/negative. A sample row in the dataset is shown below:
We will use TF-IDF, supervised learning techniques to predict the sentiment of the long text. In this project, we will build a pipeline and schedule it on a regular basis to handle the daily review streams. This analysis is needed for offline analysis, hence deployment is not needed.
We will create a review pipeline with a low-cost solution. We will use Azure for compute and storing the datasets/model. The steps include:
Create a training code for building a model to predict the review.
Build a pipeline to run the trained model against the daily reviews.
Schedule the pipeline.