Automate detection of different sentiments from textual comments and feedback, A machine learning model and Web Application is created to understand the sentiments of the restaurant reviews. The problem is that the review is in a textual form and the model should understand the sentiment of the review and automate a result. The main motive behind this project is to classify whether the given feedback or review in textual context is positive or negative. Reviews can be given to the model and it classifies the review as a negative review or a positive. This shows the satisfaction of the customer or the experience the customer has experienced. The basic approach was trying a different machine learning model and look for the one who is performing better on that data set. The restaurant reviews are very related to the project topic as reviews are made on websites and we can apply this model on such data sets to get the sentiments.
To install dependencies
- go to the directory where requirements.txt is located.
- activate your virtualenv.
- run:
pip install -r requirements.txtin your shell. We have used Python, Django, Html, CSS, Bootstrap, NLTK, and many more things....
There is always a scope of improvement. Here are a few things which can be considered to improve.
Different classifier models can also be tested. The remaining two models can be tuned for better results. For example, after plotting. AUC in Logistic Regression we may get better results. Try a different data set. Sometimes a data set plays a crucial role too. Some other tuning parameters to improve the accuracy of the model.
The motive of the model is to correctly detect the sentiments of the textual reviews or feedback. The developed model has an accuracy of 84.00% and successfully detects the sentiments of the textual reviews or feedback. The model has been tested with few of the online reviews and was found that it detects the sentiments correctly. Thus, can conclude that the motive was successful and the model can be used to detect the sentiments of the reviews and feedback.
Our team has followed Agile Methodology to develop the app. We have completed 82% of SDLC and 90% of Development phase. Further we will work on the deployment after completing rest tasks of development phase.