In this project, we model a classifier to label a given piece of text as spam or not a spam. Also, we created an API for the model, using Flask, the Python micro framework for building web applications.
The model is currently being hosted on herokuapp server - here
The classifier is trained offline with spam and non-spam messages. The trained model is deployed as a service to serve users. We have used the famous SPAM or HAM Dataset by UCI-ML.
The dataset consists of a file naming data.csv which contains one message per line. Each line is composed by two columns: v1 contains the label (ham or spam) and v2 contains the raw text.
We have used Naive Bayes and Count Vectorizer. Later its accuracy has been improved by using SVM Classifier and TF-IDF Vectorizer. It is further improved by using LSTM classifier implemented using Keras API in Python.
Main modules used are - Flask, Pandas, sklearn, numpy, PIL, Keras, etc.