Semantic Textual Similarity is the task of determining how close two pieces of text are in meaning. It has many applications such as question answering, information retrieval, recommendation systems and so on.
Here is a 1 hour NLP code-along beginners video tutorial on semantic textual similarity. The session covers the task of Automatic Question Answering from FAQs. You can find the code for our tutorial here.
Automatic Question Answering from FAQs involves finding answers to customer questions by retrieving the most relevant FAQ on the website.
Many websites have Frequently Asked Questions(FAQs) to assist users. But searching through the FAQs to find answers you need to a specific query could be tedious. They often end up sending a support email and waiting for a response instead of wading through the FAQs. Automatically answering queries from FAQs leads to good customer experience.
The dataset used is the starter FAQ-set from the MachineLearningInterview.com website. The tutorial helps understand various techniques for semantic textual similarity with various kinds of word embeddings starting with Bag of words to the state of the art BERT embeddings.