Introduction
In this session, you will learn about another method of semantic processing called “Topic Modelling”.
Suppose you are a product manager at Amazon and want to understand the features of a recently released product (say, Amazon Alexa) that customers are discussing in their reviews. Say you are able to identify that 50% of the customers are talking about the hardware 30% are talking about features related to music, and 20% are discussing the packaging of the product. this information would be pretty useful for you.
Similarly, imagine that you have a large corpus of scientific documents(such as research papers) , and you want to build a search engine for this corpus. suppose you can infer that a particular paper talks about ‘topics’ such as diabetes, cardiac arrest and obesity. with this information a topic-specific search will become easier.
This session will cover the following:
- Intuition of topic modelling
- Types of algorithms in topic modelling
- Non-Negative matrix factorisation
- Getting topics for IMDB movie reviews