Are you interested in Azure, Bayesian Theorem, text analytics or sentiment analysis?
I wrote a guest piece on the Nigel Frank International blog a few weeks ago that covers Bayesian Theorem and shows how it can be used to perform sentiment analysis.
I remember getting my head around this back in 2013 as part of a research project I was working on to help classify Twitter data.
It was quite a lot of work! I had to:
- get my head around Bayesian Theorem
- create a data model
- source training data
- cleanse and pre-process training data
- label training data
- build an API in C# to process the data
- iterating through various tests to improve the accuracy (the highest classification accuracy I hit was between 70-80%)
I ended up putting a web application in front of the C# API I built and submitted it to Twitter for an initiative I heard about:
(read about my submissions for 2016 and 2017)
Shortly after I had completed this, I started to notice Azure Cognitive Services Text Analytics and how a lot of this was made available out of the box!
In the end, I started to swap out my custom API and use the Azure Text Analytics API. It saved me the headache of managing a data model, dealing with bug fixes, and sourcing training data and so on.
Anyway, that’s enough about the background!
In the guest blog I also:
- introduce a free Bayesian Classifier called “nBayes”
- talk about training data
- run through some examples
- discuss use cases
You can read the guest blog here.
Leave a Reply