Sentiment Analysis using Power BI

Author by Concurrency Blog

Sentiment analytics can now be performed using Power Query and Azure Marketplace. In this blog, I am going to walk you through a very simple process of creating a function that can give you a sentiment score for a text. The first step is getting yourself signed up for a lexicon based sentiment analysis dictionary on azure marketplace. This is the link for https://datamarket.azure.com/dataset/aml_labs/lexicon_based_sentiment_analysis

Once you complete the free signup process, you will be able to carry out sentiment analysis on a text using Power Query and get a sentiment score. The manner in which the sentiment score is assigned is interesting. A fully positive sentiment in the text will be scored as 1, an absolute negative sentiment will be scored as -1 and a neutral sentiment will be assigned a zero value.

Now, browse to the OTHER RESOURCES tab in Power Query and select AZURE MARKETPLACE. You will then see a NAVIGATOR window fire up with lexicon based sentiment analysis API dictionary. When you double click the lexicon based sentiment analysis icon, a function called (fx) will fire up.sentiment1

sentiment2

Now, click the function (fx) which will open up a textbox where you can enter any text. I have shown two examples where I have tried posting a sentence which conveys a positive sentiment and then another one which conveys a negative sentiment.sentiment3

sentiment4sentiment5

sentiment6

You can easily see for yourself from the results that a positive sentiment gets a score of one whereas the negative sentence gets a value of negative one. I have also highlighted the function "SCORE1" which we can use to get the scores for a larger dataset. This is just the beginning and we will try to understand the further details on sentiment analysis using PowerBI in the upcoming blogs.

Author

Concurrency Blog

The latest about Concurrency