Power BI Cognitive Services

In part 1 of our Artificial Intelligence in Power BI article, we covered the use of AI in visualisations, which provide advanced analytics within the front end of a report. This is part 2 of our AI in Power BI series which covers the utilisation of Cognitive Services in Power BI.

What is Power BI Cognitive Services, and what happens when AI meets Power BI?

Cognitive Services is an Azure service provided in the cloud for developers to build intelligent applications without being trained in AI or data science. If we focus on what this means for Power BI developers, then we could conclude that Cognitive Services is intended to enrich Power BI insights with the added intelligence of AI.

There are currently 4 services supported in Cognitive Services for Power BI:

  • Sentiment Analysis
  • Key Phrase Extraction
  • Language Detection
  • Image Tagging

Let’s look at what this means for our data.

Sentiment Analysis

The sentiment function evaluates qualitative analysis such as reviews or customer satisfaction surveys and returns a sentiment score for each item. A sentiment score is a numerical representation of positive or negative ‘feeling’ with 0 being negative and 1 being positive.

Sentiment Analysis score

Many organisations gather ‘opinion’ data from their customers, however, without the added intelligence of sentiment score, which enables grouping of reviews into positive, neutral and negative, these can be extremely time-consuming and manual to work through. AI in Power BI allows customer experience teams to very quickly gain insight into how customers feel and begin to cross-reference this textual information against other data within the organisation.

Key Phrase Extraction

When analysing qualitative data, there is a lot of ‘noise’ captured. This means that some reviews might be inflated with conjuncture or flowery language that doesn’t add to the analysis of the review itself. This could have a negative impact on the accuracy of the analysis completed. This is where ‘Key Phrase Extraction’ becomes useful. By utilising this function, our review or customers feedback data can be stripped back to its key phrases, removing stop words, conjecture or irrelevant detail.

By running the same reviews through this function, we get a targeted, useful list of key phrases as below:

These key phrases will have further benefit when visualising our qualitative data. For example, a word cloud would be a great way of visualising the key phrases being used by our customers, such as in the example below.

Language Detection

Cognitive Services can be used to assess qualitative data in various languages. This can be very useful when capturing customer reviews from multiple regions and allows comparative analysis to be completed in one step rather than having to silo the data and report on it separately – or not at all depending on the skills of the users carry out the analysis. Cognitive Services recognises 120 languages.

Image Recognition

This function can tag images ranging from recognisable objects, people, scenery and actions. This is a way to automate the process of categorising image stores that your organisation may have. A collection of image tags forms the basis for an image ‘description’ displayed as human-readable language formatted in complete sentences. This is another function that takes out the arduous manual tasks of categorisation that many organisations go through.

We hope this has helped to explain how Power BI Cognitive Services works. Part 3 of the series will be AI in the Power BI Service.

If you would like to find out more about Power BI visit our Power BI page. For more information on our AI and data science visit our Data Science page.