VIDEO: How to leverage natural language in #WatsonAnalytics

Blog Home > VIDEO: How to leverage natural language in #WatsonAnalytics

VIDEO: How to leverage natural language in #WatsonAnalytics

Here’s a video from IBM Analytics Learning Services that shows  you how to leverage natural language in your analysis.

These videos from IBM Analytics Services are the result of gathering a group of likeminded IBMers together to create some videos with the intent of solving common problems.


You have been using Watson Analytics to perform analysis using personal and corporate data sources.

You and your team have uploaded multiple data sets which are available in your Personal folder and Shared folder locations. Because there are so many data sets to potentially navigate, you would simply like to pose natural language questions to help you begin your analysis. To do this, on the Data tab of Watson Analytics, type a question in the Enter a question about your data box, and then press Enter. Watson Analytics provides you with a list of relevant starting point Discoveries.

You can see that they span multiple data sets in your environment. You can also type a natural language question while on the Discover tab, and again have Watson Analytics assess the available data sets, and return relevant starting points. After entering and submitting your natural language question on either the Data or Discovery tabs, you can choose one of the starting points to begin your analysis. The Discovery and Discovery set opens, and is based on the data set on which the starting point was derived. Any additional Discoveries you create are also based on this data set. You can continue to perform analysis using this starting point Discovery, or you can create another Discovery, by submitting another natural language question, while inside a Discovery set. Any question you submit while inside a Discovery set will be assessed against only the data set currently in use. It will not be assessed against all the data sets in your environment. If you are unsure of how to phrase a question, you can get some guidance using the How to ask a question feature. In this view, you can select from a variety of keywords, and data items to help you phrase your question. Keywords are grouped into categories. For example, if your intent is to analyze totals for a given key metric, you could use the Aggregate data category. Another example might be data comparison, in which case you might want to analyze the values of a key metric, in relationship to one or more variables or dimensions. Once you are satisfied with your phrasing options, click Ask. Starting point visualizations appear, which you can open to continue your analysis. To obtain more information on using natural language questions in Watson Analytics, use the online documentation.