Picture this: Your end user is looking at a widget that breaks down the average age range of their customers. Trying to parse through the information, they think they’ve figured out what’s interesting about the data. But, they need to do some calculations to be sure. Time to whip out their calculator.
…not so fast.
Introducing: Sisense Narratives.
Here at Sisense, we’re always looking for innovative ways to allow users to interact with their data in a more intuitive way. With Sisense Narratives, we use natural language generation (NLG) to automatically present you with calculations and insights in plain, easy to understand language based on what the engine recognizes as interesting. No calculator required.
What is Natural Language Generation?
Natural language generation (NLG) is the natural language processing (NLP) task of creating instinctive, original language based on data or logic pool. Combined with Sisense, NLG does the heavy lifting for you by providing easy to understand, digestible insights from your data with the click of a button.
How does Sisense Narratives work?
You might find that some end users are less comfortable with charts and graphs, believing they are meant for data scientists and analysts. Sisense Narratives makes data more approachable by providing a consumable, textual explanation of the data. You can use this feature alongside your visualization or instead of your visualization, whichever you think will make the most impact on your end users.
Learn to make your insights shine with our on-demand webinar “Telling a Story Through Data: Dashboard Design Guidelines” now.
Narrative text isn’t limited to just your dashboards. It’s also included when you generate a PDF report of your dashboard to be shared with others.
Let’s take a look at how it works.
If we take the example from above, you can expect to see the age breakdown widget look something like this without Sisense Narratives:
But let’s say you want to enhance it with natural language text so that your end users can instantly understand calculations and insights. Here’s how you can make that happen:
The narrative text is generated by a cloud service, with all communication authenticated and encrypted. Once activated, a Narrative API token that is used to secure the HTTPS communication between Sisense and the narrative cloud service is generated. Your widget’s data is then sent to the cloud service, which generates the narrative.
From there, Designers can enable narration for widgets! It’s that simple. The narration is enabled on the widget level, meaning you select which widgets display narration, and which widgets don’t.
But that’s not where it ends. Let’s take a look at all of your customization options:
Select the position of your narrative. You can display the narration above or below your visualization, or display just the narrative.
2. Singular vs. Plural
Define how the dimension names appear in the singular and plural format in the narrated text. The default name of the dimension is taken from the label of the field that is added to the widget.
You might find that bullet points are easier for your end users to understand. Or, they might prefer to read a paragraph of text that explains to them the important information for a specific widget. Select the format of the narrative as paragraphs or bullets, test it with your end users, and switch depending on their responses with a single click of the mouse.
Select the aggregation type to define how data totals and summaries are calculated in your narrated text. For example, if you select “Sum”, Sisense calculates the totals of your data and customizes the text to fit this scenario. If you select “Average”, Sisense calculates the average and customizes the text accordingly.
Do you want to give your end users a lot of description or just cut straight to the numbers? Select how descriptive the narrated text is – either high, medium, or low.
Select the sentiment for your narrated text depending on the trends of your data. The sentiment determines if the narrated text uses positive or negative tone, based on the actual data. For example, sales figures can be narrated positively as they increase but negatively when they decrease. On the other hand, a churn visualization can be narrated negatively when it increases, but positively when it decreases.
How’s that for telling a story with data?