Description
Data visualizations are not always sufficient in understanding important aspects of data. People have to interpret charts and build narratives about it. Natural language generation technology can be used to greatly improve the communicative power of the data. In this talk I will introduce NLG and will present a case through the use of two open source tools: Accelerate Text (NLG) and Metabase (BI)
BI tools excel at visually organizing data in various dashboards. Yet data visualizations are not always sufficient in understanding important aspects of data. People have to interpret charts and build narratives about data on their own. Furthermore, different groups of people often need different narratives about the same data. For example, a story about the same sales data is different for a CFO and for an Order Fulfillment Manager.
Natural language generation technology (NLG) can be used to greatly improve the communicative power of a BI dashboards. With the use of NLG we can produce different descriptions of the same data adopted to the needs of each person reading it.
The integration between BI and NLG will be demonstrated through the use of two open source products: (a) TokenMill's open source NLG tool called Accelerated Text (https://github.com/tokenmill/acceleratedtext), to illustrate how natural language can be generated using data; (b) and the BI platform Metabase (https://github.com/metabase/metabase), to provide analytical dashboards which will include both data visualizations and automatically generated natural language.