Stories are one of the most important human communication structures and storytelling has always been a means of passing wisdom and information to other generations. They collectively help us make sense of our past and reason about the future. It is thus not surprising that Analysts who analyze data to gain business insights are dependent on data storytelling techniques to communicate their findings to decision makers. Nowadays, organizations rely on BI&A applications such as Tableau or PowerBI to transform and visualize their data. But despite the growing use of data visualization tools, it is still difficult for business users to interpret the content of those dashboards. Data Analysis must go beyond presenting the data results and instead, it must tell a story that can create a connection between the audience and the topic. Data itself does not generate value, it must be properly communicated with the audience. The aim of this thesis is to design and implement a systems that creates data stories automatically.

The overall goal of this thesis is to establish a design science research project to design, build and evaluate an automatic data storytelling system that creates compelling data stories based on various input data. For the development and testing of the system, business data sets can be provided. 

Design and implementation of an adaptive data storytelling system

Applying and adjusting a natural language generation model (e.g. SimpleNLG or RoseaNLG) to data

Evaluate your implementation - potentially a focus group discussion with business users can be arranged

Good programming skills in Python or Java

Experience in a frontend web development framework (e.g., Angular, Vue or React)

First experience in Natural Language Generation (NLG) and interest in coverting data into textual representation

Fluent in English (as the thesis has to be written in English)

Shi et al. (2020) - Calliope: Automatic Visual Data Story Generation from a Spreadsheet

Elias et al. (2013) - Storytelling in Visual Analytics Tools for Business Intelligence

Boldosova (2020) - Telling stories that sell: The role of storytelling and big data analytics in smart service sales

Lan et al. (2021) - Understanding Narrative Linearity for Telling Expressive Time-Oriented Stories

Shen et al. (2021) - Towards Natural Language Interfaces for Data Visualization: A Survey

If you have questions before, feel free to contact me anytime. If you are interested, leave me an email with a short motivation statement, your CV and your current transcript of records.

jonas.gunklach∂kit.edu