Aladdin: Using natural language to facilitate open data visualisation

Detailed pipeline of the Aladdin solution

This dissertation explores the impact of using natural language in the visualisation of open datasets, focusing on the design and evaluation of Aladdin, a system based on the DSR approach. Aladdin uses advanced natural language processing techniques to transform text queries into interactive data visualisations.

The methodological approach details the development of Aladdin through three main DSR cycles: the Rigor Cycle for the theoretical analysis and definition of objectives, the Design Cycle for the design and implementation of the solution, and the Relevance Cycle for evaluation in real-life conditions. The system is based on React for the user interface, MongoDB for data management and FastAPI for a robust API, with the GPT-4 model for the semantic analysis of queries.

The evaluation of Aladdin confirms its role in improving the accessibility and efficiency of open data visualisation, with positive feedback on the accuracy of interpretations and the quality of visualisations. However, the management of complex queries needs to be strengthened, and the shortcomings of the LLM support need to be remedied. Future prospects include integrating with other analytical platforms and improving user interaction, while extending Aladdin’s accessibility via OpenData portals to increase its functionality and adoption.

Related