The new Find Data platform aims to streamline access to vital land and water information for a diverse range of users. A 'minimum viable product' is now open to registered users for beta testing, with further development planned.
Over 12,000 publicly available land and water datasets can now be interrogated via an AI chatbot, which can recommend datasets relevant to a user's enquiry.
The Find Data platform was developed to provide fast and efficient access to land and water data by the Kuaha Matihiko: Digital Gateway project, a collaboration between Waka Digital, Massey University, and AgResearch.
The interface was designed to meet the needs of diverse users, in particular environmental consultants, Māori environmental groups, policy-makers, and researchers. The team took a bottom-up approach to actively engage with potential users of the platform, through co-design wānanga, surveys, and case-studies.
The AI component helps meet the different data requirements of these diverse data-users.
Freely available datasets from www.data.govt.nz are the initial data resource for the platform, with its 12,000 datasets meeting high quality standards. Most Our Land and Water research outputs are also included as a data resource.
In the medium-term future there is potential to upload over 100,000 datasets.
The platform offers two search engines. A traditional search engine was developed as the user interface, which sits in front of a local optimised closed-vector database that hosts the land and water information. A second AI-enhanced search component allows users to prompt a chatbot for results tailored to their land and data needs.
Over the past six months, people representing the four main user groups (research, kaimahi, consultants, policy makers) tested the alpha version of the platform. Initial testing of the AI chatbot was conducted to train it to provide insightful responses. Usage patterns and data interaction preferences helped refine the AI's understanding and response capabilities and allowed the development team to understand how prompts and AI responses can be fine-tuned to deliver the required information.
Feedback from individual testers was collected about user experience and functionality of the AI chatbot’s interaction with the data, which informed iterative improvements to the Find Data tool.
Additional user testers are now being sought to provide feedback on the beta version of the Find Data MVP prototype. Registered users will be able to actively use the tool and provide feedback. The tool is being updated frequently to enhance connectivity between data and the AI chatbot, and implement user feedback to improve usability and functionality.
If you are interested in testing the beta version of the tool, please register on the Find Data website.
More information:
Author