Artificial Intelligence for data enhancement, linking and exploration

Context

We’re working with recent advances in artificial intelligence to:

• help link disparate datasets together
• to delve within them to make data discoverable (lots of information is currently very hard to locate)
• to make the process of moving from data to knowledge easier
• to have fun and explore new ways we can engage with our amazing shared heritage – discovering new ways of linking, viewing and interacting

Aim

This project aims to help us overcome some of the critical challenges we face as we continue our transition into the digital age.

Over the last hundred years archaeologists, museums, heritage professionals, and members of the public have generated vast amounts of data. This spans the period from paper records, to early use of digital records to current ‘born digital’ approaches. This means we have built an archive of data in different formats, working on different systems, which aren’t always very accessible.

Recent advances in artificial intelligence offer us ways through which we can begin to bridge these gaps, reconnecting the disconnected, and delving deeper into records that previously have been hard to search or engage with.

The scale of this challenge is immense and provides the focus for researchers across the globe. Here, we are focusing on resolving key issues as they face the culture and heritage sector concerning marine and maritime data. This will allow us to test ideas out and see what works. We are beginning with known issues around legacy data, then will move out into more exploratory and challenging areas.

Outputs

We have been using our Artificial Intelligence/Machine Learning (AI/ML) tools to augment datasets received from our partners and shared them with our co-investigator Archaeology Data Service for inclusion in the Unpath’d Waters Portal. The codes for our various tools will be published via GitHub. We are currently also looking at confidence in the maritime record and will be sharing our findings in due course.