The ever-expanding capabilities of large-language models and other machine learning techniques can be leveraged to improve LIVE environments. A few areas of focus are identified below, and more are yet to come.
AI-Enhanced Linking Options
Large Language Models (LLMs) can be effectively used to help users automatically create links between diverse datasets with column names that don't conform to specific metadata standards (Zhang et al. 2023, Lobo et al. 2023) . By encoding the variety of columns naming conventions present in a large amount of real world data, LLMs can effectively predict when columns in different datasets refer to the same entity (a form of Entity Resolution), and provide these proposals to the user for acceptance/verification.
The LIVE team is about to deploy an AI-enhanced linking tool, so "watch this space!"
AI-Enhanced Selection Functions
Currently, many of the LIVE tools allow for brushing & linking, as well as algorithmic selection of data subsets, to enable exploratory data analysis. Many recent and ongoing research projects offer clever, often AI-enahnced, tools for selecting data subsets, and it's a near-term goal of the LIVE project to begin incorporating machine intelligence into selection options. While many display types and tables can and will benefit from smart selection, the LIVE team is especially interested in the ~unsolved problem of intelligent 3D selection withing volumetric data, so stay-tuned--and do be in touch if you have ideas!