AI-generated from publicly available materials.Elsevier and LG AI Research have launched advanced image search capabilities aimed at enhancing the discovery of visual chemistry within patents and scientific literature.
The collaboration leverages LG AI Research's chemistry-specific AI vision technology to improve the extraction and curation processes in Elsevier's Reaxys platform, a tool designed for chemistry discovery. This technology enables the rapid and accurate capture of substance information from images, which is crucial since much of the relevant chemical data is presented visually rather than textually. The ability to accurately interpret complex structures is particularly vital for applications such as novelty searching and synthesis planning.
LG AI Research's innovative model, known as MolMole, integrates multiple functionalities including molecule detection and optical chemical structure recognition. This model has demonstrated superior performance in extracting chemical information from full document pages compared to existing alternatives. The team at LG AI Research is also working on establishing a benchmark dataset to further enhance the evaluation of such models in practical settings.
Elsevier's integration of this technology ensures that the extraction processes are both accurate and scalable, allowing chemists to focus more on discovery rather than deciphering figures. The partnership aims to convert raw visual data into structured knowledge, thereby making chemistry data more accessible and actionable. Future developments will include enhancing reaction extraction capabilities, which will broaden the available chemical evidence in Reaxys, further enriching the research landscape.