Fig. 3: Comparison of PhotoCatDB and SciFinder-Photocatalysis (the photocatalysis dataset from SciFinder).
From: An artificial intelligence-driven synthesis planning platform (PhotoCat) for photocatalysis
a PhotoCatDB uses common names for photocatalysts, while b SciFinder-Photocatalysis employs lengthy IUPAC names. In this tree-map embedding, each point denotes a reaction, with spatial proximity indicating feature similarity. Clusters correspond to chemically related reactions, while isolated points highlight unique transformations, providing an overview of the dataset’s feature space. a Does not show aggregation of similar reactions as “multiple clusters” as in (b).
