2019/05/16 by Sheshera Mysore, Mysore, Sheshera, Zach Jensen +15 · 2 citations
Materials Science · Computer Science · #Machine Learning in Materials Science #Software Engineering Research #Topic Modeling
paper · pdf · doi:10.48550/arxiv.1905.06939
Materials science literature contains millions of materials synthesis\nprocedures described in unstructured natural language text. Large-scale\nanalysis of these synthesis procedures would facilitate deeper scientific\nunderstanding of materials synthesis and enable automated synthesis planning.\nSuch analysis requires extracting structured representations of synthesis\nprocedures from the raw text as a first step. To facilitate the training and\nevaluation of synthesis extraction models, we introduce a dataset of 230\nsynthesis procedures annotated by domain experts with labeled graphs that\nexpress the semantics of the synthesis sentences. The nodes in this graph are\nsynthesis operations and their typed arguments, and labeled edges specify\nrelations between the nodes. We describe this new resource in detail and\nhighlight some specific challenges to annotating scientific text with shallow\nsemantic structure. We make the corpus available to the community to promote\nfurther research and development of scientific information extraction systems.\n