Many break-through technologies require new polymers, however the time to develop, optimize and certify the new materials usually takes much longer then to develop the new technology. At the same time, an exorbitant amount of scientific studies get published each year, making it impossible for any given researcher to stay on top of all publications in their field. Knowledge of structure-property relationships of polymers can accelerate the discovery of new materials. However, it is difficult to extract chemistry-knowledge based information, because of the expressions are different for each scientist. Therefore, it is beneficial to use NLP process pipeline to understand them and extract the data.<br/>A first step in the process is to recognize chemical entities and their abbreviations. We then use table extractor to get all the information from tables (structured data) and rule based syntax to get the information from the text(non-structed data). We use our system then to normalize the data then associate them to build a structured searchable dataset. Based on chemical dictionary, table extractor and rule based NLP extract tool (SystemT), we can extract chemical entities and their associated properties, techniques, reactions, roles and relationships from scientific documents, reports, and patents. Our system is an automated, chemistry knowledge based cognitive data extraction and relation build tool.<br/> <br/>In this presentation, we will introduce a polymer data extractor based on IBM’s SystemT. We will show the process how the data is been extracted and associated together.

Introducing automated polymer data extractor tool

2019 Fall National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.