|
1000
|
Abstract/Summary
|
-
Elsevier aims to streamline taxonomy production by creating a shared infrastructure supported by automation. In this presentation we will explain the components of this infrastructure, which include a candidate pool for incoming candidate terms. Here candidates are enriched with tools such as a synonym suggestion classifier, a term categorization classifier, an ambiguity scorer, and a hierarchical relationship suggestor. In the future, we want to move to a domain-based architecture, in which pre-built branches for specific scientific domains are maintained, and the taxonomy compiler, which chooses from these”modules” to create taxonomies for specific products.
|
