Studying Biocuration Workflows

Abstract

As the first phase of a knowledge engineering study of biocuration workflows, we performed a preliminary task-modeling exercise on seven separate bioinformatics systems. This involved constructing UML activity diagrams from detailed interviews with curators in order to understand the organization of the process the biocurators used to populate their system. The objective of this work was to identify common patterns within the workflows where we might apply text mining methods to accelerate curation. We compiled a number of workflows in a common format but were largely unable to consolidate these structures into a formal structure that facilitated comparison across workflows. At present, more work is needed to perform this task.