Taxonomy - A Multi-Layered Approach to Successful Classification

Posted: 16 Dec 2019

Date Written: December 16, 2019

Abstract

In this work, we explore how the quality of a classification product does not depend on a single indexing tool or classifier, but a combination of many tools and approaches. Efforts and improvements must be made at two levels: on the indexing level, a diverse set of rule-based and machine learning tools help increase precision/recall and improve overall F measure; rarely will a single tool be able to overcome all the various issues encountered in classification (getting the context right, classifying documents of various lengths, improving precision without impacting recall and vice-versa, etc.). Once an optimal threshold of precision and recall has been attained, then having a relevancy sorting mechanism is essential to distinguish between results that are substantially about the topic versus results that only deal marginally with it. The combination of these tools and approaches are necessary to ensure that taxonomy search tools produce results that are on point and helpful to customers.

Keywords: taxonomy, classification, indexing, topic search

Suggested Citation

Lagace, Sophie, Taxonomy - A Multi-Layered Approach to Successful Classification (December 16, 2019). Proceedings of the 3rd Annual RELX Search Summit, Available at SSRN: https://ssrn.com/abstract=3504639

Sophie Lagace (Contact Author)

LexisNexis ( email )

P. O. Box 933
Dayton, OH 45401
United States

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Abstract Views
136
PlumX Metrics