Big data, little decisions: Tightening the loop between data crunching and human expertise

Author

Bennett, Zack ; L´Heureux, Marc G.

Author_Institution

LexisNexis, Dayton, OH, USA

fYear

2013

fDate

20-24 May 2013

Firstpage

65

Lastpage

66

Abstract

This presentation is a case study examining how LexisNexis uses scaled active learning on the HPCC Systems environment to focus manual topical annotations on critical documents pulled from a large corpus. The active learning system uses natural language processing and machine learning techniques to identify and present “next best” training set candidates to legal editors, combining massive parallel processing with expert human analysis to improve classifier accuracy while minimizing human effort.

Keywords

learning (artificial intelligence); natural language processing; parallel processing; text analysis; HPCC systems environment; LexisNexis; classifier accuracy; critical documents; data crunching; expert human analysis; machine learning techniques; massive parallel processing; natural language processing; scaled active learning system; text classification; Classification algorithms; Data handling; Data storage systems; Information management; Parallel processing; Support vector machines; Training; active learning; annotations; large corpus; machine learning; text classification; training set;

fLanguage

English

Publisher

ieee

Conference_Titel

Collaboration Technologies and Systems (CTS), 2013 International Conference on

Conference_Location

San Diego, CA

Print_ISBN

978-1-4673-6403-4

Type

conf

DOI

10.1109/CTS.2013.6567205

Filename

6567205