Towards incorporating derived features in dataset alignment and linking

Blauvelt, Catherine; Weigl, David M.; Downie, J. Stephen; Page, Kevin R.

Permalink

https://hdl.handle.net/2142/96727 Copy

Description

Title

Towards incorporating derived features in dataset alignment and linking

Author(s)

• Blauvelt, Catherine
• Weigl, David M.
• Downie, J. Stephen
• Page, Kevin R.

Issue Date

2017

Keyword(s)

• Linked data
• Entity alignment
• Feature extraction

Abstract

The Semantic Alignment and Linking Tool (SALT) enables scholars and domain experts to establish connections between complementary datasets describing entities such as people, works, or performances, by generating alignment candidates based on contextual cues from shared bibliographic metadata. Here, we present a redesigned user interface for SALT to address usability concerns identified during a user evaluation, and extend it to incorporate computational features as additional semantic context. These derived features quantify specific aspects of information resources such as musical recordings and textual documents, mathematically characterizing, e.g., the musical keys represented in an audio signal, or the token, line, and page counts within a text. Such metadata describing aspects of the content of information resources provide valuable additional cues, alongside bibliographic facets, to the expert user undertaking the alignment task.

Publisher

iSchools

Series/Report Name or Number

iConference 2017 Proceedings

Type of Resource

text

Language

en

Permalink

http://hdl.handle.net/2142/96727

Copyright and License Information

Copyright 2017 Catherine Blauvelt, David M. Weigl, J. Stephen Downie, and Kevin R. Page

Owning Collections