You are here

More Like This: Approaches to Recommending Related Items using Subject Headings

Kevin Beswick, NCSU Libraries

With a significant portion of the collection at our new Hunt Library being housed in an automated storage and retrieval system, several of us at NCSU Libraries have begun looking at ways to replace and improve upon the classic shelf browsing experience in order to make it easier for patrons to browse related materials. Our goal is to mimic popular services like Amazon and Netflix, which utilize recommendation engines to make it easy for users to find items similar to a particular item of interest. While there have been previous efforts in libraries to recreate this experience using circulation or call number data, we are currently investigating algorithms that focus on use of subject headings. Use of subject headings as an alternative can be particularly helpful in the case of electronic materials that do not always have call numbers or circulation data. In this talk, I will share:

  • Details of the proposed algorithms
  • How these algorithms were quickly and easily implemented using Solr.
  • Our evaluation process and its outcomes in terms of the effectiveness of the algorithms.
  • How this has (or could) impact presentation of recommended items in our discovery layer.