Collaborative Filtering: A New Approach to Searching Digital Libraries

Webster, Janet; Jung, Seikyung; Herlocker, Jonathan L.

doi:10.1080/13614570500054229

Cited by 8 publications

(6 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, collaborative filtering remains the dominant method since it is especially useful when the items to be recommended have few or no content-based features. Webster et al (2004) point out that since many traditional library resources, such as catalogues, contain only metadata about the items in a collection (i.e. there is no full text to index), content analysis techniques have limited usefulness.…”

Section: Related Workmentioning

confidence: 99%

Recommending research articles using citation data

Vellino

2015

Library Hi Tech

View full text Add to dashboard Cite

This study compares some of the characteristic differences among recommendations generated by a citation-based recommender and a user-based recommender for research articles. As with other application domains, the application of collaborative filtering methods for recommending items in a digital library suffers from a sparsity of usage data. One method for addressing this sparsity problem is to employ an article's references as a proxy for co-download information. Another method is to harvest large quantities of usage data from distributed OpenURL web logs. These different data sources produce significantly different recommendations, both at the individual and at the aggregate level. The experiments randomly selected about 9,000 articles from among 6.6 million articles as starting points for generating recommendations. Among the characteristics of the resulting recommendations are the semantic distance between the seed articles and the recommended ones, the coverage of the recommendations and the spread in publication dates between recommendations and the seed. A parallel experiment on a subset of these articles aimed to test the hypothesis that generating PageRank values for articles might be an effective method for weighting references to improve citation-based recommendations.

show abstract

Section: Related Workmentioning

confidence: 99%

Recommending research articles using citation data

Vellino

2015

Library Hi Tech

View full text Add to dashboard Cite

show abstract

“…The huge growth in available resources, particularly through the increased provision of e‐books and electronic journals, has led to a wealth of accessible information with the potential to inform scholarship, facilitate teaching and research, and satisfy readers' needs. While the delivery of that content over the web has dramatically increased the range and scope of services available to library users, it has become increasingly clear that library systems frequently fail to best connect their users with potentially relevant material (Webster et al 2004).…”

Section: Introductionmentioning

confidence: 99%

If we build it, will they come? Recommendations and WorldCat

Wakeling

Sen

Clough

et al. 2011

Proc. Am. Soc. Info. Sci. Tech.

View full text Add to dashboard Cite

This paper reports on a collaborative research project currently underway between the University of Sheffield's Information School and OCLC, the goal of which is to produce a recommender system for WorldCat.org -the publically accessible aggregated catalogue of the holdings of OCLC's member libraries. It highlights two of the key issues facing recommender systems design in the library domain: 1) Trust: users' trust both in the system itself and in collaboratively generated suggestions; and 2) Context: the difficulties of a system adapting to the huge range of information-seeking contexts that accompany a diverse user base. The paper describes how future qualitative and quantitative research will attempt to address these issues, and produce a system that best serves users of WorldCat.org.It is hoped that as well as contributing to recommender systems research in general, the results of this project will help inform debates relating to the interaction between users and systems in a library context.

show abstract

“…The huge growth in available resources, particularly through the increased provision of electronic journals and e-books, has led to a wealth of accessible information with the potential to inform scholarship and facilitate teaching and research, while the provision of that information over the web has dramatically increased the range and scope of services available to users (Webster et al, 2004). Despite these undoubted benefits, however, it has become increasingly clear that traditional models of information retrieval frequently fail to best connect library users with potentially relevant material.…”

Section: Recommender Systemsmentioning

confidence: 99%

“Readers who borrowed this also borrowed … ”: recommender systems in UK libraries

et al. 2012

View full text Add to dashboard Cite

Purpose -Moves towards more interactive services on the web have led libraries to add an increasing range of functionality to their OPACS. Given the prevalence of recommender systems on the wider web, especially in e-commerce environments, this paper aims to review current research in this area that is of particular relevance to the library community. It attempts to gauge the uptake of recommender systems in exiting OPAC services, and identify issues that might be responsible for inhibiting wider uptake. Design/methodology/approach -This paper draws on an extensive literature review, as well as original research comparing the functionality of 211 public and 118 university library OPACs in the UK. Examining current recommender systems research, it outlines the most significant recommendation models and reviews research in two key areas of recommender systems design: data acquisition, and the explanation of recommendations. It discusses three existing library recommendation systems: BibTip, LibraryThing for Libraries and the in-house system at the University of Huddersfield. Findings -The authors' analysis indicates that the incorporation of recommender systems into library services is extremely low, with only 2 per cent of public libraries and 11 per cent of university libraries in the UK offering the feature. While system limitations and budget constraints are perhaps partly to blame, it is suggested that library professionals have perhaps yet to be persuaded that the value of recommendations to library users is great enough to warrant their inclusion becoming a priority. Originality/value -This paper represents the first study of UK library OPACs to focus on the prevalence of recommender systems.

show abstract

Collaborative Filtering: A New Approach to Searching Digital Libraries

Cited by 8 publications

References 11 publications

Recommending research articles using citation data

Recommending research articles using citation data

If we build it, will they come? Recommendations and WorldCat

“Readers who borrowed this also borrowed … ”: recommender systems in UK libraries

Contact Info

Product

Resources

About