Known-item Searches Resulting in Zero Hits: Considerations for Discovery Systems

https://doi.org/10.1016/j.acalib.2016.12.002Get rights and content

ABSTRACT

The goal of this article is to understand the reasons why known-item search queries entered in a discovery system return zero hits. We analyze a sample of 708 known-item queries and classify them into four categories of zero hits with regard to whether the item is held by the library and whether the query is formulated correctly: (1) item in stock, but query incorrect, (2) item not in stock, (3) item in stock, but incomplete or erroneous metadata, (4) query is ambiguous or not understandable. The main reasons for zero hits are caused by acquisition and erroneous search queries. We discuss possible solutions for known-item queries resulting in zero hits from the side of the system and show that 30% of zero hits could easily be avoided by applying automatic spelling correction. We argue that libraries can improve their discovery systems or online catalogs by applying strategies to avoid or cope with zero hits inspired by web search engines and commercial search web sites.

Section snippets

INTRODUCTION

Web search engines in general and Google, in particular, are popular tools in information seeking today. Users perceive web search engines to be successful in providing satisfactory results and therefore to be successful in satisfying their information needs (Purcell, Brenner, & Rainie, 2012). One of the major reason for that might be that a significant number of search queries can be identified as being navigational, i.e., queries intended for a known website or a website assumed to exist (

LITERATURE REVIEW

There is no standard definition of known-item searches and interpretations of the concept behind this type of search vary. In general, the overall proportion of known-item search queries in library information systems is as significant as the number of navigational queries entered in web search engines, as we describe below. Libraries must consider retrieving the correct search result for a known-item search query in order to improve their modern information systems' retrieval effectiveness and

METHODS

We extracted 2000 search queries from transaction log files that were entered in the single search interface of EconBiz, in July 2014. EconBiz is an information portal for economic literature provided by the German National Library for Economics (ZBW - Zentralbibliothek für Wirtschaftswissenschaften)i. It has implemented the open source discovery software VuFind and combines data from multiple heterogeneous sources. Based on the queries from the log files we built a

PROPORTION OF ZERO HITS

Only 394 (56%) of the 708 queries in the sample yielded correct results, while in 314 cases (44%) zero hits occurred. Of these zero hits queries, 172 did not return any results at all, leaving 142 queries that yielded at least one result but none that would match enough query terms (see Fig. 3).

With regard to this quite large proportion of queries yielding no results at all the main question that arose was, whether the assumed to be known items could not be found simply because they had not

CONCLUSION

The goal of this article was to understand the reasons why known-item queries result in zero hits. We built a sample of 708 known-item queries and analyzed their retrieved results using an information portal for economic literature, EconBiz. Of the 708 queries, 314 queries (44%) yielded zero hits, i.e., queries that did not yield any results and queries that did return some results but not the correct one. We identified ten different categories of reasons for zero hits that we combined into

ACKNOWLEDGEMENTS

Findings presented in this article were partly examined within the research project LibRank – New Approaches to Relevance Ranking in Library Information Systems. The project was funded by the German Research Foundation (DFG - Deutsche Forschungsgemeinschaft) from March 2014 until February 2016.

REFERENCES (28)

  • B.J. Jansen et al.

    Determining the informational, navigational, and transactional intent of web queries

    Information Processing & Management

    (2008)
  • N.K. Kaske

    On my mind: Materials availability model and the internet

    The Journal of Academic Librarianship

    (1994)
  • B. Allen

    Recall cues in known-item retrieval

    Journal of the American Society for Information Science

    (1989)
  • A. Bailey et al.

    Retrieving known items with LibX

    The Serials Librarian

    (2008)
  • A. Broder

    A taxonomy of web search

    ACM SIGIR Forum

    (2002)
  • S. Chapman et al.

    Manually classifying user search queries on an academic library web site

    Journal of Web Librarianship

    (2013)
  • F.W. Chickering et al.

    Evaluation and comparison of discovery tools: an update

    Information Technology and Libraries

    (2014)
  • N. Craswell et al.

    Overview of the TREC-2004 web track. In the thirteenth text retrieval conference (TREC 2004)

    (2005)
  • J.A. Crum

    An availability study of electronic articles in an academic health sciences library

    Journal of the Medical Library Association

    (2011)
  • G. Nudelman

    Designing search: UX strategies for eCommerce success

    (2011)
  • V. Frants et al.

    Automated information retrieval: theory and methods

    (1997)
  • M.-Y. Kan et al.

    Detecting and supporting known item queries in online public access catalogs

  • P.B. Kantor

    Availability analysis

    Journal of the American Society for Information Science

    (1976)
  • M. Khabsa et al.

    Towards better understanding of academic search

    • Cited by (8)

    View all citing articles on Scopus
    View full text
    © 2016 Elsevier Inc. All rights reserved.