No Results Found

No results found” is a misleading phrase because it masquerades as a definitive answer; in reality, the library being searched may contain content that matches a user’s query. This research seeks to understand the effect of null result sets on search behavior and on the perception of contents in digital libraries. Concepts utilized by social computing frameworks and current-generation eCommerce Web sites will be examined for their efficacy in digital library search interfaces. An experiment is outlined that attempts to flesh out a connection between null result sets and authoritativeness of digital library contents. Introduction The poster child for failure in information retrieval is a slew of search results that have nothing to do with the search query presented to the system by the user. If, for example, one were to search a database of household pets for the term “dog” and only cats with the name “dog” were returned, we could easily consider this to be a failure (unless, of course, the database actually did not contain any dogs). Similarly, if one were to perform a Web search on a hypothetical 5-year-old neighbor named Michael Jackson, and the only results returned dealt with the famous musician, there would be limited recourse to explain to the system the intended meaning (see, e.g., McRae-Spencer & Shadbolt, 2006). Much research has been undertaken in the information retrieval (IR) field to address this problem, ranging from the traditional concept of search term vectors, to popularity based on interrelations, to similar word pairings (e.g., synonyms, singular/plural forms of words, and verb tenses). To a large extent these efforts have been tremendously successful; modern search engines—whether they categorize the documents on a private Intranet, the items in a digital library, or the Web at large—are much better at