Users' adjustments to unsuccessful queries in biomedical search

Biomedical researchers depend on on-line databases and digital libraries for up to date information. We introduce a pilot project aimed at characterizing adjustments made to biomedical queries that improve search results. Specifically we focus on queries submitted to PubMed®, a large sophisticated search engine that facilitates Web access to abstracts of articles in over 5,200 biomedical journals. On average 2 million users search PubMed each day. During their search, nearly 20% will experience a result page from one of their queries that has zero results. In some cases there really is no document or abstract that will satisfy a particular query. However, in analyzing one month of queries submitted to PubMed, we find that more often than not, queries that retrieved no results are queries that would retrieve something relevant if they were constructed differently. This paper describes a new effort to identify some of the characteristics of a query that produces zero results, and the changes that users most often apply in constructing new, "corrected" queries. Zero-result queries afford us an opportunity to examine changes made to queries that we know did not return relevant data, because they did not return any data. An investigation of the changes users make under these circumstances can yield insight into users' search processes.