Understanding PubMed(R) user search behavior through log analysis

Abstract

This article reports on a detailed investigation of PubMed users’ needs and behavior as a step toward improving biomedical information retrieval. PubMed is providing free service to researchers with access to more than 19 million citations for biomedical articles from MEDLINE and life science journals. It is accessed by millions of users each day. Efficient search tools are crucial for biomedical researchers to keep abreast of the biomedical literature relating to their own research. This study provides insight into PubMed users’ needs and their behavior. This investigation was conducted through the analysis of one month of log data, consisting of more than 23 million user sessions and more than 58 million user queries. Multiple aspects of users’ interactions with PubMed are characterized in detail with evidence from these logs. Despite having many features in common with general Web searches, biomedical information searches have unique characteristics that are made evident in this study. PubMed users are more persistent in seeking information and they reformulate queries often. The three most frequent types of search are search by author name, search by gene/protein, and search by disease. Use of abbreviation in queries is very frequent. Factors such as result set size influence users’ decisions. Analysis of characteristics such as these plays a critical role in identifying users’ information needs and their search habits. In turn, such an analysis also provides useful insight for improving biomedical information retrieval.Database URL: http://www.ncbi.nlm.nih.gov/PubMed

Automatic Suggestion for PubMed Query Reformulation

“…As pointed out by Islamaj-Dogan et al [1], there are a lower abandonment rate and a higher reformulation rate for PubMed, implying that PubMed users are more persistent in pursuing their information needs, than users of other search systems [17,18]. This issue would possibly be addressed by considering the sequence of query reformulations by a user (see below for further discussion).…”

“…We present a novel statistical method that computes the relative frequency of words. Since it is impractical to use the whole literature, we obtained a representative subset of the literature, by using a generic search term "gene", which is one of the most frequent words in PubMed query logs [1], to retrieve the titles and abstracts of the first 2,000 results. We call the set of the resultant documents the generic corpus (GC).…”

“…Researchers continue to revise their queries until they get what they want. In fact, 47% of all PubMed queries are followed by a new subsequent query [1]. In this paper, we present a novel approach to query reformulation, which suggests 'likely' new queries for a given user query.…”

Analyzing Medical Image Search Behavior: Semantics and Prediction of Query Results

“…Among all queries, 182,004 (90.8 %) consisted of three or fewer terms. In contrast, PubMed averages 3.54 terms per query [12], with a median of 3 terms per query; 80 % of all queries have no more than 4 terms. Figure 2 shows the number of queries given the number of terms in it.…”

“…They also have been used to analyze image search behavior [9,10]. Analysis of MedLine search behavior in the medical literature was conducted based on log files [11,12]. Closest to the presented work are the analyses of Tsikrika et al [7] and Rubin et al [13] that both used ARRS GoldMiner log files, but a much smaller set of queries (25,000 and 30,000, respectively, so around 10 % of the data used in this text).…”