Search

Transaction logs from four different Web-based information retrieval environments (bibliographic databank, OPAC, search engine, specialized search system) were analyzed for empirical regularities ill search characteristics to determine whether users engage in different behaviors in different Web-based search environments. Descriptive statistics and relative frequency distributions related to term usage, query formulation, and session duration were tabulated. The analysis revealed that there are differences in these characteristics. Users were more likely to engage in extensive searching using the OPAC and specialized search system. Surprisingly, the bibliographic databank search environment resulted in the most parsimonious searching, more similar to a general search engine. Although on the surface Web-based search facilities may appear similar, users do engage in different search behaviors. Keywords. Information retrieval: Web search characteristics; Transaction log analysis

The authors investigate the frequency distribution of the use of image tags in Web pages. Using data sampled from top level Web pages across five top level domains and from sample pages within individual websites, the authors model observed patterns in the frequency of image tag usage by fitting collected data distributions to different theoretical models used in informetrics. Models tested include the modified power law (MPL), Mandelbrot (MDB), generalized waring (GW), generalized inverse Gaussian-Poisson (GIGP), and generalized negative binomial (GNB) distributions. The GIGP provided the best fit for data sets for top level pages across the top level domains tested. The poor fits of the models to the observed data distributions from specific websites were due to the multimodal nature of the observed data sets. Mixtures of the tested models for the data sets provided better fits. The ability to effectively model Web page attributes, such as the distribution of the number of image tags used per page, is needed for accurate simulation models of Web page content, and makes it possible to estimate the number of requests needed to display the complete content of Web pages. Keywords: Informetric modeling; Cybermetrics; Image tag distributions

The emergence of general audience digital libraries (GADLs) defines a context that represents a hybrid of both 'traditional' IR, using primarily bibliographic resources provided by database vendors, and 'popular' IR, exemplified by public search systems available on the World Wide Web. Findings of a study investigating end-user searching and response to a GADL are reported. Data collected from a Web-based end-user survey and data logs of resource usage for a Web-based GADL were analyzed for user characteristics, patterns of access and use, and user feedback. Cross-tabulations using respondent demographics revealed several key differences in how the system was used and valued by users of different age groups. Older users valued the service more than younger users and engaged in different searching and viewing behaviors. The GADL more closely resembles traditional retrieval systems in terms of content and purpose of use, but is more similar to popular IR systems in terms of user behavior and accessibility. A model that defines the dual context of the GADL environment is derived from the data analysis and existing IR models in general and other specific contexts. The authors demonstrate the distinguishing characteristics of this IR context, and discuss implications for the development and evaluation of future GADLs to accommodate a variety of user needs and expectations. Keywords: Information retrieval context; Digital libraries; User studies; Internet searching

Simulation models for boolean-based information retrieval (IR) system performance under varying hypothetical system circumstances are tested to determine maximum performance and minimum space for file structures. While the hashing structure showed the best performance in retrieval time and space economy, the chained hashing structure is better for gradually decreasing term distributions because of minimal retrieval performance variations. The modified hashing structure and the double Balance Implicity Multiway tree structure, however, are best for steep term distributions and more frequently occurring terms.

Several types of information retrieval (IR) systems with different file structures, informetric database content and usage patterns are compared using simulation techniques. Simulated performance results show that a shifted Waring distribution provided the best fits for large term distribution data sets, while a shifted negative binomial distribution provided acceptable fits for models of the number of terms used per question. Meanwhile, a complex relationship between term frequency of occurrence and use is suggested by preliminary data.