Ac 2011-529: the Visualization of Database Search Results

Edward Tufte describes the visual presentation of quantitative data as “envisioning information.” A number of databases, from Google to specialized subject databases, offer a graphical representation of research results. Visualization is offered as an alternative method for evaluating search results. A survey of the current mechanisms or tools for providing a visual or graphical presentation of database information shows that this trend may be on the decline. Introduction: Edward Tufte describes the visual presentation of quantitative data as “envisioning information.” The third book of his famous trilogy is titled Visual Explanations, and best lays out his thesis that “information displays should be documentary, comparative, causal and explanatory, quantified, multivariate, exploratory.” These qualities are, for the most part, lacking from the standard display of a bibliographic database search result! Although a list of references or citations may be numbered and sorted, the display of this information offers limited guidance to the user in determining the best source to choose from the list. Within the last five or six years, a number of databases, from Google to specialized subject databases, have offered a graphical representation of research results. These visualization tools are usually offered as an alternative method for evaluating search results, through a sidebar menu or button. This trend deserves closer scrutiny, since the question of how database users evaluate their search results is central to the success of their library research. Diagramming Relevance: When the result of a database search is a long list of citations; to journal articles, books in a catalog, or web sites, the cues for selecting the “best” sources from the list are limited. Most databases sort by relevance or by date, and users have come to rely on this feature to assume that the first half-dozen results on the list are the best. More sophisticated users might look for key search words in the title or link. Only an experienced researcher might scan a citation list and select items based on familiarity with the authors or their research institution, which journal an article is published in, number of times cited, or other visual cues. When scanning the results of a Google search, similar cues are used (eg: is the URL a dot.com or a dot.gov?), and familiar web sources such as Wikipedia often appear on the first page of search results. In the last decade Groxis developed the visual search engine Grokker, creating graphical representations of search results that looked like something between a Venn diagram and a bubble chart. This sorting and grouping of search results did appear to make narrowing a search easier. The diagrams showed references in a broader context, and offered alternative avenues for research, and their relative weight based on the size of the graphics. For a time, some EbscoHost databases offered Grokker visual search results as an option, but not a prominent feature of their databases. Groxis had “ceased operations” as of 2009. Ebsco databases still offer a “visualization” tool to show how search refinements narrow results. However, this is not much different from a breadcrumb trail in the menu bar of a web browser. It does not highlight alternative avenues of research, or show context for the selected references. Web 2.0: Tag clouds or weighted lists, introduced by Flickr and del.icio.us, are databases built on individually assigned metadata “tags.” These information displays are wonderfully intuitive, highlighting the most popular or common tag words at a glance. Flickr users can search tags, and even georeference tags on maps. Thumbnail photographs display the results of an image database search, but do not organize the search results in any meaningful way. Yahoo’s “delicious” search results are displayed as text hit lists. Both fail to take full advantage of the possibilities of tag cloud as search result. Although individual users can create their own personal tag cloud, database searchers cannot see a cloud generated by a specific search, collating the collective entries of many users. The wisdom of crowds is not translated into clouds. IBM’s Many Eyes is an interesting experiment in user-generated information visualization. This site “allows users to upload data and then produce graphic representations for others to view and comment upon.” The database can be searched successfully by broad topics; the availability of specific information that a researcher might be seeking is entirely dependent on what has been uploaded by other users. Not strictly a bibliographic database, Many Eyes is close to some of the ideals put forward by Tufte for the visual presentation of information. Searching with Pictures: The most successful visualizations are incorporated into databases which index graphical information, from photographs to molecular drawings. Chemical engineers recognize the value of searching a graphical database such as SciFinder, and the power of being able to draw part of a chemical structure to search for related information. Displaying search results as chemical structure diagrams with accompanying text makes this a powerful research tool, but one that requires specialized knowledge and training to use well.