Saturday, 11:30am, Grand Salon B
One might think that search interfaces have been well worked out and reduced to standard design patterns and best practices. But one would be wrong in espousing that view. The user interfaces for search are evolving as new features and capabilities are developed. And the appropriate patterns and best practices to use in meeting user expectations are anything but clear.
One emerging capability that raises new design questions is that of federated search or
“metasearch,” a search engine that applies the user’s keyword search terms across data bases or collections of content. Many government agencies, professional organizations, and private sector entities maintain multiple collections of related publications or bibliographic content. Users of such collections are often unfamiliar with where their desired information resides and thus in which collection of material to concentrate their search.
What’s more, they may or may not care. They want to find the best information available on a topic of interest, regardless of which database delivers the goods.
By happenstance, in our consulting practice we have had three clients in the last few months with questions about the usability of metasearch interfaces. Two were government agencies (the Library of Congress and the National Institutes of Health Library) and one was a professional organization (the American Chemical Society). Their concerns were similar. What are user expectations with regard to advanced search pages and the display of search results in a metasearch environment? Will users understand the functionality being offered for browsing through search results or narrowing their search? How readily usable are their prototype interfaces for metasearch? How long are users willing to wait for federated search results? Do users care to know (and would they benefit by knowing) from which database their
“hits” are coming from and what is the best way to offer that information?
This presentation will focus on current issues in metasearch interfaces and our findings from usability tests of metasearch capabilities in these three environments, as well as related findings from past testing of other search interfaces. There was some commonality of findings across the three aforementioned metasearch environments, if not a common design approach to the interfaces. However, there were also some differences in what worked best related to the sophistication of the searchers, i.e., in the expectations and propensity to utilize certain features between the general public as opposed to scientific researchers with expert knowledge of the subject matter being searched.
From our usability test data we will illustrate some design patterns that worked well and not so well and attempt to draw some guidelines and best practices that will hopefully generalize to other metasearch interfaces. We will address progress indicators and partial displays of results, user choices in ordering search results, browsing through categories of search results, effective use of a thesaurus, and personalization features in search interfaces.
To foreshadow some of the issues to be addressed:
- The question of whether users want or need to know where the best information on their topic of interest resides affects both what options to offer on advanced search pages and how to display the search results.
- Particularly in a metasearch environment, the model of the user entering keywords, launching a search, and culling through pages of results fails to take advantage of many of the capabilities being offered. Increasingly, the user experience with search is one of interacting with the search results, iteratively refining ones terms, and narrowing in on the best information available. How to present the options available to the user becomes an issue of both educating the uninformed about what capabilities exist and making the search experience as productive as possible for users who already know about them.
- Depending on the volumes of content being searched, and the complexity of the logic in the search terms, federated searches can take a significant amount of time. Should partial results be displayed as they become available, and if so how should the display indicate that the search results are still being built? Or should the user simply have to wait, and if so how should progress and estimated time to completion be indicated?
- Many federated search products now attempt to categorize the search results into clusters of content organized by these categories through which the user can browse. How should such faceted browsing be organized and do users know how to take good advantage of these capabilities?
- Many search interfaces offer some flexibility to the user in how the search results are ordered, by relevance ranking, chronologically, alphabetically by author or title or journal, etc. Do users understand these options and utilize them effectively?
- Some databases of content have metadata that relate to a thesaurus such as the MESH index of medical terminology or the ERIC thesaurus of educational research, both of which have been built over a period of 20+ years. How should suggestions for related terms be offered on a search results page and do users know how to use a thesaurus when they see one?
- Some metasearch interfaces are starting to offer personalization features, such as the ability to save search terms or search results, the ability to configure email or RSS alerts when new content of interest becomes available, the ability to locate search results among the physical holdings of institutions by geographic proximity, etc. What best practices can be derived for the design of user interfaces with such personalization capabilities?
INSTRUCTORS
Dick Horst
Dick Horst is an experimental psychologist and user experience specialist with over 30
years’ experience in applied behavioral research, systems development, and usability engineering. He is the founder and President of UserWorks, Inc., a usability engineering consulting firm in the Washington, DC area. Dick has conducted or overseen usability engineering projects for over a hundred organizations in government, the private sector, non-profits, and academia. He has a Ph.D. in experimental psychology from Carnegie-Mellon University and is a board certified professional
ergonomist.
Dana Douglas
Dana Douglas is a user experience specialist at UserWorks. She has conducted usability testing, focus groups, and user surveys for a variety of government and private sector clients. Search capabilities have been an issue in many of the user interfaces with which she has worked. Dana has a bachelors degree in information systems from the University of Maryland Baltimore County (UMBC) and is enrolled in the masters program in human-centered computing there.
Cory Lebson
Cory Lebson is an independent consultant specializing in usability engineering and market research. He is a former employee of UserWorks and continues to consult with the company on ongoing projects. He was the lead on two of the metasearch-related usability testing projects that will be presented here. Cory has a background in cognitive psychology with masters degrees in sociology and an MBA with a specialty in technology management.