Friday, November 04, 2005

Self Discovery: Federated Search Engines & Subject Pages

Frank Cervone, Assistant University Librarian for Information Technology, Northwestern University
Darlene Fichter, Data Library Coordinator, University of Saskatchewan

Numerous usability studies demonstrate that library Web site visitors have difficulty locating and retrieving articles. The speakers describe two different approaches that work hand in hand for helping users locate relevant materials. Cervone shares recent results about the best placement of federated search boxes, display of search results, and design decisions in creating topical groups. Find out what works and what doesn’t. Fichter describes the results of a series of tests looking at the effective design of browsable library subject pages. Through a rapid cycle of testing, design changes, and retesting, subject pages were adapted and changed based on user behavior. Both discuss challenges and lessons learned.

Powerpoint presentation



My Notes:

Federated Search Services

Users understand searching Google or individual databases, but not what 'metasearch' means.

In testing, they introduced a metasearch engine and asked the users what they'd call it. Nobody used 'metasearch' or anything like that. They called it the 'catalog.'

To appeal to undergrads it needed simple interface.

Faculty not concerned about 'databases' per se, but about the journal.

In federated search engine, users prefer simple search - a Googlelike experience. Users won't use complex query statements. They feel succcesful in simple search environment because they find articles, even if the articles aren't exactly what they searched for. Unexpected advantages to this - find new resources & citations previusly unaware of, the browsing factor.

Most people who are not librarians tend to search the same way.

Their expectations: They want to see things in relevance order, even when it is not clear how relevance is determined. When results are not displayed that way, patrons often figure out display order but they don't like it and when they have the opportunity, they change over to relevance.

Overall most patrons prefer simple search interface instead of going into a long list of databases and checking them off. (made me think of something from the from the first session - too much text?) - often say that long lists like that make them feel stupid.

------------

Designing Effective Subject Pages

Testing at teh University of Saskatchewan: 66% of time their site failed to support users looking for articles. Only 33% of users could locate databases by topic. (i.e., when given a questions such as: Find two useful full text databases on religion.) Many gave up after trying for 3 or more minutes.

They found that simple terms like this sucked students in:
Databases: subject | title
Electronic Journals

Users love the feature of 'limit to scholarly peer reviewed'.

When confronted with lists of cryptic database names - what do people do when? They either click first one or they leave the page.

Lacking the librarian's mental model or organization - not looking for databases to find an article. Users headed for electronic journals instead of article databases - "if only I knew a journal title!" Faculty, when confused or confronted with intimidating interface send research assistants to do everything. Faculty know the articles they want from word of mouth and find them via Google more than the library.

How do people look for information? Undergrads look by format: article, book, then subject then journal title.

Who are you designing for? Amorphous mass of 'users' not good. U of Saskatchewan's library webpage is not designed for expert user but novice user who doens't know where to start.

Best Bets section - if you don't know where to start, start here.

When designing, think Just In Time information, not Just In Case.

- online access info (licensing restrictions)
- removing unnecessary info.
- Design for recognition, not recall.
- Give the users just enough info to make next step of journey; don't overwhelm them with verbiage.
- Less is more; the page should scan quickly, with a very brief info to reduce information pollution and cognitive overhead.
- Group similar things together (format, tutorials, library info).
- Use structure, layout, & position to organize info.
- Create zones with whitespace to route things.

When unfamiliar with a new domain, browsing may be more effective than searching

Experts don't get stuck in Best Bets-type of sections, but go down the page, so no worries.

Default to journal articles - #1 thing people look for & have trouble looking for.

Best Bests - looks kinda like Google - brief annotations. In testing, nobody clicked a detail page, just went straight to the database.

Need clear udnerstanding of purpose of page. Simple & easy to use is not always simple & easy to design.

(Thoughts for our library: add brief annotations to database alpha list.)

People tended to use the database they knew best, even if switching areas of study.

No comments: