California Digital Library Interface Design Project Task Analysis and Initial Sketches

Overview | Task Analysis | Suggested Solution | Experiment Outline

Task Analysis:

Target users and Tasks

We assume that the potential users of CDL services range from librarians, professors, and students to university administration and the general public. The existing CDL home page gives an unusually high visibility to the administration of the site. However, we want to target those who will be searching the collections of this "10th library" for specific research needs. The students, professors, and librarians who typically perform this role need to be able to quickly learn what the CDL offers and have fast access its resources. While the university administration and general public will still enter this site, we feel their needs for information about the CDL as an institution can be deemphasized. We are targeting the researchers who currently use Melvyl and systems like Gladis from home or on campus and who will need to bridge from those systems to the CDL in light of CDL's mission to help create economies of scale for the UC system's collection growth.

Our task analysis questions focus on the types of tasks that one may normally follow in the existing systems and libraries to research information. We also solicited ideas on a metaphor or model that can be used to structure the CDL interface to enhance its conceptual clarity and ease of use.

Task Analysis Questions:

The multiple part questions helped to elicit nuances from the interviewees' responses.

1. When you need to find a book or information on a topic from a library, do you use some sort of online cataloging system? How do you structure your search?
2. What other resources do you use to do this and how do you use them?
3. What do you wish you could do in a search like the ones we just discussed that you haven't seen available or, even if it was available, wasn't available easily enough to be really useful to you?
4. When you're not sure what you want at a library, and just want to browse to see what you can find, what do you do? What resources would you like to have to support that activity?
5. How do you search for a topic on a big academic website? What resources do you look for to help you do this? What have you seen in such sites that were helpful? What do you wish had been available but wasn't?
6. How do you browse a big academic website if you just want to see how it works and what it has available? What resources do you look for? What have you seen in such sites that were helpful? What do you wish had been available but wasn't?
7. If a website is offering you a link to a service with a limited, command line interface, what kind of warnings or help would you like to see?
8. If you were hired to help create the best kind of academic information website, one that people would find not just useful but a pleasure to use, what kind of picture or metaphor or structure comes to mind as you imagine someone opening up that page and getting their first look? What would they see?
9. When you think about a web-based digital library associated with a University, what possibilities do you imagine in terms of the services that it would offer? What would you like to be able to do there?

Characteristics of Interview Participants

(1) Research fellow.
(2) Former UCB library staff.
(3) Ph.D. candidate in Structural Biology (Molecular & Cell Biology Dept)
(4) Staff member Graduate Studies Library.
(5) Recent undergrad graduate. Entering graduate school in fall.

Summary of Interview Responses

1) & 2) Participants stated that they immediately used the search function of the electronic system that was offered at the library. In the case of the graduate student, Melvyl and the bibliographic databases factored in highly. (Display the results with the LOC UCB options, which gets you right to the call numbers.) Undergrads typically used Gladis. For subject specific searches, one participant simply used "typical" web search engines to find homepages for journals or authors to post to class site. Or used specific proprietary search services like Lexis-Nexus. "Browsing" was just considered a "coarser" search-less specific. Almost all participants acknowledged that for "hard searches" or if they encountered trouble with their own search methods, they ask a reference librarian. Three participants brought up some trepidation with the Web interface vs. a command line regarding speed of retrieval, ease-of-use, as well as integrity of information relayed.

3) Wishlist? Use of a librarian like in physical library who could take general requests. Perform a partial title search and limit a search to a specific location. Boolean keyword searches should be expanded to allow grouping of words or phrases. Option for computer to print out or show other options for searching related subjects. An example given of a very useful and specialized tool was found in the Web of Science Citation Index. Here you can keep track of every time your research is cited to see how the primary literature is evolving. Click on an article, check the "pedigree" of the researcher/investigator, be able to tell where you've been & autosave (topic & kw searches like in MELVYL) or delete those that you no longer want to keep on your radar screen. Click on references to find original trail, get updated on future references to follow the trail. It remembers who you are. Would like CDL to keep that type of information, too, but not as simple as dropping cookies because he is in a shared use lab. Need to have some sort of sign-in.

4) Browsing needs? A subject map broken down by subcategories was highly recommended. Separate shelves for new books (by category) and having the collection out and accessible is important. Or at least alert the patrons to the fact that some of the collection is not browsable. Ability to do full-text search of journal articles for specific topics by keywords. Browse by using cross-citations. All else fails ask a librarian.

5) Search a big academic site? Use search button, which seems to indicate the existence of content. Find the TOC or site index if there is no search function. Use the logical structure of the site (e.g., SIMS site--"things are where you'd think they'd be.") Maybe find small thumbnails of documents/resources like a document management system.

6) Browsing a big academic site, helpful? A site index again factored highly, which CDL is sorely lacking. No site search was considered frustrating. A typical web feature is a "welcome" indicating general info the site is expected to have. Visually, the preferences ranged from text only to having big categories across the top or along the side (as in frames), or having hyperlinked letters A - Z across the top of the page.

7) Link to a limited command line interface, warnings or help? The responses ranged from a firm warning to a simple notification. Some unfamiliar command line interfaces can be daunting and time-consuming, but some users said they were more comfortable with a command line UI and typically have images off on the web. Some libraries have a "you are now leaving our site, thanks please come again" window that can alert you to new language. And there was some need to indicate the integrity if it is not part of CDL then how frequently is it updated. However, where CDL links to Melvyl, one user would like to see a window open to Melvyl and as you work in Melvyl, the website retrieves the links and keeps them there for you (remembers your history.) For "Help", the e command in Melvyl is easy and all the one-page sheets he used years ago to learn the system were very helpful.

8) Metaphor or structure for site? Unfortunately, websites aren't seamless -- click & wait & click & wait. And, it's a very narrow window with no context or peripheral vision. A banner or river metaphor might help indicate the flow of information. Or, a book. It helps keep the visual representation simple on the front page...clearly indicates that "information" exists on that site (albeit using an old icon). He realizes librarians might see differently (with other info media available today), but likes the idea that your mental model thinks "books" when you think "libraries" (not files or card tabs.) Whatever it is, it should NOT stop or slow you down from getting to the Info! The structure should be able to jump 10 hits to get to what he wants when searching. Have as many abstracts as possible. Be able to search as many libraries as possible. Cut and paste easily between Gladis and Melvyl. Have a well-described info hierarchy. Don't want to wade through stuff and relations. Not too many pictorial elements to speed up the web page reload. Be able to go to the top page and the previous page without dealing with pages in between.

9) Possibilities imagined? Full text of articles, searchable, downloadable, delivered to your door on request. Email, campus mail... For on-campus people it could be free, and as a service to businesses they could charge for the information delivery... [Rachna] uses A.com as a card catalog; could CDL make similar detailed info and well- structured linkage available? Abstracts of books and the ability to search on the full text of books -- order if you want 'em. Rare books -- scan and make available. Exploit the visual nature of the web by clearly having searchable archives of art & photos. Full text of resources--not just finding aids. I want to see and use the object I'm retrieving [not thumbnails -- but do want full text!) Easy downloading of information that is readable by ENDNOTE software application that formats your citations in the appropriate format per journal you are submitting it in. Search mechanism in a natural language like AskJeeves search engine that "learns" as you work with it. Can't imagine the digital library with content. See DL as a searching tool rather than anything else. .

Representative Scenarios:

1. Browse a topic

Ken, a new system user, probably a new graduate student or UC staff member is trying to get oriented to the CDL site and to explore a topic of interest. Ken enters the CDL looking for information on Art History and needs help at each step along the way.

2. Specific item search

Phillipe has been to CDL before and wants to find a specific journal article that he believes may be available online. He only has a partial memory of the journal's name. He wants to quickly find an electronic journal on genetics without knowing the exact title, and get immediate access to the journal's web page.

3. Complete literature search and ongoing updates for specific research topic

Drusilla user is a seasoned veteran of CDL, a faculty member working on a major paper. It is important that she is kept up to date and not miss anything that is published on this topic.

Overview | Task Analysis | Suggested Solution | Experiment Outline

Updated: March 1, 1999