DLI social science team

DLI Social Science Team Home Page

send comments or questions to: l-neuma1@uiuc.edu

DLI Faculty Focus Group Summary

Note: This summary was prepared by writing up BishopÕs notes and then supplementing them with notes by Star and Ignacio. Reflected are all individual comments that were recorded, using participantsÕ own terminology as much as possible; conclusions attempt to summarize points of greatest importance and consensus. There is some redundancy as Uses lead naturally to ProblemsS to Suggested DLI features in individualÕs discussion on particular point.

Date: 11/14/94, 7:00-9:00 p.m.

Attendees:

1) Visiting Asst. Professor (Physics)
2) Professor (Physics)
3) Professor (Aero/Astro Engineering)
4) Professor (Computer Science)
5) Professor (Computer Science)
6) Associate Professor (Mech. and Industrial Eng)
7) Professor (Mech. and Industrial Eng)

Note: Only one participant was a woman. Not sure if prof is meant generically in some cases, or whether it really means full prof.

Moderator comments:

People arrived late and so we didnÕt really get started until around 7:15. A number of the participants arrived with a few notes; my strong impression was that each person really had about 3 major points they wanted to make about digital journal collections. The discussion began with participants asking a few questions about the DLI project. They were curious about the time frame, end result, nature of the testbed.

We did not follow the FG interview guide in a linear fashion. The conversation flowed freely and all of the main topics in the interview guide were addressed at one time or another. At first I was afraid that participants--given their technical backgrounds and the relative difficulty of revealing personal journal behavior--would concentrate on describing their views of the appropriate technical capabilities of the testbed. But they virtually always tied their technical comments to some kind of behavioral or functional rationale.

DISCUSSION SUMMARY

Purpose of journal use:

-- To see if anyone else has done what IÕm doing.

--To go in-depth on a topic if IÕm teaching a new course in an unfamiliar area.

--To break into emerging information... current awareness... finding out what others are doing.

-- To look at othersÕ methods and theories.

-- To see if others have same research interests.

--For student work: typical assignment is to find genesis of particular engineering problem and follow it through to present time.

-- To identify good exam problems.

--To identify all relevant material, far back in time, when writing review article.

-- Entertainment through browsing.

Nature of journal use (which pieces of document used and how):

--References very important; need to trace them backward and forward to identify related material.

--Heavy use of figures and tables. These provide most accurate indication of what the author really did. They are also a good tool for skimming to see what paper is really about. Serve as a translating mechanism, since different people and different systems use different keywords to describe work.

-- Use author, title, abstract to screen documents.

--Often print papers: to add own comments, organize and file in personal collection, take on the road.

--Often need multiple copies of specific piece, like a table, for class.

--Rarely need to read entire article carefully; only if teaching heavily from it or if it is extremely close to own work.

--Often need to transpose data presented to make it compatible with own work.

Problems in journal use:

--CanÕt trace citations forward; tracing backwards is too cumbersome.

--Identifying literature is very difficult. Often donÕt remember year, author, title. Sometimes remember the way the page looks, but canÕt search that way. Browsing shelves is frustrating and time-consuming because grouping doesnÕt always fit particular need and stuff is physically dispersed and hard to locate. Hard to find needed item, even in personal collection; stuff gets lost and multidimensional retrieval is difficult.

-- Hard to remember authorsÕ names.

-- CanÕt do specific search with Mosaic.

--Published literature is too old to be useful, in many cases. Difficult to identify and access emerging literature.

--Coverage vs. precision: want everything to be there, but to be able match very specific search request.

--Different computer formats for material make it hard to seamlessly search, capture, manipulate information.

-- CanÕt easily retrieve specific tables and figures.

--Hate IO+, IBIS and other indexing and abstracting systems (linear and slow) and Dewey Decimal System (too linear and general... couldnUt get people to elaborate on this).

--Thesaurae use terms that are out-of-date or just not useful.

--Hard to locate conference papers (would rather read than hear them). Difficult to assess their accuracy. Proceedings are very expensive.

-- Hard to find contact information for authors.

-- Hard to assess quality of published literature.

-- Hard to track material in unfamiliar field.

--Hard to get actual data that forms basis for published results.

--Citations are all formatted differently, so canÕt easily cut and paste into personal bibliography.

--Interfaces and retrieval and display options are inadequate and inflexible.

Search process (identifying and screening, organizing documents):

-- Browse.

-- Search for keywords.

--Search for publications by particular author; often want all recent work by particular person.

--Use author names, titles, abstracts to screen documents. Often move from title to abstract to full document.

--Scan home pages of companies to identify emerging information.

-- Scan tables and figures in document.

--Often locate material by remembering its physical look and feel, along with general sense of when it appeared.

-- Organization of material in personal collection:

--Heaped in piles; each pile represents particular chronological period.

-- Really important stuff is closest.

--Group into general topic categories; subdivide if gets too large. Different drawers or stacks for different topics.

--Organization is "thematic and contextual"; the categories are intrinsic to my work.

--No particular logic; but I can still find individual items.

--When I donÕt remember whatÕs in a pile any more, I throw it away.

-- Journals arranged by title.

--Search material in personal file by topic, agency/author, intended use (e.g., exam problem), method used.

--Need to mark papers in personal collection and then put them back.

--In library, use call nos. to get to right place on shelf; once a few relevant documents are found, use their citations to find more stuff.

--Document type groupings also used in library: company reports vs. journals vs. books.

--Heard paper at conference and then used ph to contact author to get hold of the paper (since it wasnÕt in IBIS). Typically have to go to conference to get access to conference papers.

--If I find a journal article that looks relevant, I photocopy it, skim it, and then file it in personal collection.

Digital library features and functions desired:

-- Must be extremely user-friendly.

--Ability to see real page images, exactly as they look in print publications. This is because people often locate material by remembering what it looks like.

--System should provide access to material in areas other than oneÕs own.

-- Include conference papers.

--Scan in individual documents on userÕs demand. One person praised Grainger for already doing this: he was at a conference and needed access to a document, so the library scanned it in and he got it over the net.

-- Trace citations backwards and, especially, forwards.

--Would like to be able to browse across several pages at once; seeing one page at a time is not enough.

--Mosaic as interface, but with specific search capability.

--"Digitize Britt Barry" (i.e., a favorite librarian who could find what you needed even if your request was "itÕs blue, the author is Smith, and I donÕt remember the year").

--Automatic accounting of history of everything an individual has looked at (often what you need is something you saw before, perhaps just one piece of a previously read paper. Sometimes you donÕt remember much about it except that you saw it before. Or, you might not remember much of a citation and you also donÕt remember that you saw it before, so the system could prompt you with items already retrieved in the past.

--Ability to step from "little info" to "more info" about a document by clicking, e.g., title to abstract to fulltext. A hierarchical display of document pieces is needed so that you can see whatever you want, but this should be flexible, determined on the fly.

--System should capture emerging information automatically by scanning home pages of relevant individuals and organizations.

--Alerting function: system should notify me when it has something in my subject area, by referring to my user profile.

--"Symbolic mapping" of what individual user needs (I think he meant as the interface; that there should be some graphic map so that documents were arranged in some user-defined, multidimensional way, e.g., by subject, method, author, intended use). Also used term "natural topology," and suggested that he might want all documents from a particular agency or all documents treating a particular topic. Interface should be graphical, like scanning shelves in library and provide similar sense of context, but should be more multidimensional.

--Ability to scan and retrieve images in a paper. A number of respondents felt that the tables and figures were a more accurate and useful surrogate for a document than were the abstract, conclusions, etc., or that the tables and figures were the most needed piece of info from the document.

--Print capability. DonÕt like to read on screen and need portability. Portability especially important for students.

--Search by author names. Often want to find work by particular person whom I know thinks like me. Include online directory of all publications by a particular author, if author agrees.

--Platform independence. Need seamless access to documents stored by different organizations in different formats (e.g., AIP uses LaTex).

--Ability to write notes on documents in the course of reading them.

--Some kind of improvement on "lexicon of keywords" (thesaurus and subject heading lists?), which are not useful... terms are too old and different people use different terms. Keyword lists need to be linked to user-built equivalence tables.

-- Ability to work from office.

--Link with other tools, e.g., Current Contents, Books in Print, handbooks, tables of integrals and functions, phone directories, personal URLs.

--Easy identification of relevant documents is more important than fulltext retrieval. Do not mind going to library to retrieve specific document; hate going to library to try to identify documents... that is the really hard part.

--Put authorsÕ names, addresses, email and phone numbers online (could be in separate file, i.e., neednÕt be linked to particular paper... just put phone and other directories online).

--Provide good documentation and make all of it available online.

--Individually-customized interface. Want to arrange things the way I want.

--Access to "secondary material" to help assess quality of document, e.g., othersÕ comments, reviews, no. of accesses.

--Access to pre-publication material, which typically is freely shared by community members. If itÕs published, itÕs probably already of no use to me.

--Graceful decay to weed, "forget" documents that havenÕt circulated.

--Access to numerical data behind figures (strong general agreement on this point)

--Live links: from references to the actual documents referenced; to proofs; to other versions of a document (e.g., conference report, full technical report, journal articles)

--Ability to paste bibliographic citations into own bibliography (so citations must be format-independent, e.g., translatable with something like BibTech).

--Contents should go back as far as possible... would like everything from last 100 years, need everything from last 15.

Other issues and comments:

--Authors must decide what other info to append to documents, e.g., interview, data, discussion.

--Privacy of authors and users must be maintained. Participants differed on whether they were eager to have their phone numbers listed.

--Intellectual property issues immensely important and complex.

--Need to be able to make 100 copies of single piece of article.

--Purpose of library is as single gateway to all information, translating among different platforms, providing service on information (which original producers donUt provide)

--Some people use ASCII terminals that donÕt support emacs option in IO. Access from home is difficult.

--CanÕt assume net literacy, even among computer scientists. One prof commented that Mosaic and gopher should be integrated and made easier; wasnÕt aware of lynx option.

--In own specialty, know how to identify people and documents. But need guidance in how to do this in unfamiliar areas.

--ducky@uiuc.edu dropped in. SheÕs interested in helping with the DLI project.

Conclusions:

--Individual customization (of interface, retrieval mechanisms, document presentation, etc.) was key theme. But I wonder how much effort an individual would be willing to put in? Seems like people should be able to do last step customization in a very intuitive manner.

--Tables and graphs and citations emerged as extremely important document features.

--People want online connections with all kinds of research tools (both additional content and various access tools).

--People want to create personal digital collection? This wasnUt noted explicitly, but personal paper collections are very important.

--System must be very friendly and simple to use; although computer literacy seems fairly ubiquitous, net literacy and information literacy are not. These cause lots of anxiety and frustration.

--System must provide means to move easily and flexibly from "little info" to "more info" about a document, i.e., to screen and zoom.