DLI social science team

DLI Social Science Team Home Page

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

DLI Grad Student Focus Group Summary

Bishop, Ignacio, Neumann, Star Date: 1/5/95, 7:00- 9:00pm

Note: This summary was prepared by writing up Bishop's notes and integrating them with the summaries prepared by Ignacio and Neumann to achieve completeness and consistency. All individual comments recorded in writing during the focus group are included in this summary and comments are recorded in the order they were made. Comments reflect the language of the participants as much as possible. Notes in square brackets reflect researchers' thoughts.

Attendees:

There were 4 male and 2 female participants. Departmental breakdown as follows: Physics (1), Computer Science (2), Aerospace Engineering (1), Mechanical Engineering (1), Computer Engineering (1).

Moderator comments: Discussion began late, due to late arrivals. The graduate students seemed to be more active information seekers and managers than either faculty or undergrads. They relied heavily on journals in their work and went to great pains to seek out articles and organize the personal collections that they subsequently created. They also seemed to spend more time manipulating (annotate, cut and paste) the information as they used and sharing material with colleagues. They could describe their activities and preferences in great detail. More precise and comprehensive searching and improved document manipulation seemed to be the most important DL features for this group.

Purpose of Journal Use [Students immediately began discussion of problems in the context of purpose of use]

--Need to do lit review when starting research project: find out what's been done (but some subtle stuff is not in the title or subject headings, so would be nice to search in the abstract, where different words are used).
--Title doesn't say what system paper is about. You wnat to search by the system name.
--Thinking of words is an art in itself.
--It'd be nice if the system suggested things.
--Also, terms are used differently in different disciplines.
--If you're working in a very specific field, you use journals like you would a textbook because there aren't any textbooks.
--To keep up with state-of-the-art technology: what people are doing, what's already been done, so you don't reinvent the wheel.
--Sometimes the article you found turns out not to be what you wanted, but the articles around it are.
--Also, it'd be nice to follow citations; need stuff back in time.
--I go to conference proceedings for the latest stuff, so use journals for background .
--I once browsed a journal for hours because I'd seen so many cites to it, but never looked at it in depth.
--Would like to identify from 40-50 articles, what is state-of-the-art, e.g., current highest speed. When writing a paper, you have to say in the intro what's done in the broader field, give benchmarks. You're setting the context, providing background info.
--You write paper for a particular journal, so you want to see the style for that journal.
--[Use to find facts?] Yes, I get the equation so I don't have to derive it. Somebody usually knows which paper has the equation I need.
--I look for specific surface tensions, experimental measurements.
--Recently looked for best efficiency for electric motor... had to just search term "electric motor."
--Tried to find baseline design.
--Needed info often provided by someone in same class; not sure how originally obtained.

Nature of Journal Use

--Look at abstract, conclusions, results to find "figure of merit" (need to compare own work on specific point)
--Scan article looking for particular word. This is hard if important words aren't in caps or italicized. It's easier online with Mosaic, but can't read on screen on old systems.
--Read abstract first to see if paper is interesting and relevant.
--Read abstract, then intro, then conclusions. I like to see the whole page; tells me how they lay out their argument.
--And figures... after title, abstracts. Figures tell me the real gist.
--When journal first comes, I like to scan the whole issue quickly to see what's in there.
--Serendipity of journals next on shelf is important.
--ToC is very important. Look for author names, because know who's important in my field, who writes in a way I can understand.
--When a particular interesting journal first came to my attention, I went to the library and scanned years of the ToC.
--Abstract is important: if interesting, keep reading til too boring. Then start skimming the figures. But can't highlight on the library's copy as I go along; could do this online.
--First I get the feel, but might want to finish reading first, then go back to refs. So highlight and bookmark as I go along.
--Sometimes the article is bad, but the references are great.
--Mosaic is too linear. I want something more like a tree, or trail, to see all what was interesting and how it is related. Ability to save search is important; often don't remember anything about a needed document except which search I found it in. [Advantages of electronic documents?]
--Can use search engine to find things more quickly, but need to know exactly what you are looking for and exactly how it would be presented.
--IEEE thing's text is too small, can't scan on screen, much less read. Have to print it out. [Organization of personal collection?]
--Alphabetically by first author. Doesn't work well if second author.
--You remember authors' names??
--I make lots of copies. Like to write on them. And also come back from new perspective. I file by author, some by topic if accessed more regularly. Plus I have an online bibliography which also has my annotations.
--Keep computer file of downloaded references and http addresses.
--File by author, journal title, year. I record documents in my online bib if personal copy made, where it's kept or found, keywords, and a summary.
--I want multiple copies or overlays; then could start fresh or work from original scratches.
--If in unfamiliar areas, I just add question marks.
--Papers organized by the project they're related to.
--In drawers by subject, initially. When you realize how you're using the info, then you make additional folder, e.g., if doing work for Rockwell, pull all papers that mention Rockwell.
--File by date.
--File by issues discussed, theoretical approach.
--It'd be nice to make own hypertext links, and give colleagues links to it.
--File by system type.
--I staple papers right into my lab notebook, so I don't lose them.
--I use flagging between related systems.
--Some articles talk about 2 things, so you want it in both categories. Or only a piece needs to be in another category.
--So can't find some things, because paper starts in subject file, then I move it for particular purpose.
--I mark and search for catchy quotes. Index first by what quote is about, because you don't remember what paper and author are associated with those "hmmmm..." phrases you jotted in the margins.
--Basically file by keywords, but those aren't the words I wrote on the folder. I use my own keywords, e.g., acronyms for different types of materials... "type x transistors" "phosphite" and I'm looking for articles with both.
--Acronyms very problematic for searching.
--Would like my keywords to be matched against incoming journals.
--Yeah, keep my profile and history.
--But SDI [selective dissemination of info: running user's query periodically and sending her the results] isn't valuable if you can't specify 2 keywords. My research interest is too diverse and interdisciplinary, so I just scan the new journal shelf.
--I file by keywords--but they change!

Problems

--Info needed on something that would not appear in the title; need to search text.
--Need access to thesaurus to find related terms--a synonym finder.
--Words have different meanings in different fields.
--Finding specific measurement or formula; can spend hours looking.
--Titles don't give enough info.
--Can't tell what's most up-to-date info--what has or has not been disputed by others in later articles.
--Seemed to be anxiety about missing important information... don't know how thorough search is, how extensive database is]
--I want to use subject terms, but might only be interested in theoretical or practical aspects; hard to get the right words.
--Hard to copy bound volumes.
--When new to library and don't know basic arrangement, need call no. Also, get too many hits for IEEE Computer.
--Extensiveness of what's indexed: IEEE system has only IEEE stuff; conf proceedings to recent, ephemeral to be included.
--Sometimes conf paper not even in the proceedings; in that case, having contact info for author would help.
--Need complete citation info on screen so you can reference it.
--Can't expand search without starting over.

Other Information Sources Important in Work

--Theses.
--More details than in journal, so need tech reports.
--Equipment spec sheets.
--Expanded Thomas Register.
--Technology overview books: collections of article.
--Need to include errata, corrections.
--Subsequent reactions, letters, comments.
--Government standards.

Ideal Digital Library Features

--Want to search forward, find when article cited.
--Reference links--great to get whole article, but would also like just basic reference info, i.e., where it is, if checked out, abstract, call no.
--But have to have mechanism to tell you where you are and have the option to go back.
--Video to get look and feel of systems you're studying.
--Digitize graphs, so you can manipulate the data.
--Copy just specific sections.
--Ability to use from office.
--Ability to do what I want with bibliographic data.
--Several browsing levels: shelf, ToC, paper.
--Access yearly indexes.
--Browse for keywords. Want automatic thesaurus list of suggestions, use these as main words or modifiers, but also have option to never see the words, have the computer just add them into the search.
--Design own frontend.
--Use real thesaurus in word processor because, keyword searching so bad.
--Gave up on keywords: sometimes paper in different field and different term used. Just follow cites.
--Finding right acronym.
--Type in few letters of keyword, then see thesaurus matches.
--Have to ask colleagues for current, system terms.
--Use keywords from found articles.
--Should be able to choose either broad or narrow searches.
--Ability to access Library of Congress indexes.
--Find frequency of citations, how many time certain article has been cited.
--Ability to save searches automatically.
--Want to be able to read text while graphics fill in. Mosaic takes too long, Netscape better for this.
--Don't limit search by discipline; often you need to span disciplines.
--Include movies of conference proceedings so you can actually see presentations and the discussion afterwards.
--Be able to highlight parts of text, take notes and have that saved with the articles--as a sort of overlay, BUT have option to suppress those notes and take new set, or add to old set, or switch between sets.
--Make your own hypertext links, own homepage to related topics.
--Be able to search "excepts" ... e.g., find topic except when something else is also in the article.
--[Need several browsing/searching features: by topic, journal, date]
--Computer should be able to cross-tabulate references... which refs appear in all these articles?
--Ref links more important than keywords, but just following refs could give too narrow an outlook on the field.