Conference Reports

Science and Technology Research Forum

I. "Stretching the Dollars for library Materials in Science and Technology in Illinois"

Powerpoint slides from the talk are available on the web at http://ux1.cts.eiu.edu/~cfvab/ALA/ [Note: Broken link removed 1/2/03 by aduda@istl.org]

The speakers explained the history and purpose of the Illinois Cooperative Collection Management Program (CCMP), a consortium of 70 academic libraries.  The program's goal is to enrich and strengthen the collective information resources available to the consortium libraries and the citizens of the State of Illinois.  Controlling principles of the program are universal benefit to all participants, open physical access, contractual responsibility, local sufficiency, universal input, and multiple funding modes.  Work in the CCMP began in 1994 through a pilot project called "Model Disciplines" which was an effort to get subject librarians to examine goals for cooperative collection development.  The goals echo traditional cooperative collection development ideas including: building research level collections, documenting strengths in participants' collections, eliminating duplication, carrying out in-kind contributions, finding support for new serial titles, and establishing a last copy repository. From fiscal year 1995 through fiscal year 1998, the CCMP has spent $389,700 in disciplines of education, biosciences, materials technology, gerontology, and philosophy.

Problems surfaced in choosing which disciplines would participate.  Research collections are defined differently from what the CCMP calls "pockets of excellence" developed in smaller libraries.  Such libraries did not want to be limited by the CCMP's policy of non-duplication. 

Two meetings of two days each were held to foster collection partnerships and enable subject librarians to get to know one another.  Program managers hoped librarians would share information and draft written agreements such as long-term collecting commitments, last copy commitments, plans for evaluation, and descriptions of open access policies.  Co-authoring a grant proposal for program funding and developing web pages are other projects partners could do together.

Benefits of the program include enabling individual institutions to concentrate purchasing in selected subjects, local savings in other areas, identification of new journals, and the ability to purchase expensive titles.

The CCMP sought volunteers to continue work in science model disciplines both for continued collection building as well as for consortial journal purchases. Biosciences and materials technology emerged as areas for CCMP support.  One of the speakers' slides lists the schools participating in the bioscience effort with specific areas they agreed to support.  Six institutions participated in the first year, and two others joined for the second year.  The materials technology area had circulation and interlibrary loan statistics included in its evaluation plan. A database of monograph and journal titles held by participating institutions was posted on a Web site.  Slides from the presentation indicate the number of titles purchased with CCMP grant support in each discipline, the amount of money spent, average cost per title, the number of unique titles owned by each participating institution, and number of titles owned by only one other institution in the consortium.

Increasing costs of scientific and technical journals were addressed through consortial examination of titles held, cooperative purchase of some journals and databases, and cooperative deselection.  The CCMP purchased Academic Press's IDEAL Full Text service for access to scientific journals.  Fourteen libraries ranging from community colleges to large universities participated.  At least one of the institutions subscribed to each of 143 titles.  thirty-six of the titles were not owned in print by any of the participating libraries.  Electronic access to full text allowed duplicate titles to be canceled by participating libraries.  A spread sheet was distributed to participating libraries for annotation to indicate willingness to commit to maintaining the print subscription indefinitely, titles they had already canceled, or willingness to subscribe tot titles held by no other library in the consortium.  Of the total 179 journals, 66 had archive commitments from 7 participating libraries; only 7 of those were duplicate titles.  Fifty titles had been canceled or considered for cancellation by one of the libraries.  One institution expressed willingness to subscribe in print to two titles.

The major benefits of the CCMP are strengthened collections overall as a result of participating libraries' commitments to carry out intense collecting in selected disciplines and last copy agreements.  Communication among subject librarians throughout the state of Illinois increased and improved.  The consortial purchase of IDEAL allows consortium participants to cancel 211 journals freeing funds for other uses.

II.  "The Work demands and Information Seeking Behavior of Ph.D. Physics Students"

Kevin set out to examine characteristics of Ph.D. students in physics according to common types such as precandidates, instructors, and candidates.  The first question he asked is, "How does information seeking behavior of graduate students vary by role?"  He reviewed differences in demands of each stage and sought to discover how students use technology, people, and the library to get information in each one.

Reviewing existing literature, Kevin found a few reports (see citations below), and he characterized the authors' approaches.  Generally, they carried out quantitative studies using survey instruments.  They did not investigate behavior in light of the work environment the students anticipated for themselves or that faculty anticipated for their students.  The studies heavily emphasized use of library resources and services rather than attempting to discover all resources graduate students use; the studies did not explore uses of technological tools for information gathering.  The investigators did not integrate the impact of work demands with analysis of information seeking behavior. [the three citations are 1) Farid, M. 1984. A study of information seeking behavior of Ph.D. students in selected disciplines.  Final report. ERIC Document 252213; 2) Parrish, M.M. 1989a. Academic community analysis: discovering research needs of graduate students at Bowling Green State University. College and Research Libraries News. no. 8: 644-646; 3) Parrish, M.M. 1989b. Analysis of graduate students research at Bowling Green State University. ERIC Document 309771]

Kevin's approach involved qualitative research using structured interviews which provide what he calls "rich" data and an integrated account of students' experience.  Interview scripts included 18 to 26 questions; Kevin taped interviews, which averaged one hour in length.  He used transcriptions of the interviews and assigning codes to blocks of text in order to find patterns and quotations. 

Core interview questions:

• Please tell me a little bit about the work involved.  I am interested in any demands that teaching [or taking classes or conducting research] places on your time.
• Describe your most successful means of getting information.
• What roles do computers/technology play in how you get information?
• What roles do people play in how you get information?
• How do you use the library to meet the demands present in your role as precandidate [or instructor or candidate]?

The population interviewed was seventeen graduate students in the Department of Physics at the University of Michigan; 5 were precandidates; 5 instructors; 7 candidates.  Fourteen male students were interviewed, and that number represents 16% of the number of males in the physics graduate student population (89); three women were interviewed, 17% of the population of 18 female physics graduate students.

Kevin described the activities of precandidates, instructors, and candidates and the nature of demand their work represents, and the information sources they use.

Precandidates take three to four classes each term and often pursue basic courses on theoretical principles in their first year and specialized, advanced courses in their second year.  Basic courses usually involve work on problem sets; they tend to be complex and time intensive.  Advanced courses focus on latest research in an area, problem sets encountered in them are less theoretical, and students are often required to a write a paper or give a presentation.  Precandidates rely on their books, notes and personal thought for information.  They make limited use of technology and the library, and other graduate students and faculty are important.  Kevin was surprised not to hear second year students mention using INSPEC and other electronic tools.

Instructors spend one fourth of their time as teaching assistants, typically for two classes each term.  The courses are usually in electricity/magnetism and mechanics, and they work with undergraduate laboratory sections.  Content for teaching comes from the lab manual.  Instructors' work involves reviewing experiments, preparing handouts and quizzes, in-class demonstration of experiments and explanation of concepts, responding to students' questions, grading, and office hours.  Most instructors do not teach undergraduates to search for information, whether using the library or the Web. Information sources instructors rely on include their own personal knowledge and manuals for the course and instructor.  Kevin detected no use of the library and only limited user of technology and e-mail.  Fellow graduate students are important as well as previous instructors and faculty supervisors.

Precandidates and instructors appear to seek information internally from their personal knowledge, notes, class and lab materials, and fellow classmates or instructors.

Kevin found it difficult to generalize about candidate's experience since work in theoretical or applied physics varied along with work groups and laboratory settings.  Demands and information needs change.  Differences among candidates depend on area of concentration.  The research group has a distinct impact on the way they get information.  Kevin would like to have had more subjects for his investigation.

The work demands on candidates are primarily learning and reading, building and maintaining equipment, collecting and analyzing data, and writing up results.  They are the most productive of the types of physics graduate students and regularly present at departmental seminars/colloquia and conferences.  They reported relying on their advisors and fellow graduate students for information.  They also indicate that computers are integral to their work not only for data gathering and analysis but also for email collaboration, www searching, access to the library catalog and INSPEC.  All the candidates Kevin interviewed reported using the library for books and journals.  They view the library as a storage facility, use the online catalog and INSPEC via remote access, and stay only briefly in the library when they visit.

Summary points are that no use of the library was detectable among precandidates and instructors.  Candidates' view the library as a storage facility.  Their access of the catalog and INSPEC from outside the building reiterates the need for outreach in the electronic environment.  Kevin detected frustrations in candidates' lack of knowledge about searching; they lacked the option to ask for help because they rely on remote access.

III.  "An Investigation of the Efficacy of Intelligent Agents in Collection Development"

Hypothetically, Web intelligent agents should allow searchers to find a greater number of sources in less time and more easily than traditional search techniques.  Jennifer reviewed what intelligent agents are, described the way she chose an agent and other search engines to test, and explained the methods she followed.

First, Jennifer reviewed for the audience Cornell's criteria for selecting Internet resources.  They follow principles similar to those applied to selection of any materials and examine sources for their scope, content, authority, and quality.  Cornell's librarians review 'Net resources title by title, not by entire sites.  They seek standard data formats and select resources which offer substantial content worthy of cataloging rather than those which supply only links to other sites.

Intelligent agents are created for use with dynamic systems of information like the web, not regular books, journals, and fixed information.  They are communicative and reactive, they interact with their users,  and are goal oriented, unbiased, and trainable.  Desirable features of intelligent agents useful for collection development would be desktop applications which search the entire Web rather than only specific servers.  Results they return would be organized with some descriptive information.  Agent software should provide the option of saving searches and results and should exhibit some intelligence in sites retrieved. The agent should remove duplicate sites and download the URLs retrieved.

The two systems Jennifer selected to compare thoroughly are an intelligent agent product called Autonomy and a meta search engine called Echo Search.  Important features Jennifer noted about Autonomy are that it is inexpensive ($50), it is trainable based on selections from searches it performs, and it overwrites search results as new searches are performed.  Echo Search uses seven different Web search engines and indexes, counts, and times its search results.

Jennifer began her study by comparing seven search engines and examined the top fifty items each one retrieved.  The topics she pursued were electronic journals in the biosciences, entomology, lepidoptera, and a tick known as ixodes scarpularis (not a member of lepidoptera).

The method Jennifer used to compare performance of the search engine, meta search, and intelligent agent was to conduct searches in two cycles.  Ensuring consistency among the searches using all three types of search assistants was difficult because the intelligent agent gauges its progress based on the time elapsed, and the other two by the number of sites retrieved.  Jennifer used search topics of increasing specificity and evaluated the results.  In Cycle I the intelligent agent and meta search engine retrieved less than the traditional search engines.

For search Cycle 2 Jennifer used the top two search engines from cycle 1 and analyzed 150 if the sites retrieved.  The intelligent agent ran for 150 minutes and retrieved a larger number of sites.  No useful sites for lepidoptera were retrieved.  The search engine retrieved fewer total sites.

Autonomy was useful for discovering hidden sites and turned up more sites the longer it was allowed to run.  It runs slowly, however, requires a lot of computer memory, comes with poor documentation, and performs inconsistently.

Echo Search likewise returned more results the longer it was allowed to run.  It identified duplicate sites retrieved, removed them from the results presented, and noted sites experiencing network errors.  Its results were inconsistent and did not always look at all the search engines it was requested to search.

In evaluating the project, Jennifer reported that very few of the sources retrieved were selected to be included in the library's collection.  Web searching is difficult to define, and comparing results from different tools is enormously time consuming.

We welcome your comments about this article. Please fill out this form for possible inclusion in a future issue.