Invest in Data, Speculate on Interfaces: The Long-Term Strategy of the HCI Bibliography

Gary Perlman, director@hcibib.org, http://www.hcibib.org

Note: Red and *'s indicate "hotness".

How the HCI Bibliography Started

The idea for the HCI Bibliography started with an online bibliography for a curriculum module on UI Development prepared for the Software Engineering Institute in 1988. I thought the resource was useful because it had abstracts of articles and tables of contents of books. It had 196 records. I had the idea that a distributed group project could coordinate gathering bibliographic records, possibly with authors contributing abstracted entries out of self-interest. Although that did not work for a variety of reasons, people were generally supportive of the idea, and the project started releasing files in 1992 primarily based on work-study student and later OCR input.

Another Early Effort

Around the same time, the HILITES project was trying to move from a research project into a commercial "Information Service for the World HCI Community" (Shackel et al, 1992). I sent a survey to about 500 registered user/sites and concluded that HILITES was priced out of the reach of most users, especially when matched up against a free service like the HCI Bibliography (even though it offered less coverage and fewer services). HILITES disappeared.

Shackel. et al (1992) "HILITES -- The Information Service for the World HCI Community" SIGCHI Bulletin, 1992 24:3 40-49.

HCIBIB.ORG

Initially, the HCI Bibliography was simply a database at Ohio State University; there was no internal support to provide the resources for an online search service. Services occasionally were offered, but usually without updating the data, and people complained to the project. ACM SIGCHI, a long-time supporter of the project, agreed to fund a site, which I requested on its own domain (hcibib.org) in case the database and service ever had to move (and to maintain some legal distance between the project and ACM). A search service, based on the glimpse search engine, started in April of 1998.

HCIBIB Promotion and Usage

The HCI Bibliography has been promoted primarily through informal announcements (magazines, newsletters, Usenet, listserv). It is listed, although often incorrectly, on a variety of services like Yahoo. The project home page has received about 100 "hits" per day, and the search service has processed over 5000 searches per month. The main conclusion one can draw from the search logs is that most searches are not nearly as effective as they could be (and this probably generalizes to ineffective use of other search services).

HCIBIB Organization *

The advent of the Web during the life of the HCIBIB simplified the delivery issues for the data and supporting files. Still, it is a fundamental principle that the data files (journal volumes, conference proceedings) be independent of how they are delivered. The data files remained in their original extended UNIX Refer format, but could be reformatted and displayed in HTML. Software tools on the server side could readily manipulate and summarize the data (e.g., reports of the most frequent authors, number of publications by year, tables of journals by years, etc.).

Goals for the Future **

The goals for the near future of the hcibib.org is to better integrate itself into the research needs of the HCI community. This has already started with increased access to the world's HCI publications: Integrating links to Web resources and to full text do not really require much insight or expertise. They are simply information to be added to records (although potentially more volatile than information about more traditional forms of publication).

More difficult to estimate are the ways of better supporting research in HCI, itself a research area far broader than HCI. Some active areas of work include:

Measuring Success *

Measuring success in these areas is difficult. Subjective satisfaction may be high even though important information has been missed (either inside or outside the database coverage). There is also the of making X% of the information so much more accessible than the other 100-X% that the remainder is never found by many researchers.

Conclusion **

Search engines, data visualizations, and the like will continue to improve, but many users are not finding the information they hope for, and they are not using it as effectively as possible. Long-term, the data in the HCI Bibliography will survive and be useful for a variety of purposes. If an online resource hopes to remain useful, it must separate data from its use.