Methodology
CSLAC brings together public information about Computer Science at U.S. national liberal arts colleges — who teaches it, what they research, what they publish, and what courses are offered — into a single dataset. Everything here is assembled automatically from public sources; this page explains how, and where the numbers should be read with caution.
Which colleges are included
The starting point is the set of institutions ranked by U.S. News & World Report as National Liberal Arts Colleges. A college is included once we can identify a Computer Science department — or a combined department, such as Mathematics & Computer Science — with a public faculty listing and information about Computer Science as a major.
Faculty
For each college we read the department’s public faculty page and extract every listed member’s name, title, and — where available — a link to their personal or institutional web page. Titles are grouped into broad categories (tenured, tenure-track, teaching, visiting, and adjunct) so that roles are comparable across schools that describe them differently.
Only faculty whose work is identifiably in Computer Science are kept. In combined departments this classification is made automatically by reading each person’s public profile, so the boundary between “CS” and “not CS” is a best-effort judgment rather than an official designation.
Research fields
To describe what each person works on, we collect the text of their public website and Google Scholar interests and use a language model to infer a primary field and up to five research subfields, drawn from a fixed taxonomy of Computer Science areas. These labels are inferred, not self-reported.
Publications and citations
Publication records come from OpenAlex, an open catalog of scholarly works. We collect Computer Science papers affiliated with each college and match their authors back to our faculty list by name and institution. Each paper’s venue is tagged with its research area and, where we can match it, a conference ranking from CSRankings and CORE (A*/A/B/C), or a journal quartile from Scimago (Q1–Q4).
Citation counts and h-index come from Google Scholar for faculty whose profile we can confidently match and verify; for everyone else, we fall back to author metrics from OpenAlex. Because the two sources index different bodies of work, their numbers are not directly comparable.
Courses
Where a college publishes its course schedule or registration catalog, we collect the Computer Science offerings term by term. Each course title is automatically sorted into a subfield (or a general bucket such as core, seminar, or other) so that offerings can be filtered the same way as faculty and publications.
Degrees awarded
The count of CS degrees conferred by each college comes from IPEDS, the U.S. Department of Education’s higher-education data system.
Updates
The dataset is refreshed periodically — roughly quarterly. The date of the most recent refresh is shown in the header of the main dashboard, and the full dataset can be downloaded as JSON.
Limitations
This is an automated, best-effort snapshot built entirely from public data. It is not affiliated with or endorsed by any college, and it will contain errors. The most important caveats:
- Coverage and freshness. Faculty are read from department pages as they appeared on the day they were scraped. Recent hires and departures may lag, people listed only on other pages may be missed, and pages that change format may be captured incompletely.
- Who counts as CS. In combined departments, deciding who is a Computer Science faculty member is an automated judgment that will sometimes include or exclude people incorrectly. Emeritus, affiliated, and courtesy appointments are handled heuristically.
- Inferred research areas. Fields and subfields are inferred by a language model from public text. They can be wrong or incomplete, especially when someone has little public material.
- Profile matching. Google Scholar and OpenAlex profiles are matched by name and affiliation. Common names, name changes, and shared profiles can produce false matches or misses, and not everyone maintains a profile — so citation metrics are absent or understated for some.
- Point-in-time metrics. Citation counts and h-index are snapshots that change over time and differ between sources; treat them as approximate, not authoritative.
- Publication matching. OpenAlex’s subject filter and author disambiguation both miss some papers and include others; interdisciplinary work in particular is undercounted. Venue rankings are matched heuristically and are themselves imperfect proxies for quality.
- Course data. We can only see what each college’s catalog exposes publicly. Some systems show only currently registerable terms, some require a login, and formats vary — so course coverage is uneven, and the automatic subfield labeling is approximate.
Found something wrong? Corrections and suggestions are welcome on GitHub.