What `cat.ademic` adds

cat.ademic is a thin domain wrapper around cat.stack that adds OpenAlex-based paper fetching plus academic prompt framing. You can:

Fetch papers from a journal or topic via OpenAlex (journal_name, journal_issn, journal_field, topic_name, topic_id) and classify them in one call.
Classify text you already have (abstracts, full text, or PDFs) as a plain character vector or file directory.

Everything else — supported models, output format, ensemble voting, batch mode — is identical to cat.stack.

Install

install.packages(
  "cat.ademic",
  repos = c("https://chrissoria.r-universe.dev",
            "https://cloud.r-project.org")
)
library(cat.ademic)

Classify abstracts you already have

abstracts <- c(
  "We use mixed-methods to study labor market outcomes for...",
  "This paper develops a formal model of bargaining under...",
  "A systematic review of 47 studies on educational interventions..."
)

results <- classify(
  categories = c("Empirical-quantitative",
                 "Empirical-qualitative",
                 "Theoretical-formal",
                 "Review/meta-analysis",
                 "Other"),
  input_data = abstracts,
  mode       = "text",
  api_key    = Sys.getenv("OPENAI_API_KEY"),
  user_model = "gpt-4o-mini"
)

Fetch papers from a journal

cat.ademic connects to OpenAlex — a free, open scholarly database — to fetch papers by journal, field, or topic. Set polite_email (your email) for higher rate limits.

results <- classify(
  categories   = c("Quantitative", "Qualitative", "Mixed Methods"),
  journal_name = "American Sociological Review",
  paper_limit  = 100L,
  date_from    = "2024-01-01",
  polite_email = "[email protected]",
  api_key      = Sys.getenv("OPENAI_API_KEY")
)

Or by ISSN for unambiguous journal identification:

results <- classify(
  categories   = c("Empirical", "Theoretical", "Review"),
  journal_issn = "0003-1224",                # AJS
  paper_limit  = 50L,
  polite_email = "[email protected]",
  api_key      = Sys.getenv("OPENAI_API_KEY")
)

Fetch papers by topic

OpenAlex auto-tags papers with research topics. You can pull all papers on a topic across journals:

results <- classify(
  categories   = c("Causal-identification", "Descriptive",
                   "Theoretical", "Other"),
  topic_name   = "climate change adaptation",
  paper_limit  = 200L,
  date_from    = "2023-01-01",
  polite_email = "[email protected]",
  api_key      = Sys.getenv("OPENAI_API_KEY")
)

Classify full-text PDFs

Pass a directory or a vector of file paths. cat.ademic extracts the text (or renders pages as images for vision models) and classifies:

# One-time: install PDF extras
# cat.stack::install_cat_stack(pdf = TRUE)

results <- classify(
  categories = c("Has-DGP-assumption", "No-DGP-assumption",
                 "Unclear", "Other"),
  input_data = "./papers/",                  # directory of PDFs
  mode       = "image",                      # rendered-page vision mode
  api_key    = Sys.getenv("OPENAI_API_KEY"),
  user_model = "gpt-4o"                      # vision-capable model
)

Summarize before classifying

For long full-text inputs, summarizing first can improve downstream classification quality (and reduce token cost):

summaries <- summarize(
  input_data   = "./papers/",
  description  = "Sociology articles",
  instructions = "Summarize methodology and key findings in 3 sentences",
  format       = "paragraph",
  api_key      = Sys.getenv("OPENAI_API_KEY"),
  user_model   = "gpt-4o-mini"
)

results <- classify(
  categories = c("Causal", "Descriptive", "Theoretical", "Other"),
  input_data = summaries$summary,
  api_key    = Sys.getenv("OPENAI_API_KEY")
)

Tips for academic work

Always set polite_email when fetching from OpenAlex — without it you’re throttled to a low rate limit.
Abstract vs. full text. Abstracts are cheap and fast; full-text classification (PDF input) is more accurate for methodological categories but costs more. Use abstracts for screening, full text for in-depth coding.
Cite OpenAlex if you publish on data fetched through it — see openalex.org for citation guidance.

Where to learn more

Full Getting Started guide: vignette("getting-started", package = "cat.llm")
Per-function reference: ?cat.ademic::classify, ?cat.ademic::extract, ?cat.ademic::explore, ?cat.ademic::summarize
OpenAlex docs: https://docs.openalex.org

Classifying Academic Papers

What cat.ademic adds