{
  "version": "2026-05-21",
  "description": "Strict phrase and alias rules used to filter OpenAlex candidates into the TSFM corpus. A candidate is admitted iff its (title + reconstructed abstract) text, lowercased, contains at least one of these patterns. Phrases with required_co_occurrence must additionally contain at least one of the listed terms.",
  "strict_phrases": [
    {"pattern": "time series foundation model"},
    {"pattern": "time series foundation models"},
    {"pattern": "foundation model for time series"},
    {"pattern": "foundation models for time series"},
    {"pattern": "pretrained time series"},
    {"pattern": "pre-trained time series"},
    {"pattern": "large time series model"},
    {"pattern": "time series pretraining"},
    {"pattern": "time series pre-training"},
    {"pattern": "timegpt"},
    {"pattern": "timesfm"},
    {"pattern": "lag-llama"},
    {"pattern": "lag llama"}
  ],
  "phrases_with_context": [
    {"pattern": "moirai", "required_co_occurrence_any": ["time series"]},
    {"pattern": "tsfm", "required_co_occurrence_any": ["time series"]}
  ],
  "candidate_query_terms": [
    "\"time series foundation model\"",
    "\"time series foundation models\"",
    "\"foundation model for time series\"",
    "\"foundation models for time series\"",
    "\"pretrained time series\"",
    "\"pre-trained time series\"",
    "\"large time series model\"",
    "\"time series pretraining\"",
    "\"time series pre-training\"",
    "\"TimeGPT\"",
    "\"TimesFM\"",
    "\"Lag-Llama\"",
    "\"Moirai\"",
    "\"TSFM\""
  ],
  "date_range": {"from": "2020-01-01", "to": "2026-05-31"},
  "candidate_pool_size": 3009,
  "admitted_corpus_size": 679,
  "arxiv_indexed_subset": 360,
  "known_gaps": [
    "Papers introducing a TSFM under an unanticipated name without using any TSFM-specific phrase.",
    "Named-model-only papers (e.g., a Chronos benchmark paper that does not also use one of the strict phrases) — these are caught only when the model name co-occurs with a strict phrase from another release in the same paper.",
    "May 2026 partial: OpenAlex indexing lag of 2–8 weeks for new arXiv submissions.",
    "ICLR 2026 conference-proceedings versions: not yet indexed in OpenAlex as of 2026-05-21."
  ]
}