# WP21 — Search strategy (Phase B)

Date of searches: 2026-10-20. Period covered: publications dated 1970-01-01 to 2026-10-20.

# Sources actually queried (status recorded in reports/search_audit.md)

1. OpenAlex REST API (api.openalex.org/works, polite pool with mailto): full-text search (search=) and title/abstract search (title_and_abstract.search=), 1970–2026, all languages, cursor pagination, up to 10,000 records per query; fields: id, doi, title, authors, year, venue, type, language, abstract (inverted index), concepts/topics, OA status and PDF URL, cited-by count, referenced works (for backward snowballing).
2. Crossref REST API (query.bibliographic): used for metadata verification of every cited study and for the foundational studies; not used as a primary discovery source because it has no abstracts for most economics journals.
3. Semantic Scholar Graph API (api.semanticscholar.org/graph/v1/paper/search, no key): keyword searches, used for forward citations of the three foundational studies and of the main competing reviews (citation snowballing).
4. arXiv API (export.arxiv.org/api/query): econ.EM, econ.GN, cs.LG, cs.CV, stat.AP queries for machine-learning / vision / text valuation papers.
5. RePEc/IDEAS, EconPapers, SSRN, CORE, DOAJ, Google Scholar: IDEAS and CORE have REST endpoints that are attempted; SSRN and Google Scholar have no compliant bulk API and are not scraped (recorded as a coverage limitation).
6. Backward snowballing: referenced works of the eligible studies with the highest citation counts and of the foundational reviews; forward snowballing: citing works of Sirmans et al. (2005, 2006) and Rosen (1974) via OpenAlex cites:.

# Query set (English; translated variants in scripts/discovery/queries.yaml)

hedonic housing price(s); hedonic property valuation; residential property price determinants; housing characteristics and prices; implicit prices of housing attributes; real estate hedonic regression; spatial hedonic model(s); housing price meta-analysis; automated valuation model(s); text-based real estate valuation; computer vision property valuation; machine learning hedonic pricing; hedonic pricing house(s); house price hedonic; apartment price hedonic; rent hedonic; willingness to pay housing attributes hedonic. Translations: French (prix hédoniques logement / immobilier), Spanish (precios hedónicos vivienda), German (hedonische Preise Wohnimmobilien), Portuguese (preços hedônicos imóveis), Italian (prezzi edonici abitazioni), Chinese (特征价格 住房 / 房价 特征价格模型), Japanese (ヘドニック 住宅価格), Korean (헤도닉 주택가격), Russian (гедонистическая модель цены жилья), Turkish (hedonik fiyat konut), Polish (hedoniczny model cen mieszkań), Dutch (hedonische prijzen woningen).

# Deduplication

DOI (lower-cased, normalised); OpenAlex ID; normalised title (ASCII, lower, alphanumerics) + first author surname + year ±1; exact title match across sources; preprint/published pairs linked when titles match and years differ by ≤3.

# Records kept

data/metadata/openalex_records.jsonl (raw), data/metadata/records_dedup.parquet / .csv, data/metadata/search_log.csv (query, source, date, hits, retrieved), data/metadata/retrieval_failures.csv.