coop/rose-ash
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

search: boolean-filtered ranked search + 11 tests
Some checks failed
Test, Build, and Deploy / test-build-deploy (push) Failing after 40s
Details

Browse Source 
searchRankTfIdf/searchRankBm25 parse a boolean query, filter docs via evalQuery,
then rank survivors by relevance over the query's leaf terms (queryTerms) — the
filter-then-rank pattern. 225/225.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
giles
2026-06-06 23:58:37 +00:00
parent cfa68c3db3
commit db2a5dc6ab
7 changed files with 98 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
lib/search/rankq.sx Normal file
View File

@@ -0,0 +1,11 @@
;; search boolean-filtered ranked search — Haskell source fragment.
;; Depends on parse (parseQuery/Query), query (evalQuery), rank (tfidfDoc/bm25Doc/
;; cmpScore). Filters by the boolean query, then ranks the surviving docs by
;; relevance over the query's leaf terms — the real-world filter-then-rank pattern.
;; queryTerms :: Query -> [Term]
;; searchRankTfIdf :: String -> Index -> [DocId]
;; searchRankBm25 :: Float -> Float -> String -> Index -> [DocId]
(define
search/rankq-src
"queryTerms (Term t) = [t]\nqueryTerms (And a b) = queryTerms a ++ queryTerms b\nqueryTerms (Or a b) = queryTerms a ++ queryTerms b\nqueryTerms (Not a) = queryTerms a\nqueryTerms (Phrase ts) = ts\nmkSubPair f terms idx d = (f terms idx d, d)\nrankSubsetWith f terms docs idx = map snd (sortBy cmpScore (map (mkSubPair f terms idx) docs))\nsearchRankTfIdf s idx = let q = parseQuery s in rankSubsetWith tfidfDoc (queryTerms q) (evalQuery idx q) idx\nsearchRankBm25 k1 b s idx = let q = parseQuery s in rankSubsetWith (bm25Doc k1 b) (queryTerms q) (evalQuery idx q) idx\n")