BayesianScoreQuery raises the minimum competitive score to at least 1e-7 before passing it to its inner query, so default top-k search silently drops the best hits when every competitive calibrated score is below 1e-7

Finding LFV-164Area: search / top-k dynamic pruning (BayesianScoreQuery: BayesianScoreScorer#setMinCompetitiveScore)Priority: lowImpact: silent wrong resultsNovelty: new

Finding page for LFV-164

AI-assisted finding, confirmed by automated checks, not yet triaged by a human Lucene committer. This site files nothing upstream.

What it shows

You run an ordinary top-k search such as searcher.search(query, 10) with a BayesianScoreQuery whose calibrated scores (probabilities) are all below 1e-7, and more than 1,000 documents match. You expect the same top hits as an exhaustive search, because the scores are a monotone transform of the inner BM25 scores and explain() reports exactly those scores. Instead, once the collector starts pruning, BayesianScoreQuery rounds the requested minimum score up to 1e-7 before turning it into a BM25 threshold for the inner query, so the inner query skips documents that are still competitive and the search returns weaker hits (in the repros, in effect the first matches in doc-id order) with no error.

The test shows that BayesianScoreQuery raises the minimum competitive score to 1e-7 before passing it to its inner query, so a default top-k search drops the best hit when every calibrated score is below 1e-7 and more than 1,000 documents match.

Expected vs actual

Expected search(q, 1) to return the exhaustive top hit, doc 5000 (1.8876845E-8); actual: doc 3900 (1.16228565E-8) (expected:<5000> but was:<3900>), with -ea and -da.

Files

• src/org/apache/lucene/search/TestBayesianScoreTopScores.java (JUnit test)

Run it

• Tested on Lucene main e1faeba08bde (2026-11-02): still reproduces.
• Tested on Lucene main 645d4fe14a20 (2026-10-28): reproduced.
• Recorded outcome of run 1 on Lucene main 645d4fe14a20: 1 test(s) run, 1 failed.
• How to read the outcome: a failing test is the finding reproducing (the failure message states expected against actual).
• Seeds: a rerun with the same seed replays the same random choices. Assertions off (-da -dsa -Dtests.asserts=false) is the production-like setting.

# Lucene checkout at the tested commit, built with: ./gradlew --no-daemon assemble   (JDK 25)
LUCENE=/path/to/lucene; V=12.0.0-SNAPSHOT
CP=$LUCENE/lucene/core/build/libs/lucene-core-$V.jar
CP=$CP:$LUCENE/lucene/codecs/build/libs/lucene-codecs-$V.jar
CP=$CP:$LUCENE/lucene/test-framework/build/libs/lucene-test-framework-$V.jar
CP=$CP:randomizedtesting-runner-2.9.2.jar:junit-4.13.2.jar:hamcrest-3.0.jar   # test-framework dependencies (Maven Central)
CP=$CP:junit-platform-console-standalone-6.1.3.jar   # JUnit Platform console launcher (Maven Central)

javac -d classes -cp "$CP" $(find src -name '*.java')

# run 1: assertions on, seed DEADBEEF
java -ea -esa --add-modules jdk.incubator.vector --enable-native-access=ALL-UNNAMED -Dtests.seed=DEADBEEF \
     -cp "classes:$CP" org.junit.platform.console.ConsoleLauncher \
     execute --select-class org.apache.lucene.search.TestBayesianScoreTopScores

Gradle alternative: copy the test into the matching module's src/test/ tree of the checkout and run ./gradlew -p lucene/<module> test --tests <class> -Ptests.seed=<seed> (add -Ptests.asserts=false for a run without assertions).

Feedback

Open an issue in lucene-ai-bug-hunt. Please do not open issues or pull requests in apache/lucene about this finding unless a Lucene committer agrees it should be filed.

Snapshot of 2026-11-04. AI-assisted; not yet triaged by a human Lucene committer. Feedback: open an issue in lucene-ai-bug-hunt.