QualityStats.average(stats).getPrecisionAt(n) is not the mean per-query precision at n when n is above the mean result count: the mean that average() stored is replaced by a second extrapolation (mean P@10 0.4 instead of 0.6)

Finding LFV-935Area: benchmark / search-quality evaluation (QualityStats)Priority: lowImpact: minorNovelty: new

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

What it shows

You evaluate search quality with QualityBenchmark, average the per-query results with QualityStats.average(stats) and ask the average for its precision at 10, expecting the mean of the per-query P@10 values (which, like trec_eval's P_10, already count missing ranks as non-relevant). If the queries returned different numbers of results and their mean is below 10, getPrecisionAt(10) ignores the correct mean that average() stored and extrapolates again from the precision at the mean result count: with setMaxResults(10), a query whose 10 hits are all relevant (P@10 = 1.0) and a query whose only 2 hits are relevant (P@10 = 0.2) give 0.4 instead of 0.6. There is no error, and the printed SUMMARY never shows the affected cutoffs.

The test shows that QualityStats.average(stats).getPrecisionAt(10) for a query with 10 relevant hits (P@10 1.0) and a query with 2 relevant hits (P@10 0.2) is not the mean of the per-query P@10 values.

Expected vs actual

Expected 0.6 (the mean that average() stores); actual 0.4, because the averaged object's numPoints is the mean result count 6 and getPrecisionAt(10) extrapolates again from P@6, with -ea and -da.

Files

src/org/apache/lucene/benchmark/quality/TestQualityStatsAveragePrecisionAt.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=$CP:$LUCENE/lucene/benchmark/build/libs/lucene-benchmark-$V.jar
CP=$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.benchmark.quality.TestQualityStatsAveragePrecisionAt

The module(s) benchmark also need their third-party runtime jars; ./gradlew -p lucene/<module> dependencies lists them.

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 2026-11-03. AI-assisted; not yet triaged by a human Lucene committer. Feedback: open an issue in lucene-ai-bug-hunt.