Nori UserDictionary picks a user noun's right context id (last syllable with a final consonant, without one, or not Hangul) from the last character of the whole entry line instead of the noun, so a trailing space or an inline comment ('학교 # memo') changes how the word after the noun is analysed

Finding LFV-654Area: analysis / nori UserDictionary (right context id of user nouns, used by the Viterbi connection costs)Priority: lowImpact: silent wrong resultsNovelty: new

Finding page for LFV-654

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

What it shows

You load a nori user dictionary whose line for the noun 학교 ('school') ends with a space, or carries an inline comment as in '학교 # memo' (the parser deletes '#' to the end of the line but keeps the space before it), and expect the same analysis as with the plain line '학교', because the parser splits the line on whitespace and gets the same noun with no segmentation. Instead UserDictionary.java:94 takes the noun's right context id, which tells the tokenizer whether the noun's last syllable ends in a consonant, from the last character of the whole line, here the space, so the noun gets the generic id and the word after it is analysed differently: on 9.12.3 '학교이' gives 이/VCP with '학교' but 이/J with '학교 ', and KoreanAnalyzer turns '학교가방을 샀다' into [학교][가방][사] with '학교' but [학교][방][사] with '학교 ', losing the term 가방 ('bag'). The two line forms disagree in every release from 7.4.0 to 10.6.0 (10.6.0 by reading) and on main (main: '학교이' gives 이/JKS vs 이/VCP, '사람이나' gives [사람] vs [사람][이]), and a shorter segmentation such as '세종시 세종' takes the id from 종 (final consonant) instead of 시 (none) in the same way.

The test shows that a nori user dictionary line 사람 (trailing space) changes how KoreanAnalyzer analyses the text after the noun, compared with the line 사람, because UserDictionary takes the noun's right context id from the last character of the line.

Expected vs actual

Expected the same terms for "사람이나" with both lines ([사람]); actual expected:<[사람]> but was:<[사람, 이]> (an extra term), with -ea and -da.

Files

Run it

# 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/analysis/common/build/libs/lucene-analysis-common-$V.jar
CP=$CP:$LUCENE/lucene/analysis/nori/build/libs/lucene-analysis-nori-$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.analysis.ko.TestUserDictionaryTrailingSpace

The module(s) analysis-nori 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 2026-11-04. AI-assisted; not yet triaged by a human Lucene committer. Feedback: open an issue in lucene-ai-bug-hunt.