languagetool-org/languagetool · error · RuntimeException

Not found in set '

Error message

Not found in set '

What it means

getConfusionString looks up the confusion-set entry that case-insensitively equals the token from the text. This RuntimeException is thrown when no entry matches, meaning the token extracted from text is not present in its expected confusion set.

Solutions

1. Inspect the token and set in the message; verify the token's exact casing/content.
2. Check the confusion data for the missing word and re-add it if removed.
3. Adjust token normalization before lookup if case/locale transformations change the token.

Example fix

// before (data updated but word removed)
// confusion set: {there} while text token is 'their'
// after
// re-add to confusion data:
their

// pre-check that the token appears in the confusion set before matching
boolean present = set.stream().anyMatch(s -> s.getString().equalsIgnoreCase(token));
if (!present) log.warn("Token not in confusion set: " + token);

try { rule.match(...); } catch (RuntimeException e) { if (e.getMessage().startsWith("Not found in set")) { log.error("Token missing from confusion data: " + e.getMessage()); } throw e; }

When it happens

Trigger: stringFromText extracts a token (often uppercase or differently inflected) that does not case-insensitively equal any confusion-set entry, then getConfusionString fails.

Common situations: Case conversion differences (e.g. sentence-initial capitalization combined with all-uppercase tokens); language data updates removing a word still matched by the rule pattern.

Understand the failure class

Background: Record Not Found Errors: "not found", RecordNotFound, and "was not found" — what they mean and how to fix them — this error's family across 28 libraries.

Related errors

• Confusion set must be of size 2:
• grams must be between 1 and 5:
• No alternative found for:
• AbstractAdvancedSynthesizerFilter only works with pattern…
• beolingusFile not found:

AI-assisted analysis of languagetool-org/languagetool@2e990059ce (2026-09-06). Data as JSON: /api/errors/efd850f5fef910ae. Report an issue: GitHub.

Appendix: source

Thrown at languagetool-core/src/main/java/org/languagetool/rules/ngrams/ConfusionProbabilityRule.java:318

    return getBetterAlternativeOrNull(token, tokens, other, factor);
  }

  private ConfusionString getAlternativeTerm(List<ConfusionString> confusionSet, GoogleToken token) {
    for (ConfusionString s : confusionSet) {
      if (!s.getString().equals(token.token)) {
        return s;
      }
    }
    throw new RuntimeException("No alternative found for: " + token);
  }

  private ConfusionString getConfusionString(List<ConfusionString> confusionSet, GoogleToken token) {
    for (ConfusionString s : confusionSet) {
      if (s.getString().equalsIgnoreCase(token.token)) {
        return s;
      }
    }
    throw new RuntimeException("Not found in set '" + confusionSet + "': " + token);
  }

  private ConfusionString getBetterAlternativeOrNull(GoogleToken token, List<GoogleToken> tokens, ConfusionString otherWord, long factor) {
    String word = token.token;
    double p1;
    double p2;
    if (grams == 3) {
      p1 = LanguageModelUtils.get3gramProbabilityFor(language, lm, token, tokens, word);
      p2 = LanguageModelUtils.get3gramProbabilityFor(language, lm, token, tokens, otherWord.getString());
    } else if (grams == 4) {
      p1 = LanguageModelUtils.get4gramProbabilityFor(language, lm, token, tokens, word);
      p2 = LanguageModelUtils.get4gramProbabilityFor(language, lm, token, tokens, otherWord.getString());
    } else {
      throw new RuntimeException("Only 3grams and 4grams are supported");
    }
    debug("%.90f <- P(" + word + ") \n", p1);
    debug("%.90f <- P(" + otherWord + ")\n", p2);
    return p2 >= MIN_PROB && p2 > p1 * factor ? otherWord : null;

View on GitHub (pinned to 2e990059ce)