What causes "Found a relation with a missing entity: ${p}" in stanfordnlp/CoreNLP?

Add the named node '=entity' to the Semgrex pattern (e.g. add "{...}=entity")

Found a relation with a missing entity

Error message

Found a relation with a missing entity: ${p}

What it means

KBPSemgrexExtractor.matches iterates SemgrexMatcher node matches and expects named nodes 'entity' and 'slot' in each pattern. If the 'entity' named node is absent from the matched graph, it warns with the pattern p and skips that match, treating it as a rule-authoring bug rather than a runtime failure.

Solutions

Add the named node '=entity' to the Semgrex pattern (e.g. add "{...}=entity")
Check the rule resources/patterns file for missing group names and fix the authoring
Log the offending pattern and remove/disable it if it's obsolete
Verify with a small Semgrex test that getNode("entity") returns non-null for intended matches

Example fix

// before
String p = "{tag:/NN.*/}=slot <nsubj {tag:/NN.*/}";
// after
String p = "{tag:/NN.*/}=entity <nsubj {tag:/NN.*/}=slot";

Defensive patterns

Strategy: validation

Validate before calling

SemgrexPattern p = SemgrexPattern.compile(pattern);
// ensure required named groups exist in the pattern string
if (!pattern.contains("=entity")) throw new IllegalArgumentException("pattern missing =entity: " + pattern);

Try / catch

SemgrexMatcher m = p.matcher(graph);
while (m.find()) {
  IndexedWord entity = m.getNode("entity");
  if (entity == null) { log.warn("no entity for " + pattern); continue; }
  // ...
}

Prevention

Always name both =entity and =slot groups in KBP Semgrex rules
Unit-test each rule on annotated sample sentences
Review rule resource files after edits for dropped group names
Log and quarantine patterns that repeatedly produce null nodes

When it happens

Trigger: A Semgrex pattern used for KBP extraction that does not (or conditionally does not) bind the 'entity' named node, e.g. pattern string missing '{word:/.../}=entity' naming, and the matcher finds a hit without that node.

Common situations: Hand-written or edited Semgrex rules missing one of the required named groups; rules copied from other pipelines with different group names; optional groups that fail to bind on some matches.

Understand the failure class

Background: "This is a bug, please report it": internal invariant violations, unreachable panics, and SNH errors explained — this error's family across 47 libraries.

Related errors

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/f3a1ceea42e4daf7. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/ie/KBPSemgrexExtractor.java:121

        tokens.get(i).setNER(input.subjectType.name);
      }
    }
    for (int i : input.objectSpan) {
      if ("O".equals(tokens.get(i).ner())) {
        tokens.get(i).setNER(input.objectType.name);
      }
    }

    for (SemgrexPattern p : rulesForRel) {

      SemgrexMatcher n = p.matcher(graph);
      while (n.find()) {
        IndexedWord entity = n.getNode("entity");
        IndexedWord slot = n.getNode("slot");
        if (entity == null) {
          // really this is a hideous bug, right?  these rules
          // should all have entity and slot set
          logger.warn("Found a relation with a missing entity: " + p);
          continue;
        }
        if (slot == null) {
          logger.warn("Found a relation with a missing slot: " + p);
          continue;
        }
        boolean hasSubject = entity.index() >= input.subjectSpan.start() + 1 && entity.index() <= input.subjectSpan.end();
        boolean hasObject  = slot.index() >= input.objectSpan.start() + 1 && slot.index() <= input.objectSpan.end();
        
        if (hasSubject && hasObject) {
          return p;
        } 
      }
    }
    return null;
  }

View on GitHub (pinned to 1b7edd19c4)