stanfordnlp/CoreNLP · error · UnknownVertexException

Unknown vertex

Error message

Unknown vertex

What it means

SemanticGraph.descendants(vertex) computes the transitive set of nodes governed by the given vertex via depth-first search, and throws UnknownVertexException if the vertex is not present in this graph. Each graph method validates membership first, so passing a node from a different (or rebuilt) graph always throws.

Solutions

1. Check graph.containsVertex(vertex) before calling descendants
2. Ensure the vertex came from the same graph instance (same parse), not another graph or a stale reference
3. Use the Safe getter methods to re-resolve the vertex by index on the current graph
4. Restructure code so vertex references do not outlive the graph they belong to

Example fix

// before
Set<IndexedWord> desc = graph.descendants(v);
// after
if (graph.containsVertex(v)) {
  Set<IndexedWord> desc = graph.descendants(v);
} else { /* re-resolve by index or handle */ }

if (!graph.containsVertex(v)) { throw new IllegalStateException("vertex not in graph"); }

IndexedWord resolve(SemanticGraph g, IndexedWord maybeStale, int index) {
  return g.containsVertex(maybeStale) ? maybeStale : g.getNodeByIndexSafe(index);
}

try {
  Set<IndexedWord> desc = graph.descendants(v);
} catch (UnknownVertexException e) {
  // re-resolve v by index on the current graph
}

When it happens

Trigger: Calling descendants with an IndexedWord obtained from another SemanticGraph instance, or from a vertex whose backing document/graph was re-created, or after the node was removed from this graph.

Common situations: Comparing nodes across basic vs enhanced graphs; keeping IndexedWord references across pipeline stages that rebuild the parse; copy constructors that create a new graph while old vertex references are reused.

Understand the failure class

Background: 'Could not be found', 'does not exist', 'not found in database': the resource-not-found family when an ID, slug, key, or URI lookup comes back empty — this error's family across 20 libraries.

Related errors

• No SemanticGraph vertex with index + index
• No SemanticGraph vertex with index + index + and…

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/ef5ee5ca9f5f7dda. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/semgraph/SemanticGraph.java:701

  public List<IndexedWord> getAllNodesByPartOfSpeechPattern(String pattern) {
    Pattern p = Pattern.compile(pattern);
    List<IndexedWord> nodes = new ArrayList<>();
    for (IndexedWord vertex : vertexSet()) {
      String pos = vertex.tag();
      if ((pos == null && pattern == null) || pos != null && p.matcher(pos).matches()) {
        nodes.add(vertex);
      }
    }
    return nodes;
  }

  /**
   * Returns the set of descendants governed by this node in the graph.
   *
   */
  public Set<IndexedWord> descendants(IndexedWord vertex) {
    if (!containsVertex(vertex)) {
      throw new UnknownVertexException(vertex, this);
    }
    // Do a depth first search
    Set<IndexedWord> descendantSet = wordMapFactory.newSet();
    descendantsHelper(vertex, descendantSet);
    return descendantSet;
  }

  private void descendantsHelper(IndexedWord curr, Set<IndexedWord> descendantSet) {
    if (descendantSet.contains(curr)) {
      return;
    }
    descendantSet.add(curr);
    for (IndexedWord child : getChildren(curr)) {
      descendantsHelper(child, descendantSet);
    }
  }

  /**

View on GitHub (pinned to 1b7edd19c4)