Fixed another bug in LDA doc-topic mapping

markgw · Feb 10, 2020 · cbab9e8 · cbab9e8
1 parent 49af96b
commit cbab9e8
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/src/python/pimlico/modules/gensim/lda_doc_topics/execute.py b/src/python/pimlico/modules/gensim/lda_doc_topics/execute.py
@@ -16,7 +16,7 @@ def process_document(worker, archive_name, doc_name, doc):
     # Use the LDA model to infer a topic vector for the document
     topic_weights = dict(worker.model[bow])
     # The weights are a sparse vector: fill in the relevant values and leave the rest as 0
-    return [topic_weights.get(i, 0.) for i in range(worker.model.num_topics)]
+    return worker.info.document(vector=[topic_weights.get(i, 0.) for i in range(worker.model.num_topics)])
 
 
 ModuleExecutor = multiprocessing_executor_factory(process_document, worker_set_up_fn=worker_set_up)