`_extract_representative_docs` uses text-based `in` matching, causing wrong documents when duplicates exist

[pidefrem]

Describe the bug

_extract_representative_docs() maps selected documents back to their original indices using text membership testing:

doc_ids = [selected_docs_ids[index] for index, doc in enumerate(selected_docs) if doc in docs]

When the same document text appears in multiple topics (common with short texts, boilerplate, or near-duplicates), doc in docs matches the first occurrence regardless of which topic the document actually belongs to. This causes:

• Documents skipped entirely (their index never matched for the right topic)
• Misaligned doc_ids ↔ selected_docs pairs — wrong similarity scores assigned to wrong documents
• representative_docs_ containing documents that don't belong to their assigned topic

This is a separate bug from the replace=True sampling issue (#2491). That one creates duplicates going into the function; this bug is in the output mapping logic.

Reproduction

from bertopic import BERTopic

# Create a dataset where some documents appear verbatim in multiple topics
docs = [
    "machine learning is great",  # shared text
    "deep learning neural networks",
    "machine learning is great",  # same text, different context
    "topic modeling with BERTopic",
    # ... enough docs to form 2+ topics with shared text
]
topic_model = BERTopic(min_topic_size=2)
topics, _ = topic_model.fit_transform(docs)
# → representative_docs_ may contain docs mapped to the wrong topic

BERTopic Version

0.17.4

Your contribution

I've already worked through a fix in my fork (replacing the text-based membership lookup with positional indexing), with tests. Happy to open a PR if you're on board with the approach.