fix: topic analysis false-matches keywords as substrings (e.g. 'ai' in 'email') (#687)

* fix: match topic keywords on word boundaries, not substrings

* fix: apply word-boundary matching to topic example snippets too

* test: topic keywords match whole words, not substrings
This commit is contained in:
Afonso Coutinho
2026-06-02 03:42:04 +01:00
committed by GitHub
parent 9d8eebfa63
commit 48d3b7abab
2 changed files with 32 additions and 2 deletions

View File

@@ -60,11 +60,11 @@ def analyze_topics(session_manager, owner: str = None) -> Dict[str, Any]:
for topic, keywords in TOPIC_KEYWORDS.items():
for kw in keywords:
if kw in content:
if re.search(rf"\b{re.escape(kw)}\b", content):
topic_counts[topic] += 1
sentences = re.split(r'[.!?]', str(content_raw))
for sentence in sentences:
if kw in sentence.lower():
if re.search(rf"\b{re.escape(kw)}\b", sentence.lower()):
topic_matches[topic].append({
"session_id": session_id,
"session_name": session_name,

View File

@@ -0,0 +1,30 @@
"""Tests for topic keyword matching (src/topic_analyzer.py)."""
from types import SimpleNamespace
from src.topic_analyzer import analyze_topics
def _sm(*messages):
history = [{"role": "user", "content": c} for c in messages]
return SimpleNamespace(sessions={"s1": {"owner": None, "name": "S", "history": history}})
def _freq(result):
return {t["topic"]: t["frequency"] for t in result["topics"]}
def test_substring_does_not_false_match_technology():
# Regression: "ai" matched inside "email"/"again"/"rain"/"wait", flagging
# Technology for messages with no technical content at all.
result = analyze_topics(_sm("Can you send me an email again about the rain? I will wait."))
assert "Technology" not in _freq(result)
def test_real_keywords_still_match():
result = analyze_topics(_sm("I wrote some Python code to test the algorithm."))
assert _freq(result).get("Technology", 0) >= 1
def test_multiword_keyword_matches():
result = analyze_topics(_sm("Can you explain how to set this up?"))
assert "Learning" in _freq(result)