jill/LLMsForDisinformationAnalysis
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add better scoring, ignoring duplicates, catching under and over confidence. Showing difference between "FINE" and "PERFECT"

Browse Source
This commit is contained in:
William Jeynes
2026-03-13 12:18:52 +00:00
parent 8311556855
commit d5c6cb444d
1 changed files with 31 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files
supporting/scorer/views/stats.py
+31 -7
View File
@@ -65,6 +65,9 @@ def render():
st.subheader(f"File: {file_path.name}")
confidence_counter = Counter()
overconfident_docs = []
underconfident_docs = []
dup_counter = 0
# ---- Read file line by line ----
with open(file_path, "r", encoding="utf-8") as f:
@@ -73,18 +76,24 @@ def render():
entry = json.loads(line)
except json.JSONDecodeError:
continue
doc_id = entry.get("documentUrl", "UNKNOWN")
for event in entry.get("events", []):
score = event.get("score", None)
extra_lower = (event.get("extra_info", "") or "").strip().lower()
print(extra_lower)
# print(extra_lower)
if score is not None:
if score > THRESH and extra_lower == "perfect":
confidence_counter["Correct-TRUE"] += 1
elif score > THRESH and extra_lower != "perfect":
if "duplicate" in extra_lower:
dup_counter += 1
elif score > THRESH and extra_lower == "perfect":
confidence_counter["Correct-PERFECT"] += 1
elif score > THRESH and extra_lower == "":
confidence_counter["Correct-FINE"] += 1
elif score > THRESH and extra_lower != "perfect" and extra_lower != "":
confidence_counter["Over-confident"] += 1
elif score < THRESH and extra_lower == "perfect":
overconfident_docs.append(doc_id)
elif score < THRESH and (extra_lower == "perfect" or extra_lower == ""):
confidence_counter["Under-confident"] += 1
underconfident_docs.append(doc_id)
else:
confidence_counter["Correct-FALSE"] += 1
@@ -105,10 +114,25 @@ def render():
ax.set_title(file_path.name)
total = sum(confidence_counter.values())
correct = confidence_counter["Correct-TRUE"] + confidence_counter["Correct-FALSE"]
correct = confidence_counter["Correct-PERFECT"] + confidence_counter["Correct-FINE"] + confidence_counter["Correct-FALSE"]
corr_percent = (correct / total) * 100
st.markdown(f"**Correct: {corr_percent:.2f}% ({correct}/{total})**")
st.markdown(f"Duplicates: {dup_counter}")
st.pyplot(fig, width=500)
st.subheader("Over-Confident Document IDs")
if overconfident_docs:
st.container(height=200).write(sorted(set(overconfident_docs)))
else:
st.info("None")
st.subheader("Under-Confident Document IDs")
if underconfident_docs:
st.container(height=200).write(sorted(set(underconfident_docs)))
else:
st.info("None")
else:
st.info("No score data available in this file.")