changes

Set default model to gpt-5-mini-2025-08-07 across all components

Set default model to gpt-5-mini-2025-08-07 across all components

Fix state propaganda bias direction: score as RIGHT per MBFC methodology

Fix state propaganda bias direction: score as RIGHT per MBFC methodology

Fix ModuleNotFoundError: change 'from ddgs import DDGS' to 'from duckduckgo_search import DDGS'

Fix ModuleNotFoundError: change 'from ddgs import DDGS' to 'from duckduckgo_search import DDGS'

Add debug trace JSON with pipeline step timing and log capture

Add debug trace JSON with pipeline step timing and log capture

Respect --model CLI arg across all entry points

Respect --model CLI arg across all entry points

Save full intermediate metric data to JSONL and add offline recompuation

Save full intermediate metric data to JSONL and add offline recomputation

Add per-outlet results logging and benchmark analysis report

Add per-outlet results logging and benchmark analysis report

new results

Improve system mode bias/factuality MAE and FACTScore precision

Improve system mode bias/factuality MAE and FACTScore precision

Fix more AttributeErrors in _calibrate_scores_with_llm

Fix more AttributeErrors in _calibrate_scores_with_llm

Fix AttributeError: EditorialBiasResult has no 'bias_summary' attribute

Fix AttributeError: EditorialBiasResult has no 'bias_summary' attribute

Add LLM score calibration and --cached-only filter for system mode

Add LLM score calibration and --cached-only filter for system mode

upds

Fix MAE computation to use ordinal class mapping

Fix MAE computation to use ordinal class mapping

Add rescore script to evaluate benchmark on cached-only outlets

Add rescore script to evaluate benchmark on cached-only outlets

test

Add unit tests for WHOIS, ownership research, and funding detection

Add unit tests for WHOIS, ownership research, and funding detection

Merge pull request #57 from mbaisbay/claude/media-profiling-mbfc-Hb8Kh

Add factuality descriptions, positive fact-check assertion, fix transparency default

Fix system mode centrality bias: improve MAE scores and FACTScore precision

Improve FACTScore precision: negative assertion filtering, dense retrieval, LLM synthesis mode

Improve FACTScore precision: negative assertion filtering, dense retrieval, LLM synthesis mode

changes to the hybrid, fixing FC detect and etc

Revert LLM score adjustment to maintain strict MBFC formula compliance

Improve system/hybrid to top benchmark positions

Improve benchmark scores: fix FC detection, add calibration anchors

Post-hoc replace bias_category_description with non-MBFC text

Post-hoc replace bias_category_description with non-MBFC text

Fix system mode boilerplate leakage: rewrite bias category descriptions

Fix system mode boilerplate leakage: rewrite bias category descriptions

Expand audit to all 6 modes, skip missing JSONL files gracefully

Expand audit to all 6 modes, skip missing JSONL files gracefully

Also sanitize log files: remove stale [Search] LEAKAGE warning lines

Also sanitize log files: remove stale [Search] LEAKAGE warning lines

Add post-hoc sanitization script for existing JSONL results