Cleanup

RSS/parameter configuration UI, feedparser migration, effective-params logging

HuggingFace Spaces deploy + Neo4j Aura support

Refactor + fixes

Set FAKE_THRESHOLD=0.75, benchmark results P=0.909 R=0.227 F1=0.364 Acc=0.650 FP=1 FN=34

integrate coverage_factor into TCS formula (0.85+0.15*cov), unify severity labels, upload.py error handling, dead code cleanup, TextHighlight regex fix, InconsistencyList TYPE_LABELS complete, Swagger description updated

threshold sweep + TF-IDF/RF/SVM baselines, event-date KG +64 canonical events (1169->1233 facts), V8 action-before-office check, _check_event_date() in external.py

Fix

DATE_TOLERANCE 400->200, cross-entity overlap check with same_role_category filter, Reference KG +22 manual facts (Nixon/Ford/LBJ/JFK/Truman/Blair/Thatcher/Starmer/Macron/Tusk + legislation) — F1 0.345->0.393, Recall 0.227->0.273, Sep +0.129->+0.151

RSS verification panel, benchmark split (100/150), remove Ollama/REBEL from frontend, sync Qwen3 labels

Some fix

Replace Ollama/llama3 with spacy-llm + Qwen3-1.7B (Pipeline B + Explainer), archive REBEL (F1=0.038), fix TCS n_claims=0 score, add ISOT/RAGuard eval scripts, update start.sh and requirements

RSS feeds updated + Reference KG 517->1161 facts + entity_label fix + Neo4j reconnect

RSS Stream C3b + HITL endpoint + frontend toggles

Refactor

Pipeline C REBEL-large extractor + orchestrator integration

Testes + refactor

Testes + refactor

New translation

Wikidata inverse, Neo4j cache decoupled, auto-title, cross-article button, batch checkboxes, threshold 0.55

Add new articles and script for liar dataset

Entity subject matching threshold + V5 HOLDS_POSITION only + KG tolerance 400d

Persist article title, source and analyzed_at in Article node

Translate all comments, logs and docstrings to English across 22 backend files

Refactor and eliminate non-relevant files

Suite pytest 33 teste C1→C4 + orchestrator (mock Ollama/Neo4j)

LLMExplainer llama3 cu fallback static + integrare orchestrator

Neo4j cache Wikidata 3-level + cross-article in orchestrator

Refactor

Add internal validation rules and expand external Wikidata checks

Enhance spacy facts coverage and refine llm logging

Migrate to llama3 and enhance temporal extraction prompt

Move dependencies.py up in folder

Add neo4j and cross article verification

Few-shot examples in LLM prompt — improved extraction quality

Externalize model config to .env, auto-install spaCy in startup scripts

Multi-model backend — orchestrator model_name, flexible /compare any-2-models, GET /models endpoint

Fix - source_sentence LLM, shared dependencies"

Add test notebooks (sprint3 + SPARQL verification) + main.py

Update orchestrator - Pipeline B support via extractor registry

Add FastAPI app + routers (analyze, compare, health)

Add explainer.py - natural language explanations

Add llm_extractor.py - Pipeline B via Ollama

Search_entity returns QID, rewrite SPARQL query

Adapt to new search_entity return type

Add tcs.py - implement TCSCalculator with primary formula and weighted variant, plus timeline builder

Add orchestrator.py - implement PipelineOrchestrator with lazy-load and batch support

Add external.py - implement ExternalVerifier with Reference KG and Wikidata comparison

Add internal.py - implement InternalVerifier with cycle detection, causal violation and ordering error checks

Add wikidata.py - implement WikidataClient with SPARQL temporal queries