Diagnostics / Semantic pipeline report
Close
Read-only report from existing DB results. It does not run ingestion, matching, clustering, or embeddings.
Pipeline status snapshot
last_run_at=not available
status=not available
topic=AI & Tech / 1cebc882-7c2f-4da8-9dba-ca5a7011d055
ingestion_count=not available
feature_extraction_count=not available
matching=not available
clustering=not available
summaries_generated=not available
errors_warnings=not available
Semantic pipeline config snapshot
FEWS_ENABLE_SEMANTIC_PIPELINE=True
FEWS_EMBEDDING_BACKEND=sentence_transformers
FEWS_EMBEDDING_MODEL=paraphrase-multilingual-MiniLM-L12-v2
FEWS_MATCHING_SCORE_MODE=semantic_primary
FEWS_CLUSTERING_ALGORITHM=agglomerative_semantic
FEWS_SEMANTIC_MATCHING_THRESHOLD=0.5
FEWS_SEMANTIC_CLUSTERING_THRESHOLD=0.62
min_pair_similarity=0.59
bucket_hours=12
max_posts_for_clustering=500
llm_summary_enabled=True
Latest runtime result
semantic_pipeline_enabled_used=not available
embedding_backend_used=not available
matching_score_mode_used=not available
clustering_algorithm_used=not available
matching_runtime_seconds=not available
clustering_runtime_seconds=not available
events_created=not available
summaries_generated=not available
summaries_provider_used=not available
summaries_provider_configured=not available
summaries_provider_effective=not available
llm_summary_enabled=not available
llm_summary_calls=not available
llm_summary_skipped_cache=not available
llm_summary_skipped_run_limit=not available
llm_summary_fallback_count=not available
llm_summary_errors=not available
llm_summary_estimated_input_chars=not available
llm_summary_model_used=not available
fingerprint_action_mismatch_rejections=not available
fingerprint_object_mismatch_rejections=not available
fingerprint_location_mismatch_rejections=not available
fingerprint_direction_mismatch_rejections=not available
claim_type_mismatch_rejections=not available
low_information_excluded_count=not available
digest_excluded_count=not available
same_source_batch_splits=not available
analysis_guard_rejections=not available
broad_actor_topic_rejections=not available
weak_keyphrase_overlap_splits=not available
opinion_analysis_isolated_count=not available
ad_promotional_isolated_count=not available
semantic_subtype_incompatible_rejections=not available
rhetoric_propaganda_isolated_count=not available
digest_roundup_isolated_count=not available
announcement_stub_isolated_count=not available
action_mismatch_rejections=not available
object_mismatch_rejections=not available
location_mismatch_rejections=not available
same_source_different_action_splits=not available
reprocess_total_runtime_seconds=not available
reprocess_settings_seconds=not available
reprocess_select_posts_seconds=not available
reprocess_reset_seconds=not available
reprocess_mark_posts_seconds=not available
reprocess_reset_commit_seconds=not available
reprocess_feature_extraction_seconds=not available
reprocess_rebuild_loop_seconds=not available
reprocess_summary_stage_seconds=not available
Topic language
configured_topic_primary_language=auto
resolved_topic_output_language=ru
output_language_fallback=auto_fallback_no_language_data
post_language_distribution={'ru': 1}
multilingual_events_count=0
Event feed quality metrics
total_events=1
events_with_latest_summary=1
events_missing_latest_summary=0
total_mentions_posts=1
total_sources=1
avg_mentions_per_event=1.0
singleton_events_count=1
multi_post_events_count=0
digest_singleton_events_count=0
events_missing_summary_short=0
events_missing_summary_long=0
events_missing_summary_detailed=0
events_with_stale_latest_activity_at=0
events_with_conflict_flags=1
events_with_english_boilerplate=0
events_with_generic_title=0
events_with_single_source_provisional_text=0
events_with_low_information_mentions=0
Summary completeness
summary_short_coverage_pct=100.0
summary_long_coverage_pct=100.0
summary_detailed_coverage_pct=100.0
detailed_with_key_claims_count=0
Conflict flags distribution
events_with_real_conflict_flags=1
top_real_conflict_flag_keys={}
real_conflict_flags_distribution={}
events_with_metadata_flags=1
top_metadata_flag_keys={'is_digest': 1, 'post_semantic_subtype': 1, 'post_semantic_subtypes': 1, 'post_types': 1, 'reference_numbers': 1}
metadata_flags_distribution={'is_digest=false': 1, 'post_semantic_subtype="diplomatic_statement"': 1, 'post_semantic_subtypes=["diplomatic_statement"]': 1, 'post_types=["summary"]': 1, 'reference_numbers=[]': 1}
Semantic clustering diagnostics
Event
Mentions
Sources
Latest
Digest
Summary
Flags
Fingerprint
Post diagnostics
Similarity
4d54f9a2-a800-4dae-b6ef-dab9075c8ee3
1
1
2026-07-08T12:10:56+03:00
no
short=yes,
long=yes,
detailed=yes
{'reference_numbers': [], 'is_digest': False, 'post_types': ['summary'], 'post_semantic_subtypes': ['diplomatic_statement'], 'post_semantic_subtype': 'diplomatic_statement'}
actor=amazon kindle colorsoft
action=statement
object=цветной
low_info=0
representative=1acc90f1-3109-4b80-bfe8-878140ef4764
validation=[]
post=1acc90f1-3109-4b80-bfe8-878140ef4764
source=Protech
type=summary low=False
action=statement
not persisted / not persisted
Top suspicious events
Event
Title
Reasons
Conflict flags
Mentions
Latest
4d54f9a2-a800-4dae-b6ef-dab9075c8ee3
Пользователь сравнил Kindle Colorsoft и Paperwhite
['singleton_event', 'has_conflict_flags', 'summary_detailed_missing_key_claims']
{"is_digest": false, "post_semantic_subtype": "diplomatic_statement", "post_semantic_subtypes": ["diplomatic_statement"], "post_types": ["summary"], "reference_numbers": []}
1
2026-07-08T12:10:56+03:00
Summary quality sample
Event
Title / counts
Summary short
Summary long
Detailed preview
Quality flags
4d54f9a2-a800-4dae-b6ef-dab9075c8ee3
Пользователь сравнил Kindle Colorsoft и Paperwhite
mentions=1 sources=1
Длительное использование показало, что цветной ридер не всегда лучше, так как у Colorsoft есть недостатки, и многие не должны платить больше за цвет.
Длительное использование показало, что цветной ридер не всегда лучше, так как у Colorsoft есть недостатки, и многие не должны платить больше за цвет.
Цветной — не всегда лучший! Долго пользовался ридером Amazon Kindle Colorsoft и сравнил его со своим чёрно‑белым Paperwhite. Могу сказать, что далеко не всё так однозначно. Многим действительно нет смысла переплачивать за цвет, так как есть ряд неприятных недостатков, которые вы получите в довесок. Подробнее рассказал на YouTube и ВК!
{'contains_english_boilerplate': False, 'contains_single_source_provisional_text': False, 'summary_too_short': False, 'summary_detailed_missing_key_claims': True, 'title_too_generic': False}
STEP 7 avg_pair_similarity/min_pair_similarity are not persisted; Get results does not recompute them.
Per-stage pipeline counts are shown only when present in pipeline_jobs.result_json.
Store future run-stage diagnostics in pipeline_jobs.result_json or a dedicated run diagnostics table.
Per-post similarity_to_representative is not persisted; Get results does not recompute embeddings.
Copy-friendly diagnostics block
Copy to clipboard