News Analytics
FastAPI-backed source, lens, workflow, and event diagnostics
The news surface reads from the backend contract and exposes exploratory comparisons, data quality checks, workflow health, and same-event analysis.
Index
Pages
All routes below stay inside the same shared analytics surface.
News Group Latent Space
Live FastAPI data
What this page does
Map sources, topics, and tags into the shared lens PCA/MDS space to compare centroid position, dispersion, and nearest neighbors.
Latent Groups
Group Latent-Space Status
Sources, topics, and tags are projected into the existing global lens PCA/MDS space so group positions stay comparable.
ok
Controls
Group View
Switch group type, then pick one group to highlight across the maps and detail tables.
Centroid Maps
Group Centroid Maps
Bubble size follows article count. The selected group is highlighted; low-sample groups remain visible but should be interpreted cautiously.
Topics PCA Clusters
Clusters group nearby centroids in shared PC1-PC3 space for the active group type.
| Cluster | Representative Groups | Groups | Articles | Defining Lens Deviations |
|---|---|---|---|---|
| Untagged, world | Untagged, world | 2 | 172 | Emotional Intensity Lens (12.4), Causal Attribution Lens (7.7), Credibility Lens (7.2), Epistemic Modality and Certainty Lens (5.3) |
Tag Profiles
Tag Lens PCA
A separate PCA where each tag is represented by its mean lens-score profile. Use this to see which tags have similar framing signatures.
ok
| Tag | Articles | Sources | Cluster | PC1 | PC2 |
|---|---|---|---|---|---|
| politics | 30 | 5 | security, Ukraine, Russia | -1.445 | -1.090 |
| UK | 21 | 2 | security, Ukraine, Russia | -1.612 | -0.727 |
| health | 20 | 3 | health | -0.681 | 1.254 |
| Ebola | 17 | 2 | security, Ukraine, Russia | 2.279 | 1.566 |
| military | 12 | 5 | security, Ukraine, Russia | -0.451 | -1.145 |
| sports | 11 | 5 | sports | -6.023 | 0.443 |
| World Cup | 9 | 4 | World Cup | -5.106 | 0.998 |
| conflict | 8 | 3 | conflict | 2.609 | -1.006 |
| DRC | 8 | 2 | security, Ukraine, Russia | 2.011 | 3.089 |
| Iran | 8 | 4 | Iran, US | 1.373 | -2.148 |
| security | 8 | 5 | security, Ukraine, Russia | 0.395 | -0.455 |
| Russia | 6 | 2 | security, Ukraine, Russia | 1.858 | 0.142 |
| Kenya | 5 | 2 | security, Ukraine, Russia | 2.215 | 1.079 |
| Ukraine | 5 | 2 | security, Ukraine, Russia | 1.787 | 0.255 |
| US | 5 | 4 | Iran, US | 0.789 | -2.256 |
Component-Defining Tags
Positive and negative extremes show which tags define each PCA axis. These are tag-profile positions, not popularity rankings.
| Component | Positive Tags | Negative Tags |
|---|---|---|
| PC1 | conflict (2.61), Ebola (2.28), Kenya (2.22), DRC (2.01) | sports (-6.02), World Cup (-5.11), UK (-1.61), politics (-1.44) |
| PC2 | DRC (3.09), Ebola (1.57), health (1.25), Kenya (1.08) | US (-2.26), Iran (-2.15), military (-1.15), politics (-1.09) |
| PC3 | UK (1.68), Russia (1.28), politics (1.12), military (1.12) | sports (-1.51), Kenya (-1.38), US (-1.19), Iran (-1.07) |
Tag PCA Clusters
Clusters group nearby tag profiles in PC1-PC3 space. Use them as exploratory neighborhoods, not ground-truth topic labels.
| Cluster | Tags | Articles | Sources | Defining Lenses |
|---|---|---|---|---|
| security, Ukraine, Russia | security, Ukraine, Russia, military, Ebola | 112 | 7 | Objectivity vs Opinion Lens (98.0), Linguistic Quality Lens (94.5), Sentiment Clarity Lens (93.6), Entity-Level Sentiment Lens (92.3) |
| Iran, US | Iran, US | 13 | 4 | Objectivity vs Opinion Lens (100.0), Entity-Level Sentiment Lens (90.8), Linguistic Quality Lens (90.6), Sentiment Clarity Lens (89.4) |
| health | health | 20 | 3 | Sentiment Clarity Lens (97.5), Objectivity vs Opinion Lens (95.0), Entity-Level Sentiment Lens (93.5), Linguistic Quality Lens (93.0) |
| sports | sports | 11 | 5 | Linguistic Quality Lens (96.4), Sentiment Clarity Lens (96.4), Entity-Level Sentiment Lens (95.5), Objectivity vs Opinion Lens (91.4) |
| World Cup | World Cup | 9 | 4 | Linguistic Quality Lens (98.9), Sentiment Clarity Lens (96.7), Entity-Level Sentiment Lens (95.6), Objectivity vs Opinion Lens (87.2) |
| conflict | conflict | 8 | 3 | Objectivity vs Opinion Lens (100.0), Entity-Level Sentiment Lens (91.9), Authority and Source Positioning Lens (90.0), Linguistic Quality Lens (90.0) |
Selected Group
Untagged
Centroid position, dispersion, sample size, and top lens deviations for the active group.
ok
| Neighbor | PCA Distance |
|---|---|
| world | 1.619 |
Composition
Selected Group Composition
The selected group's source, topic, and tag mix. These counts help separate coverage composition from latent-space position.
Sources
| BBC News | 61 |
|---|---|
| NPR | 51 |
| Fox News | 11 |
| PBS NewsHour | 5 |
| CBS News | 2 |
| ABC News | 1 |
| NBC News | 1 |
Topics
| Untagged | 132 |
|---|
Tags
| politics | 28 |
|---|---|
| UK | 19 |
| health | 11 |
| military | 11 |
| sports | 10 |
| conflict | 8 |
| Iran | 8 |
| World Cup | 8 |
Reference Table
Group Centroids
Ranked group centroid table for the active group type.
| Group | Status | Articles | Sources | PC1 | PC2 | MDS1 | MDS2 | Cluster | PCA Dispersion |
|---|---|---|---|---|---|---|---|---|---|
| Untagged | ok | 132 | 7 | -0.370 | -0.057 | 0.370 | 0.057 | Untagged, world | 2.163 |
| world | ok | 40 | 1 | 1.220 | 0.190 | -1.220 | -0.190 | Untagged, world | 1.803 |