Merge pull request #26 from webstackdev/feature/topic-centroid-feedback-loop

Feature/topic centroid feedback loop

Author: webstackdev

core/admin.py

@@ -6,10 +6,13 @@
 """
 
 import json
+from urllib.parse import urlencode
 
 from django import forms
 from django.contrib import admin, messages
-from django.db.models import Avg
+from django.db.models import Avg, Max, QuerySet
+from django.http import HttpRequest
+from django.urls import reverse
 from django.utils import timezone
 from django.utils.html import format_html
 from django.utils.safestring import mark_safe
@@ -35,6 +38,7 @@
     ReviewQueue,
     SkillResult,
     SourceConfig,
+    TopicCentroidSnapshot,
     UserFeedback,
 )
 from core.plugins import get_plugin_for_source_config, validate_plugin_config
@@ -64,6 +68,94 @@ def _score_color(value) -> str:
     return "red"
 
 
+def _drift_card_color(value) -> str:
+    """Return an admin card severity for centroid drift percentages."""
+
+    if value is None:
+        return "info"
+    numeric_value = float(value)
+    if numeric_value <= 0.15:
+        return "success"
+    if numeric_value <= 0.35:
+        return "warning"
+    return "danger"
+
+
+def _format_snapshot_freshness(computed_at) -> str:
+    """Return a compact human-readable age for the latest snapshot."""
+
+    if computed_at is None:
+        return "-"
+    age = timezone.now() - computed_at
+    total_hours = max(0, int(age.total_seconds() // 3600))
+    if total_hours < 24:
+        return f"{total_hours}h ago"
+    return f"{max(1, total_hours // 24)}d ago"
+
+
+def _freshness_card_color(computed_at) -> str:
+    """Return an admin card severity based on snapshot recency."""
+
+    if computed_at is None:
+        return "warning"
+    age = timezone.now() - computed_at
+    age_hours = age.total_seconds() / 3600
+    if age_hours <= 24:
+        return "success"
+    if age_hours <= 72:
+        return "warning"
+    return "danger"
+
+
+def _build_topic_centroid_project_drilldowns(queryset, changelist_url: str):
+    """Build one filtered-history drilldown row per project.
+
+    The changelist widget needs stable project links even on SQLite, so this keeps
+    the grouping logic in Python instead of relying on database-specific distinct-on
+    behavior.
+    """
+
+    latest_by_project: dict[int, TopicCentroidSnapshot] = {}
+    snapshot_counts: dict[int, int] = {}
+    ordered_snapshots = queryset.select_related("project").order_by(
+        "project_id", "-computed_at"
+    )
+
+    for snapshot in ordered_snapshots:
+        project_id = snapshot.project_id
+        snapshot_counts[project_id] = snapshot_counts.get(project_id, 0) + 1
+        latest_by_project.setdefault(project_id, snapshot)
+
+    project_drilldowns = []
+    for snapshot in sorted(
+        latest_by_project.values(),
+        key=lambda value: value.project.name.lower(),
+    ):
+        project_drilldowns.append(
+            {
+                "project_id": snapshot.project_id,
+                "project_name": snapshot.project.name,
+                "snapshot_count": snapshot_counts[snapshot.project_id],
+                "centroid_active": snapshot.centroid_active,
+                "feedback_count": snapshot.feedback_count,
+                "latest_snapshot": _format_snapshot_freshness(snapshot.computed_at),
+                "drift_from_previous": (
+                    f"{_score_to_percent(snapshot.drift_from_previous):.1f}%"
+                    if snapshot.drift_from_previous is not None
+                    else "n/a"
+                ),
+                "drift_from_week_ago": (
+                    f"{_score_to_percent(snapshot.drift_from_week_ago):.1f}%"
+                    if snapshot.drift_from_week_ago is not None
+                    else "n/a"
+                ),
+                "href": f"{changelist_url}?{urlencode({'project__id__exact': snapshot.project_id})}",
+            }
+        )
+
+    return project_drilldowns
+
+
 class BlueskyCredentialsAdminForm(forms.ModelForm):
     """Admin form that accepts a plaintext Bluesky app credential input."""
 
@@ -223,6 +315,15 @@ class ProjectConfigAdmin(admin.ModelAdmin):
         "upvote_authority_weight",
         "downvote_authority_weight",
         "authority_decay_rate",
+        "recompute_topic_centroid_on_feedback_save",
+    )
+    list_filter = ("recompute_topic_centroid_on_feedback_save",)
+    fields = (
+        "project",
+        "upvote_authority_weight",
+        "downvote_authority_weight",
+        "authority_decay_rate",
+        "recompute_topic_centroid_on_feedback_save",
     )
 
 
@@ -325,6 +426,112 @@ def display_components(self, obj):
         )
 
 
+@admin.register(TopicCentroidSnapshot)
+class TopicCentroidSnapshotAdmin(admin.ModelAdmin):
+    """Admin view for persisted topic-centroid history and drift."""
+
+    list_before_template = "admin/topic_centroid_snapshot_changelist_widget.html"
+    list_display = (
+        "project",
+        "centroid_active",
+        "feedback_count",
+        "display_drift_from_previous",
+        "display_drift_from_week_ago",
+        "computed_at",
+    )
+    list_filter = (
+        "centroid_active",
+        ("project", admin.RelatedOnlyFieldListFilter),
+        "computed_at",
+    )
+    search_fields = ("project__name",)
+    autocomplete_fields = ("project",)
+
+    @admin.display(description="Drift vs Previous", ordering="drift_from_previous")
+    def display_drift_from_previous(self, obj):
+        """Render cosine-distance drift from the previous active snapshot."""
+
+        if obj.drift_from_previous is None:
+            return "n/a"
+        return f"{_score_to_percent(obj.drift_from_previous):.1f}%"
+
+    @admin.display(description="Drift vs 7d", ordering="drift_from_week_ago")
+    def display_drift_from_week_ago(self, obj):
+        """Render cosine-distance drift from the nearest week-old snapshot."""
+
+        if obj.drift_from_week_ago is None:
+            return "n/a"
+        return f"{_score_to_percent(obj.drift_from_week_ago):.1f}%"
+
+    def changelist_view(self, request, extra_context=None):
+        """Augment the changelist with centroid freshness and drift summary cards."""
+
+        queryset = self.get_queryset(request)
+        changelist_url = reverse(
+            f"{self.admin_site.name}:{self.model._meta.app_label}_{self.model._meta.model_name}_changelist"
+        )
+        metrics = queryset.aggregate(
+            avg_drift_from_previous=Avg("drift_from_previous"),
+            avg_drift_from_week_ago=Avg("drift_from_week_ago"),
+            latest_snapshot_at=Max("computed_at"),
+        )
+        project_count = queryset.values("project_id").distinct().count()
+        active_project_count = (
+            queryset.filter(centroid_active=True)
+            .values("project_id")
+            .distinct()
+            .count()
+        )
+
+        extra_context = extra_context or {}
+        extra_context["dashboard_stats"] = [
+            {
+                "title": "Active Centroids",
+                "value": (
+                    f"{active_project_count} / {project_count}"
+                    if project_count
+                    else "0 / 0"
+                ),
+                "icon": "hub",
+                "color": (
+                    "success"
+                    if active_project_count == project_count and project_count
+                    else "warning"
+                ),
+            },
+            {
+                "title": "Avg Drift vs Previous",
+                "value": (
+                    f"{_score_to_percent(metrics['avg_drift_from_previous']):.1f}%"
+                    if metrics["avg_drift_from_previous"] is not None
+                    else "-"
+                ),
+                "icon": "show_chart",
+                "color": _drift_card_color(metrics["avg_drift_from_previous"]),
+            },
+            {
+                "title": "Avg Drift vs 7d",
+                "value": (
+                    f"{_score_to_percent(metrics['avg_drift_from_week_ago']):.1f}%"
+                    if metrics["avg_drift_from_week_ago"] is not None
+                    else "-"
+                ),
+                "icon": "timeline",
+                "color": _drift_card_color(metrics["avg_drift_from_week_ago"]),
+            },
+            {
+                "title": "Latest Snapshot",
+                "value": _format_snapshot_freshness(metrics["latest_snapshot_at"]),
+                "icon": "schedule",
+                "color": _freshness_card_color(metrics["latest_snapshot_at"]),
+            },
+        ]
+        extra_context["centroid_project_drilldowns"] = (
+            _build_topic_centroid_project_drilldowns(queryset, changelist_url)
+        )
+        return super().changelist_view(request, extra_context=extra_context)
+
+
 @admin.register(EntityMention)
 class EntityMentionAdmin(admin.ModelAdmin):
     """Admin view for extracted tracked-entity mentions."""
@@ -404,7 +611,11 @@ def reject_selected_candidates(self, request, queryset):
         )
 
     @admin.action(description="Merge selected candidates into existing entities")
-    def merge_into_existing_entities(self, request, queryset):
+    def merge_into_existing_entities(
+        self,
+        request: HttpRequest,
+        queryset: QuerySet[EntityCandidate],
+    ) -> None:
         """Merge candidates when a same-name entity already exists in the project."""
 
         merged_count = 0

core/api.py

@@ -8,7 +8,7 @@
 import logging
 from typing import Any
 
-from django.db.models import Count, Prefetch
+from django.db.models import Avg, Count, Prefetch, Q
 from drf_spectacular.utils import (
     OpenApiExample,
     OpenApiParameter,
@@ -40,6 +40,7 @@
     ReviewQueue,
     SkillResult,
     SourceConfig,
+    TopicCentroidSnapshot,
     UserFeedback,
 )
 from core.serializers import (
@@ -55,6 +56,8 @@
     ReviewQueueSerializer,
     SkillResultSerializer,
     SourceConfigSerializer,
+    TopicCentroidObservabilitySummarySerializer,
+    TopicCentroidSnapshotSerializer,
     UserFeedbackSerializer,
 )
 
@@ -1036,6 +1039,63 @@ class SourceConfigViewSet(ProjectOwnedQuerysetMixin, viewsets.ModelViewSet):
     queryset = SourceConfig.objects.select_related("project")
 
 
+@document_project_owned_viewset(
+    resource_plural="topic centroid snapshots",
+    resource_singular="topic centroid snapshot",
+    create_description="Topic centroid snapshots are pipeline-managed history rows and are exposed read-only for observability.",
+    tag="Observability",
+    action_overrides=build_crud_action_overrides(
+        TopicCentroidSnapshotSerializer,
+        resource_plural="topic centroid snapshots for the selected project",
+        resource_singular="topic centroid snapshot",
+    ),
+)
+class TopicCentroidSnapshotViewSet(
+    ProjectOwnedQuerysetMixin, viewsets.ReadOnlyModelViewSet
+):
+    """Inspect persisted centroid history and aggregate drift for a project."""
+
+    serializer_class = TopicCentroidSnapshotSerializer
+    queryset = TopicCentroidSnapshot.objects.select_related("project")
+
+    @extend_schema(
+        summary="Get topic centroid summary",
+        description=(
+            "Return aggregate centroid observability metrics for the selected project, "
+            "including average drift and the latest persisted snapshot."
+        ),
+        request=None,
+        responses={
+            200: TopicCentroidObservabilitySummarySerializer,
+            403: AUTHENTICATION_REQUIRED_RESPONSE,
+        },
+        tags=["Observability"],
+    )
+    @action(detail=False, methods=["get"], url_path="summary")
+    def summary(self, request, *args, **kwargs):
+        """Return centroid observability summary metrics for the current project."""
+
+        queryset = self.get_queryset()
+        metrics = queryset.aggregate(
+            snapshot_count=Count("id"),
+            active_snapshot_count=Count("id", filter=Q(centroid_active=True)),
+            avg_drift_from_previous=Avg("drift_from_previous"),
+            avg_drift_from_week_ago=Avg("drift_from_week_ago"),
+        )
+        serializer = TopicCentroidObservabilitySummarySerializer(
+            {
+                "project": self.get_project().id,
+                "snapshot_count": metrics["snapshot_count"],
+                "active_snapshot_count": metrics["active_snapshot_count"],
+                "avg_drift_from_previous": metrics["avg_drift_from_previous"],
+                "avg_drift_from_week_ago": metrics["avg_drift_from_week_ago"],
+                "latest_snapshot": queryset.order_by("-computed_at").first(),
+            },
+            context=self.get_serializer_context(),
+        )
+        return Response(serializer.data)
+
+
 @document_project_owned_viewset(
     resource_plural="review queue entries",
     resource_singular="review queue entry",

core/api_urls.py

@@ -11,6 +11,7 @@
     ReviewQueueViewSet,
     SkillResultViewSet,
     SourceConfigViewSet,
+    TopicCentroidSnapshotViewSet,
     UserFeedbackViewSet,
 )
 
@@ -40,6 +41,11 @@
 project_router.register(
     r"source-configs", SourceConfigViewSet, basename="project-source-config"
 )
+project_router.register(
+    r"topic-centroid-snapshots",
+    TopicCentroidSnapshotViewSet,
+    basename="project-topic-centroid-snapshot",
+)
 project_router.register(
     r"review-queue", ReviewQueueViewSet, basename="project-review-queue"
 )