anpassungen Kantenvergeleich
This commit is contained in:
parent
857ba953e3
commit
0a429e1f7b
|
|
@ -1,7 +1,10 @@
|
|||
"""
|
||||
FILE: app/core/graph/graph_derive_edges.py
|
||||
DESCRIPTION: Hauptlogik zur Kanten-Aggregation und De-Duplizierung.
|
||||
AUDIT: Integriert parse_link_target für saubere Graphen-Topologie.
|
||||
AUDIT:
|
||||
- Nutzt parse_link_target
|
||||
- Übergibt Section als 'variant' an ID-Gen
|
||||
- Dedup basiert jetzt auf Edge-ID (erlaubt Multigraph für Sections)
|
||||
"""
|
||||
from typing import List, Optional, Dict, Tuple
|
||||
from .graph_utils import (
|
||||
|
|
@ -18,11 +21,11 @@ def build_edges_for_note(
|
|||
note_level_references: Optional[List[str]] = None,
|
||||
include_note_scope_refs: bool = False,
|
||||
) -> List[dict]:
|
||||
"""Erzeugt und aggregiert alle Kanten für eine Note (WP-15b)."""
|
||||
"""Erzeugt und aggregiert alle Kanten für eine Note."""
|
||||
edges: List[dict] = []
|
||||
note_type = _get(chunks[0], "type") if chunks else "concept"
|
||||
|
||||
# 1) Struktur-Kanten (belongs_to, next/prev)
|
||||
# 1) Struktur-Kanten
|
||||
for idx, ch in enumerate(chunks):
|
||||
cid = _get(ch, "chunk_id", "id")
|
||||
if not cid: continue
|
||||
|
|
@ -52,76 +55,78 @@ def build_edges_for_note(
|
|||
if not cid: continue
|
||||
raw = _get(ch, "window") or _get(ch, "text") or ""
|
||||
|
||||
# Typed & Candidate Pool (WP-15b Integration)
|
||||
# Typed
|
||||
typed, rem = extract_typed_relations(raw)
|
||||
for k, raw_t in typed:
|
||||
t, sec = parse_link_target(raw_t, note_id)
|
||||
if not t: continue
|
||||
|
||||
payload = {
|
||||
"chunk_id": cid, "edge_id": _mk_edge_id(k, cid, t, "chunk", "inline:rel"),
|
||||
"chunk_id": cid,
|
||||
# Variant=sec sorgt für eindeutige ID pro Abschnitt
|
||||
"edge_id": _mk_edge_id(k, cid, t, "chunk", "inline:rel", variant=sec),
|
||||
"provenance": "explicit", "rule_id": "inline:rel", "confidence": PROVENANCE_PRIORITY["inline:rel"]
|
||||
}
|
||||
if sec: payload["target_section"] = sec
|
||||
|
||||
edges.append(_edge(k, "chunk", cid, t, note_id, payload))
|
||||
|
||||
# Semantic AI Candidates
|
||||
pool = ch.get("candidate_pool") or ch.get("candidate_edges") or []
|
||||
for cand in pool:
|
||||
raw_t, k, p = cand.get("to"), cand.get("kind", "related_to"), cand.get("provenance", "semantic_ai")
|
||||
t, sec = parse_link_target(raw_t, note_id)
|
||||
if t:
|
||||
payload = {
|
||||
"chunk_id": cid, "edge_id": _mk_edge_id(k, cid, t, "chunk", f"candidate:{p}"),
|
||||
"chunk_id": cid,
|
||||
"edge_id": _mk_edge_id(k, cid, t, "chunk", f"candidate:{p}", variant=sec),
|
||||
"provenance": p, "rule_id": f"candidate:{p}", "confidence": PROVENANCE_PRIORITY.get(p, 0.90)
|
||||
}
|
||||
if sec: payload["target_section"] = sec
|
||||
|
||||
edges.append(_edge(k, "chunk", cid, t, note_id, payload))
|
||||
|
||||
# Callouts & Wikilinks
|
||||
# Callouts
|
||||
call_pairs, rem2 = extract_callout_relations(rem)
|
||||
for k, raw_t in call_pairs:
|
||||
t, sec = parse_link_target(raw_t, note_id)
|
||||
if not t: continue
|
||||
|
||||
payload = {
|
||||
"chunk_id": cid, "edge_id": _mk_edge_id(k, cid, t, "chunk", "callout:edge"),
|
||||
"chunk_id": cid,
|
||||
"edge_id": _mk_edge_id(k, cid, t, "chunk", "callout:edge", variant=sec),
|
||||
"provenance": "explicit", "rule_id": "callout:edge", "confidence": PROVENANCE_PRIORITY["callout:edge"]
|
||||
}
|
||||
if sec: payload["target_section"] = sec
|
||||
|
||||
edges.append(_edge(k, "chunk", cid, t, note_id, payload))
|
||||
|
||||
# Wikilinks & Defaults
|
||||
refs = extract_wikilinks(rem2)
|
||||
for raw_r in refs:
|
||||
r, sec = parse_link_target(raw_r, note_id)
|
||||
if not r: continue
|
||||
|
||||
# Explicit Reference
|
||||
payload = {
|
||||
"chunk_id": cid, "ref_text": raw_r, "edge_id": _mk_edge_id("references", cid, r, "chunk", "explicit:wikilink"),
|
||||
"chunk_id": cid, "ref_text": raw_r,
|
||||
"edge_id": _mk_edge_id("references", cid, r, "chunk", "explicit:wikilink", variant=sec),
|
||||
"provenance": "explicit", "rule_id": "explicit:wikilink", "confidence": PROVENANCE_PRIORITY["explicit:wikilink"]
|
||||
}
|
||||
if sec: payload["target_section"] = sec
|
||||
|
||||
edges.append(_edge("references", "chunk", cid, r, note_id, payload))
|
||||
|
||||
# Defaults (nur einmal pro Target, Section hier irrelevant für Typ-Logik, oder?)
|
||||
# Wir erzeugen Defaults auch pro Section, um Konsistenz zu wahren.
|
||||
for rel in defaults:
|
||||
if rel != "references":
|
||||
def_payload = {
|
||||
"chunk_id": cid, "edge_id": _mk_edge_id(rel, cid, r, "chunk", f"edge_defaults:{rel}"),
|
||||
"chunk_id": cid,
|
||||
"edge_id": _mk_edge_id(rel, cid, r, "chunk", f"edge_defaults:{rel}", variant=sec),
|
||||
"provenance": "rule", "rule_id": f"edge_defaults:{rel}", "confidence": PROVENANCE_PRIORITY["edge_defaults"]
|
||||
}
|
||||
if sec: def_payload["target_section"] = sec
|
||||
edges.append(_edge(rel, "chunk", cid, r, note_id, def_payload))
|
||||
|
||||
# Für Note-Scope Sammlung nutzen wir den Original-String zur Dedup, aber gesäubert
|
||||
refs_all.extend([parse_link_target(r, note_id)[0] for r in refs])
|
||||
|
||||
# 3) Note-Scope & De-Duplizierung
|
||||
if include_note_scope_refs:
|
||||
# refs_all ist jetzt schon gesäubert (nur Targets)
|
||||
# note_level_references müssen auch gesäubert werden
|
||||
cleaned_note_refs = [parse_link_target(r, note_id)[0] for r in (note_level_references or [])]
|
||||
refs_note = _dedupe_seq((refs_all or []) + cleaned_note_refs)
|
||||
|
||||
|
|
@ -136,10 +141,13 @@ def build_edges_for_note(
|
|||
"provenance": "rule", "confidence": PROVENANCE_PRIORITY["derived:backlink"]
|
||||
}))
|
||||
|
||||
unique_map: Dict[Tuple[str, str, str], dict] = {}
|
||||
# Deduplizierung: Wir nutzen jetzt die EDGE-ID als Schlüssel.
|
||||
# Da die Edge-ID nun 'variant' (Section) enthält, bleiben unterschiedliche Sections erhalten.
|
||||
unique_map: Dict[str, dict] = {}
|
||||
for e in edges:
|
||||
key = (str(e.get("source_id")), str(e.get("target_id")), str(e.get("kind")))
|
||||
if key not in unique_map or e.get("confidence", 0) > unique_map[key].get("confidence", 0):
|
||||
unique_map[key] = e
|
||||
eid = e["edge_id"]
|
||||
# Bei Konflikt (gleiche ID = exakt gleiche Kante und Section) gewinnt die höhere Confidence
|
||||
if eid not in unique_map or e.get("confidence", 0) > unique_map[eid].get("confidence", 0):
|
||||
unique_map[eid] = e
|
||||
|
||||
return list(unique_map.values())
|
||||
|
|
@ -41,10 +41,19 @@ def _dedupe_seq(seq: Iterable[str]) -> List[str]:
|
|||
seen.add(s); out.append(s)
|
||||
return out
|
||||
|
||||
def _mk_edge_id(kind: str, s: str, t: str, scope: str, rule_id: Optional[str] = None) -> str:
|
||||
"""Erzeugt eine deterministische 12-Byte ID mittels BLAKE2s."""
|
||||
def _mk_edge_id(kind: str, s: str, t: str, scope: str, rule_id: Optional[str] = None, variant: Optional[str] = None) -> str:
|
||||
"""
|
||||
Erzeugt eine deterministische 12-Byte ID mittels BLAKE2s.
|
||||
|
||||
WP-Fix: 'variant' (z.B. Section) fließt in den Hash ein, um mehrere Kanten
|
||||
zum gleichen Target-Node (aber unterschiedlichen Abschnitten) zu unterscheiden.
|
||||
"""
|
||||
base = f"{kind}:{s}->{t}#{scope}"
|
||||
if rule_id: base += f"|{rule_id}"
|
||||
if rule_id:
|
||||
base += f"|{rule_id}"
|
||||
if variant:
|
||||
base += f"|{variant}" # <--- Hier entsteht die Eindeutigkeit für verschiedene Sections
|
||||
|
||||
return hashlib.blake2s(base.encode("utf-8"), digest_size=12).hexdigest()
|
||||
|
||||
def _edge(kind: str, scope: str, source_id: str, target_id: str, note_id: str, extra: Optional[dict] = None) -> dict:
|
||||
|
|
|
|||
Loading…
Reference in New Issue
Block a user