File size: 7,539 Bytes
66f9450 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 | #!/usr/bin/env python3
"""Clasificadores base para comparar con PI-HGAT-T.
Incluye:
- reglas deterministas (oracle de la especificación)
- regresión logística sobre features one-hot de fase+posición
- MLP sin estructura de grafo
"""
from __future__ import annotations
import csv
import json
import random
from pathlib import Path
import numpy as np
import torch
from torch import nn
from torch.nn import functional as F
ROOT = Path(__file__).resolve().parents[2]
CLASSES = ("none", "mirrorOf", "creates")
CLASS_ID = {n: i for i, n in enumerate(CLASSES)}
def seed_everything(seed: int = 20261008) -> None:
random.seed(seed)
np.random.seed(seed)
torch.manual_seed(seed)
torch.use_deterministic_algorithms(True)
torch.set_num_threads(1)
def features_for_node(node_id: str) -> list[float]:
parts = node_id.split("_")
phase, position = parts[1], int(parts[2][1:])
return [float(phase == "adv"), float(phase == "ret"),
float(position == 0), float(position == 1), float(position == 2)]
def parse_pair_index(node_id: str) -> int:
phase = node_id.split("_")[1]
position = int(node_id.split("_")[2][1:])
return position if phase == "adv" else 3 + position
def read_pairs():
pairs: dict[int, list[tuple[str, str, int]]] = {i: [] for i in range(1, 14)}
with (ROOT / "datasets" / "relation_candidates.csv").open(encoding="utf-8", newline="") as fh:
for row in csv.DictReader(fh):
pairs[int(row["triad_index"])].append(
(row["source"], row["target"], CLASS_ID[row["relation"]])
)
return pairs
def admissible(label: str, source: int, target: int) -> bool:
source_phase, source_pos = (0 if source < 3 else 1), source % 3
target_phase, target_pos = (0 if target < 3 else 1), target % 3
if source_phase == target_phase:
return False
if label == "mirrorOf":
return target_pos == 2 - source_pos
if label == "creates":
return (source_pos in (0, 2) and target_pos == 1) or (source_pos == 1 and target_pos in (0, 2))
return True
def deterministic_rule_predict(src_idx: int, tgt_idx: int) -> int:
if admissible("mirrorOf", src_idx, tgt_idx):
return CLASS_ID["mirrorOf"]
if admissible("creates", src_idx, tgt_idx):
return CLASS_ID["creates"]
return CLASS_ID["none"]
def pair_feature(src_id: str, tgt_id: str) -> list[float]:
return features_for_node(src_id) + features_for_node(tgt_id)
def scores(gold: list[int], pred: list[int]) -> dict:
matrix = [[0] * 3 for _ in range(3)]
for a, b in zip(gold, pred):
matrix[a][b] += 1
per_class = {}
f1s = []
for i, name in enumerate(CLASSES):
tp = matrix[i][i]
fp = sum(matrix[r][i] for r in range(3) if r != i)
fn = sum(matrix[i][c] for c in range(3) if c != i)
prec = tp / (tp + fp) if tp + fp else 0.0
rec = tp / (tp + fn) if tp + fn else 0.0
f1 = 2 * prec * rec / (prec + rec) if prec + rec else 0.0
per_class[name] = {"precision": prec, "recall": rec, "f1": f1, "support": tp + fn}
f1s.append(f1)
acc = sum(a == b for a, b in zip(gold, pred)) / max(len(gold), 1)
return {
"accuracy": round(acc, 6),
"macro_f1": round(sum(f1s) / 3, 6),
"per_class": per_class,
"confusion_matrix": matrix,
}
class SoftmaxRegression(nn.Module):
def __init__(self, in_dim: int, n_classes: int = 3):
super().__init__()
self.linear = nn.Linear(in_dim, n_classes)
def forward(self, x):
return self.linear(x)
class SimpleMLP(nn.Module):
def __init__(self, in_dim: int, hidden: int = 32, n_classes: int = 3):
super().__init__()
self.net = nn.Sequential(
nn.Linear(in_dim, hidden),
nn.GELU(),
nn.Linear(hidden, hidden),
nn.GELU(),
nn.Linear(hidden, n_classes),
)
def forward(self, x):
return self.net(x)
def train_sklearn_like(model: nn.Module, X: torch.Tensor, y: torch.Tensor,
epochs: int = 200, lr: float = 0.01) -> None:
opt = torch.optim.AdamW(model.parameters(), lr=lr, weight_decay=1e-4)
for _ in range(epochs):
opt.zero_grad()
logits = model(X)
loss = F.cross_entropy(logits, y)
loss.backward()
opt.step()
def leave_one_triad_out(pairs: dict, model_factory, epochs: int = 200) -> dict:
all_gold, all_pred = [], []
fold_accs = []
for held in sorted(pairs):
seed_everything(20261008 + held - 1)
train_X, train_y = [], []
test_X, test_y, test_meta = [], [], []
for tid, plist in pairs.items():
for src, tgt, label in plist:
feat = pair_feature(src, tgt)
if tid == held:
test_X.append(feat)
test_y.append(label)
test_meta.append((src, tgt))
else:
train_X.append(feat)
train_y.append(label)
Xtr = torch.tensor(train_X, dtype=torch.float32)
ytr = torch.tensor(train_y, dtype=torch.long)
Xte = torch.tensor(test_X, dtype=torch.float32)
model = model_factory(Xtr.shape[1])
train_sklearn_like(model, Xtr, ytr, epochs=epochs)
with torch.no_grad():
pred = model(Xte).argmax(dim=-1).tolist()
all_gold.extend(test_y)
all_pred.extend(pred)
fold_accs.append(sum(a == b for a, b in zip(test_y, pred)) / max(len(test_y), 1))
result = scores(all_gold, all_pred)
result["fold_accuracies"] = fold_accs
result["fold_accuracy_mean"] = round(float(np.mean(fold_accs)), 6)
result["fold_accuracy_std"] = round(float(np.std(fold_accs)), 6)
return result
def run_deterministic(pairs: dict) -> dict:
gold, pred = [], []
for plist in pairs.values():
for src, tgt, label in plist:
gold.append(label)
s_idx = parse_pair_index(src)
t_idx = parse_pair_index(tgt)
pred.append(deterministic_rule_predict(s_idx, t_idx))
return scores(gold, pred)
def run() -> dict:
seed_everything()
pairs = read_pairs()
report = {
"description": "Comparación de PI-HGAT-T con clasificadores base sobre el mismo leave-one-triad-out",
"baselines": {},
}
report["baselines"]["deterministic_rules"] = {
"description": "Oracle que aplica exactamente las reglas SHACL de mirrorOf/creates",
"metrics": run_deterministic(pairs),
}
report["baselines"]["logistic_regression"] = {
"description": "Softmax linear sobre concatenación de one-hot fase+posición (10 dim)",
"metrics": leave_one_triad_out(pairs, lambda d: SoftmaxRegression(d), epochs=300),
}
report["baselines"]["mlp_no_graph"] = {
"description": "MLP de dos capas ocultas sin mensaje-passing ni atención",
"metrics": leave_one_triad_out(pairs, lambda d: SimpleMLP(d, hidden=32), epochs=200),
}
out = ROOT / "experiments" / "results" / "baselines_comparison.json"
out.parent.mkdir(parents=True, exist_ok=True)
out.write_text(json.dumps(report, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")
print("Baselines escritos en", out)
for name, block in report["baselines"].items():
m = block["metrics"]
print(f" {name}: accuracy={m['accuracy']:.4f} macro_f1={m['macro_f1']:.4f}")
return report
if __name__ == "__main__":
run()
|