File size: 4,468 Bytes
0dd5490 | 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 | """Tests for src.metric. Run with `python -m pytest tests` or `python tests/test_metric.py`."""
import math
import os
import sys
import tempfile
from pathlib import Path
sys.path.insert(0, str(Path(__file__).resolve().parents[1]))
from src.metric import (breakdown, entity_scores, f05_entity, macro_f05, # noqa: E402
read_id_list_tsv, write_id_list_tsv)
def test_edge_both_empty():
assert f05_entity(set(), set()) == 1.0
def test_edge_true_empty_pred_nonempty():
assert f05_entity({"S2-1"}, set()) == 0.0
def test_edge_true_nonempty_pred_empty():
assert f05_entity(set(), {"S2-1"}) == 0.0
def test_edge_no_overlap():
assert f05_entity({"S2-1", "S2-2"}, {"S3-9"}) == 0.0
def test_problem_statement_example():
pred = {"S2-00047", "S2-00193", "S3-00812"}
true = {"S2-00047", "S3-00812"}
f = f05_entity(pred, true)
assert round(f, 3) == 0.714
assert math.isclose(f, 5 * 2 / (4 * 3 + 2)) # F = 5 tp / (4|S| + |T|)
def test_known_values():
# one correct out of k true: 1.25 / (0.25k + 1)
for k, want in [(1, 1.0), (2, 1.25 / 1.5), (3, 1.25 / 1.75), (4, 1.25 / 2.0)]:
true = {f"S2-{i}" for i in range(k)}
assert math.isclose(f05_entity({"S2-0"}, true), want)
# one wrong alongside one right, one true: P=.5, R=1 -> 0.5556
assert math.isclose(f05_entity({"S2-0", "S2-x"}, {"S2-0"}), 1.25 * 0.5 / (0.125 + 1))
def test_macro_missing_key_counts_as_empty():
truth = {"a": set(), "b": {"x"}, "c": {"y", "z"}}
pred = {"b": {"x"}, "zzz": {"q"}} # 'a' missing -> empty -> 1; 'c' missing -> 0
assert math.isclose(macro_f05(pred, truth), (1.0 + 1.0 + 0.0) / 3)
def test_read_write_tsv_roundtrip():
with tempfile.TemporaryDirectory() as d:
p = os.path.join(d, "gt.tsv")
with open(p, "w") as fh:
fh.write("source1_entity_id\tmatched_entity_ids\n")
fh.write("S1-1\tS2-1,S3-4\n")
fh.write("S1-2\t\n") # empty list
fh.write("S1-3\n") # missing second column
fh.write('S1-4\t"S2-9, S2-8"\n') # quoted with spaces
fh.write("S1-1\tS2-7\n") # repeated id -> union
got = read_id_list_tsv(p)
assert got == {"S1-1": {"S2-1", "S3-4", "S2-7"}, "S1-2": set(), "S1-3": set(),
"S1-4": {"S2-9", "S2-8"}}
p2 = os.path.join(d, "cand.tsv")
with open(p2, "w") as fh:
fh.write("source1_entity_id\tcandidate_entity_ids\r\nS1-9\tS2-1\r\n")
assert read_id_list_tsv(p2) == {"S1-9": {"S2-1"}}
p3 = os.path.join(d, "out.tsv")
write_id_list_tsv(got, p3, all_ids=["S1-1", "S1-2", "S1-3", "S1-4", "S1-5"])
back = read_id_list_tsv(p3)
assert back == {**got, "S1-5": set()}
assert macro_f05(back, got) == 1.0
def test_breakdown():
truth = {"a": set(), "b": {"x"}, "c": {"y", "z"}, "d": {"1", "2", "3", "4", "5", "6"}}
pred = {"a": {"q"}, "b": {"x"}, "c": {"y"}, "d": {"1"}}
groups = {"a": "US", "b": "US", "c": "IN"} # 'd' -> <missing>
bd = breakdown(pred, truth, groups)
overall = bd.loc[("overall", "all")]
assert math.isclose(overall["macro_f05"], macro_f05(pred, truth))
assert overall["n"] == 4
assert overall["singleton_acc"] == 0.0
# precision over non-empty preds: a 0, b 1, c 1, d 1 -> 0.75
assert math.isclose(overall["mean_precision"], 0.75)
# recall over non-empty truths: b 1, c 1/2, d 1/6
assert math.isclose(overall["mean_recall"], (1 + 0.5 + 1 / 6) / 3)
s = entity_scores(pred, truth)
assert math.isclose(s["f05"].mean(), overall["macro_f05"])
assert math.isclose(bd.loc[("group", "US"), "macro_f05"], 0.5)
assert bd.loc[("group", "<missing>"), "n"] == 1
assert bd.loc[("true_count", "5+"), "n"] == 1
assert bd.loc[("true_count", "0"), "singleton_acc"] == 0.0
assert math.isclose(bd.xs("true_count")["loss_share"].sum(), 1.0)
def _run_all():
mod = sys.modules[__name__]
names = [n for n in dir(mod) if n.startswith("test_") and callable(getattr(mod, n))]
failed = 0
for n in names:
try:
getattr(mod, n)()
print(f"PASS {n}")
except Exception as e: # noqa: BLE001
failed += 1
print(f"FAIL {n}: {type(e).__name__}: {e}")
print(f"{len(names) - failed}/{len(names)} passed")
return failed
if __name__ == "__main__":
raise SystemExit(1 if _run_all() else 0)
|