Update llm assisted annotation

This commit is contained in:
Xin Wang
2026-09-14 23:04:21 +08:00
parent 068b7ab9e6
commit 06b5e6c385
12 changed files with 1348 additions and 238 deletions

View File

@@ -5,7 +5,7 @@ from __future__ import annotations
import unittest
from derive_easycom_interaction_events import Clip, Utterance
from prepare_easycom_eot_int_review import derive_protocol_candidates
from prepare_easycom_eot_int_review import CandidateConfig, derive_protocol_candidates
def utterance(uid: str, participant: int, start: float, end: float, text: str) -> Utterance:
@@ -50,18 +50,83 @@ class EOTINTReviewCandidateTests(unittest.TestCase):
def derive(self, rows: list[Utterance]):
return derive_protocol_candidates(rows, [clip()], session="Session_4")
def test_every_wearer_onset_is_an_unresolved_entry_candidate(self) -> None:
def test_eot_candidate_pairs_release_and_behavior_anchors(self) -> None:
rows = [
utterance("other", 4, 0.0, 1.0, "Your turn."),
utterance("wearer", 2, 1.2, 2.0, "My response."),
]
event = next(row for row in self.derive(rows) if row.candidate_source == "WEARER_ONSET")
self.assertEqual(event.automatic_protocol_label, "unsure")
self.assertEqual(event.rule_candidate_type, "ENTRY_CANDIDATE")
self.assertEqual(event.rule_candidate_type, "EOT_OPPORTUNITY_CANDIDATE")
self.assertEqual(event.anchor_type, "wearer_onset")
self.assertAlmostEqual(event.event_time, 1.2)
self.assertAlmostEqual(event.temporal_evidence["opportunity_anchor_seconds"], 1.0)
self.assertAlmostEqual(event.temporal_evidence["behavior_anchor_seconds"], 1.2)
self.assertAlmostEqual(event.temporal_evidence["response_delay_seconds"], 0.2)
self.assertAlmostEqual(event.temporal_evidence["wearer_speech_duration_after_onset_seconds"], 0.8)
def test_long_silence_does_not_cancel_eot_opportunity(self) -> None:
rows = [
utterance("other", 4, 0.0, 0.5, "Your turn."),
utterance("wearer", 2, 3.5, 4.5, "Delayed response."),
]
event = next(row for row in self.derive(rows) if row.candidate_source == "WEARER_ONSET")
self.assertAlmostEqual(event.temporal_evidence["response_delay_seconds"], 3.0)
def test_competing_other_onset_terminates_eot_opportunity(self) -> None:
rows = [
utterance("release", 4, 0.0, 1.0, "First turn."),
utterance("competitor", 3, 1.5, 3.0, "I take the floor."),
utterance("wearer", 2, 2.5, 3.5, "Wearer later."),
]
onset = next(
row for row in self.derive(rows)
if row.anchor_dense_seconds == 2.5 and row.anchor_type == "wearer_onset"
)
self.assertEqual(onset.candidate_source, "WEARER_ONSET_CONTROL")
self.assertEqual(onset.automatic_protocol_label, "not_applicable")
def test_short_wearer_pause_without_other_takeover_becomes_resume_control(self) -> None:
rows = [
utterance("wearer-1", 2, 0.2, 1.0, "I was saying"),
utterance("wearer-2", 2, 1.5, 2.5, "and then this."),
]
derived = self.derive(rows)
resume = next(row for row in derived if row.candidate_source == "WEARER_RESUME")
self.assertEqual(resume.task, "eot_response_control")
self.assertEqual(resume.rule_candidate_type, "WITHIN_TURN_RESUME_CONTROL")
self.assertEqual(resume.automatic_protocol_label, "not_applicable")
self.assertTrue(resume.temporal_evidence["floor_episode_merged"])
self.assertAlmostEqual(resume.temporal_evidence["previous_wearer_silence_duration_seconds"], 0.5)
self.assertFalse(any(
row.candidate_source == "WEARER_ONSET" and row.anchor_dense_seconds == 1.5
for row in derived
))
def test_short_feedback_does_not_split_wearer_floor_episode(self) -> None:
rows = [
utterance("wearer-1", 2, 0.2, 1.0, "First clause"),
utterance("feedback", 4, 1.1, 1.4, "Yeah."),
utterance("wearer-2", 2, 1.6, 2.5, "same turn continues."),
]
resume = next(row for row in self.derive(rows) if row.candidate_source == "WEARER_RESUME")
self.assertEqual(resume.temporal_evidence["feedback_like_other_ids_during_gap"], [4])
self.assertFalse(resume.temporal_evidence["sustained_other_takeover_during_gap"])
def test_short_gap_with_sustained_other_takeover_preserves_floor_reentry_but_not_eot(self) -> None:
rows = [
utterance("wearer-1", 2, 0.2, 1.0, "First wearer turn."),
utterance("other", 4, 1.0, 2.2, "I have the floor now."),
utterance("wearer-2", 2, 1.6, 2.5, "Wearer re-enters."),
]
reentry = next(
row for row in self.derive(rows)
if row.candidate_source == "WEARER_ONSET_CONTROL" and row.anchor_dense_seconds == 1.6
)
self.assertFalse(reentry.temporal_evidence["floor_episode_merged"])
self.assertTrue(reentry.temporal_evidence["sustained_other_takeover_during_gap"])
self.assertEqual(reentry.temporal_evidence["sustained_competing_other_ids_during_gap"], [4])
def test_same_other_resume_is_eot_midturn_hard_negative(self) -> None:
rows = [
utterance("first", 4, 0.0, 1.0, "First clause."),
@@ -93,7 +158,7 @@ class EOTINTReviewCandidateTests(unittest.TestCase):
event.temporal_evidence["incoming_speaker_id"],
)
def test_every_wearer_offset_is_an_unresolved_yield_candidate(self) -> None:
def test_other_crossing_wearer_offset_is_an_unresolved_yield_candidate(self) -> None:
rows = [
utterance("wearer", 2, 0.0, 2.0, "Wearer is speaking."),
utterance("other", 4, 1.0, 3.0, "I will take over now."),
@@ -103,6 +168,63 @@ class EOTINTReviewCandidateTests(unittest.TestCase):
self.assertEqual(event.rule_candidate_type, "YIELD_CANDIDATE")
self.assertEqual(event.anchor_type, "wearer_offset")
self.assertAlmostEqual(event.event_time, 2.0)
self.assertEqual(event.temporal_evidence["tolerance_crossing_other_ids"], [4])
self.assertEqual(event.temporal_evidence["eligible_new_other_onset_ids"], [4])
self.assertAlmostEqual(
event.temporal_evidence["eligible_new_other_onsets"][0]["onset_seconds"], 1.0
)
self.assertTrue(event.temporal_evidence["int_ordering_satisfied"])
def test_preexisting_other_crossing_offset_is_not_an_int_candidate(self) -> None:
rows = [
utterance("other", 4, 0.0, 3.0, "Other already holds the floor."),
utterance("wearer", 2, 1.0, 2.0, "Wearer enters and stops."),
]
event = next(row for row in self.derive(rows) if row.anchor_type == "wearer_offset")
self.assertEqual(event.candidate_source, "WEARER_OFFSET_CONTROL")
self.assertEqual(event.temporal_evidence["preexisting_crossing_other_ids"], [4])
self.assertFalse(event.temporal_evidence["int_ordering_satisfied"])
def test_other_starting_after_wearer_offset_is_a_separate_control(self) -> None:
rows = [
utterance("wearer", 2, 0.0, 2.0, "Wearer finishes."),
utterance("other", 4, 2.2, 3.0, "Later turn."),
]
event = next(row for row in self.derive(rows) if row.anchor_type == "wearer_offset")
self.assertEqual(event.candidate_source, "WEARER_OFFSET_CONTROL")
self.assertEqual(event.rule_candidate_type, "WEARER_OFFSET_CONTROL")
self.assertEqual(event.automatic_protocol_label, "not_applicable")
self.assertFalse(event.temporal_evidence["other_crosses_wearer_offset"])
def test_one_frame_boundary_tolerance_is_configurable(self) -> None:
rows = [
utterance("wearer", 2, 0.0, 2.0, "Wearer finishes."),
utterance("other", 4, 1.0, 2.0, "Boundary-aligned takeover."),
]
tolerant = self.derive(rows)
strict = derive_protocol_candidates(
rows,
[clip()],
session="Session_4",
config=CandidateConfig(int_offset_tolerance_seconds=0.0),
)
tolerant_offset = next(row for row in tolerant if row.anchor_type == "wearer_offset")
strict_offset = next(row for row in strict if row.anchor_type == "wearer_offset")
self.assertEqual(tolerant_offset.candidate_source, "WEARER_OFFSET")
self.assertEqual(strict_offset.candidate_source, "WEARER_OFFSET_CONTROL")
def test_internal_raw_offset_is_not_an_int_candidate(self) -> None:
rows = [
utterance("wearer-1", 2, 0.0, 1.0, "First clause."),
utterance("other", 4, 0.8, 1.3, "Yeah."),
utterance("wearer-2", 2, 1.2, 2.5, "Same floor episode."),
]
event = next(
row for row in self.derive(rows)
if row.anchor_type == "wearer_offset" and row.anchor_dense_seconds == 1.0
)
self.assertEqual(event.candidate_source, "WEARER_OFFSET_CONTROL")
self.assertTrue(event.temporal_evidence["raw_offset_is_within_merged_floor_episode"])
def test_brief_feedback_is_int_backchannel_hard_negative(self) -> None:
rows = [