| """PropBank-style semantic-role label vocabulary for SRL supervision. |
| |
| The relation space is *shared* with UD: UD deprels keep ids ``[0, NUM_DEPREL)`` |
| (unchanged), and SRL roles occupy the contiguous block starting at ``SRL_BASE == |
| NUM_DEPREL``. SRL losses operate on the ``edge_logits`` slice |
| ``[SRL_BASE : SRL_BASE + NUM_SRL_ROLES]`` using **local** role ids ``[0, |
| NUM_SRL_ROLES)`` where local id 0 is ``NONE`` (this ordered pair is not an |
| argument edge). A model that does SRL therefore needs |
| ``graph_relation_types >= RELATION_VOCAB_SIZE``. |
| """ |
|
|
| from __future__ import annotations |
|
|
| from strata.data.ud_labels import IGNORE_INDEX, NUM_DEPREL |
|
|
| |
| SRL_ROLES: tuple[str, ...] = ( |
| "NONE", |
| "ARG0", "ARG1", "ARG2", "ARG3", "ARG4", "ARG5", "ARGA", |
| "ARGM-TMP", "ARGM-LOC", "ARGM-MNR", "ARGM-CAU", "ARGM-DIS", "ARGM-ADV", |
| "ARGM-MOD", "ARGM-NEG", "ARGM-PRP", "ARGM-DIR", "ARGM-EXT", "ARGM-PRD", |
| "ARGM-GOL", "ARGM-COM", "ARGM-REC", "ARGM-LVB", "ARGM-OTHER", "ARG-OTHER", |
| ) |
| SRL_ROLE_TO_LOCAL: dict[str, int] = {role: i for i, role in enumerate(SRL_ROLES)} |
| NONE_LOCAL = 0 |
| NUM_SRL_ROLES = len(SRL_ROLES) |
|
|
| SRL_BASE = NUM_DEPREL |
| RELATION_VOCAB_SIZE = SRL_BASE + NUM_SRL_ROLES |
|
|
| |
| PREDICATE_MARKER = "V" |
| NO_ROLE = "_" |
|
|
|
|
| def srl_role_to_local(role: str) -> int: |
| """Map a UP role string to a local SRL id (never ``NONE`` for a real role). |
| |
| Continuation/reference prefixes (``C-`` / ``R-``) are dropped to their base; |
| unrecognised core / modifier roles fall back to generic buckets. |
| """ |
|
|
| r = role.strip().upper() |
| if r.startswith(("C-", "R-")): |
| r = r[2:] |
| if r in SRL_ROLE_TO_LOCAL and r != "NONE": |
| return SRL_ROLE_TO_LOCAL[r] |
| if r.startswith("ARGM"): |
| return SRL_ROLE_TO_LOCAL["ARGM-OTHER"] |
| if r.startswith("ARG"): |
| return SRL_ROLE_TO_LOCAL["ARG-OTHER"] |
| return SRL_ROLE_TO_LOCAL["ARG-OTHER"] |
|
|
