An Argument-Marker Model for Syntax-Agnostic Proto-Role Labeling

Proceedings of the Eighth Joint Conference on Lexical and Computational Semantics (*SEM), pages 224–234 Minneapolis, June 6–7, 2019. c⃝2019 Association for Computational Linguistics 224 An Argument-Marker Model for Syntax-Agnostic Proto-Role Labeling Juri Opitz and Anette Frank Research Training Group AIPHES, Leibniz ScienceCampus “Empirical Linguistics and Computational Language Modeling” Department for Computational Linguistics 69120 Heidelberg {opitz,frank}@cl.uni-heidelberg.de Abstract Semantic proto-role labeling (SPRL) is an alternative to semantic role labeling (SRL) that moves beyond a categorical definition of roles, following Dowty’s feature-based view of proto-roles. This theory determines agent- hood vs. patienthood based on a participant’s instantiation of more or less typical agent vs. patient properties, such as, for example, voli- tion in an event. To perform SPRL, we develop an ensemble of hierarchical models with self- attention and concurrently learned predicate- argument-markers. Our method is competitive with the state-of-the art, overall outperforming previous work in two formulations of the task (multi-label and multi-variate Likert scale pre- diction). In contrast to previous work, our re- sults do not depend on gold argument heads derived from supplementary gold tree banks. 1 Introduction Deciding on a linguistically sound, clearly de- fined and broadly applicable inventory of seman- tic roles is a long-standing issue in linguistic the- ory and natural language processing. To alleviate issues found with classical thematic role invento- ries, Dowty (1991) argued for replacing categori- cal roles with a feature-based, composite notion of semantic roles, introducing the theory of seman- tic proto-roles (SPR). At its core, it proposes two prominent, composite role types: proto-agent and proto-patient. Proto-roles represent multi-faceted, possibly graded notions of agenthood or patient- hood. For example, consider the following sen- tence from Bram Stoker’s Dracula (1897): (1) He opened it [the letter] and read it gravely. ‘Davidsonian’ analyses based on SR and SPR of the event open are displayed in Figure 1. The SPR analysis provides more detail about the event and the roles of the involved entities. Whether an SR: ∃e � open(e) ∧ agent(e, c∗) ∧ theme(e, l∗) � SPR: ∃e � open(e) ∧ volition(e, c∗) ∧ aware(e, c∗) ∧ sentient(e, c∗) ∧ manipulated(e, l∗) ∧ changes- state(e, l∗) ∧ ... � Figure 1: Two different ‘Davidsionian’ event analyses of open. c∗ and l∗ refer to the count and letter entities. argument is considered an agent or patient follows from the proto-typical properties the argument ex- hibits: e.g., being manipulated is proto-typical for patient, while volition is proto-typical for an agent. Hence, in both events of (1) the count is deter- mined as agent, and the letter as patient. Only recently two SPR data sets have been published. Reisinger et al. (2015) developed a property-based proto-role annotation schema with 18 properties. One Amazon Mechanical Turk crowd worker (selected in a pilot annotation) an- swered questions such as how likely is it that the argument mentioned with the verb changes loca- tion? on a 5-point Likert or responded inapplica- ble. This dataset (news domain) will henceforth be denoted by SPR1. Based on the experiences from the SPR1 annotation process, …

Content truncated to 500 words