A substantial amount of both computational and linguistic research on event representations has focused on categorical, coarse thematic roles such as AGENT and PATIENT. In spite of this, there is evidence going back to Dowty 1991 that this is the wrong grain size to look for thematic information: the appearance of roles like AGENT represents a conglomeration of finer-grained thematic properties such as causation, volition, change-of-state, etc.
In this line of work, we develop a semantic proto-role annotation task, and shown that it readily leads to quality annotations with minimal training in crowd-sourcing situations. Decomposing thematic roles into fine-grained proto-role properties provides both theoretical and practical gains: not only does it lead to a better account of the mapping of semantic representations to syntax, it also leads to more tractable semantic annotation tasks.
Data
| Train | Dev | Test | Download | Citation |
|---|---|---|---|---|
| 7800 | 969 | 969 | v1 (tar.gz) | Reisinger et al. 2015 |
| 4877 | 632 | 582 | v2 (tar.gz) | White et al. 2016 |
References
Researchers
 Rachel Rudinger |
 Aaron Steven White |
 Benjamin Van Durme |
 Kyle Rawlins |
 Frank Ferraro |
 Craig Harman |