The goal of this theme session, proposed for the 9th International Conference on Construction Grammar, is to bring together (frame) semanticists, (construction) grammarians, and computational linguists interested in advancing the role(s) of frames and constructions in computational semantics—or conversely, advancing the role of computational approaches in the characterization of frames and constructions. We seek contributions that address the following question broadly construed: How does the work inform the understanding of computing the meaning of a frame or construction?
We invite abstracts describing computational work involving semantic frames (Fillmore 2012, inter alia) and/or grammatical constructions, i.e. form-meaning pairings (Fillmore et al. 2012). We are especially interested in novel approaches to familiar natural language processing techniques and tasks. Submissions should use theoretical or empirical methods to address the questions such as: How do frames/constructions figure into computational models of human processing, production, and language acquisition? How can natural language processing systems exploit frames/constructions for semantic interpretation and other tasks? Topics of interest include, but are not limited to:
- Semantic role labeling
- Semantic parsing with frames and/or constructions
- Models of language acquisition & change
- Lexicon/grammar induction with frames and/or constructions
- Selectional preferences
- Disambiguation of semantic relations (e.g. in compounds or other multiwords)
- Events, temporal structure, causality
- Script/narrative induction & reasoning
- Multiword expressions & idioms
- Construction detection/disambiguation/parsing
- Computational construction grammar formalisms (e.g. Embodied Construction Grammar, Fluid Construction Grammar, Sign-Based Construction Grammar, HPSG)
- Grounded semantics (e.g. vision, space & time, actions/robotics,neuroscience)
- Frames or constructions in semantic similarity, paraphrasing, entailment, generation, discourse processing, translation, sentiment analysis, and other applications.
This theme session is specifically designed for work that involves algorithms, formalisms, or systems, not (exclusively) data resources or annotation schemes.
Please submit your abstract (max. 500 words) in .pdf format directly to: miriamp [at] icsi.berkeley.edu.
Important Dates:
- Abstract submission: January 5, 2016
- Preliminary notification of acceptance: January 31, 2016
Presentation at the theme session is contingent on the acceptance of the theme session to the conference. If an abstract is not accepted for the theme session, authors can submit abstracts to the general conference. Should the theme session NOT be accepted, authors may submit abstracts to the general conference as well.
We are delighted to announce that Springer is interested in publishing papers from this theme session in Text, Speech, and Language Technology (Nancy Ide, series editor), subject to review. We are open to published material for the session, but only previously unpublished work can appear in the book.
References:
- Fillmore, Charles J. 2012. Encounters with Language. Computational Linguistics, 38.4: 701-718.
- Fillmore, Charles J., Russell Lee-Goldman, and Russell Rhodes. 2012. The FrameNet Constructicon. In Sag, Ivan. A. and Hans. C. Boas (eds.) Sign-based Construction Grammar. Stanford: CSLI.