Combining symbolic and corpus-based approaches for the generation of successful referring expressions
نویسندگان
چکیده
We present an approach to the generation of referring expressions (REs) which computes the unique RE that it predicts to be fastest for the hearer to resolve. The system operates by learning a maximum entropy model for referential success from a corpus and using the model’s weights as costs in a metric planning problem. Our system outperforms the baselines both on predicted RE success and on similarity to human-produced successful REs. A task-based evaluation in the context of the GIVE-2.5 Challenge on Generating Instructions in Virtual Environments verifies the higher RE success scores of the system.
منابع مشابه
The Potsdam NLG systems at the GIVE-2.5 Challenge
We present the Potsdam natural language generation systems P1 and P2 of the GIVE-2.5 Challenge. The systems implement two different referring expression generation models from Garoufi and Koller (2011) while behaving identically in all other respects. In particular, P1 combines symbolic and corpus-based methods for the generation of successful referring expressions, while P2 is based on a purel...
متن کاملSpatial Descriptions as Referring Expressions in the MapTask Domain
We discuss work-in-progress on a hybrid approach to the generation of spatial descriptions, using the maps of the Map Task dialogue corpus as domain models. We treat spatial descriptions as referring expressions that distinguish particular points on the maps from all other points (potential ‘distractors’). Our approach is based on rule-based overgeneration of spatial descriptions combined with ...
متن کاملG-TUNA: a corpus of referring expressions in German, including duration information
Corpora of referring expressions elicited from human participants in a controlled environment are an important resource for research on automatic referring expression generation. We here present G-TUNA, a new corpus of referring expressions for German. Using images of furniture as stimuli similarly to the TUNA and D-TUNA corpora, our corpus extends on these corpora by providing data collected i...
متن کاملProceedings of the Corpus Linguistics 2005 Workshop on Using Corpora for Natural Language Generation
We report on a set of experiments using corpora to evaluate referring expression generation for spoken dialog. In dialog, participants frequently converge on the same referring expressions even if those referring expressions are inefficient. Existing rule-based algorithms for referring expression generation do not adequately model this adaptation. We extended two such algorithms with simple mod...
متن کاملJustifying Corpus-Based Choices in Referring Expression Generation
Most empirically-based approaches to NL generation elaborate on co-occurrences and frequencies observed over a corpus, which are then accommodated by learning algorithms. This method fails to capture generalities in generation subtasks, such as generating referring expressions, so that results obtained for some corpus cannot be transferred with confidence to similar environments or even to othe...
متن کامل