Abstract | ||
---|---|---|
This report documents our efforts to develop a Generation Challenges 2011 surface realization system by converting the shared task deep inputs to ones compatible with OpenCCG. Although difficulties in conversion led us to employ machine learning for relation mapping and to introduce several robustness measures into OpenCCG's grammar-based chart realizer, the percentage of grammatically complete realizations still remained well below results using native OpenCCG inputs on the development set, with a corresponding drop in output quality. We discuss known conversion issues and possible ways to improve performance on shared task inputs. |
Year | Venue | Keywords |
---|---|---|
2011 | ENLG | surface realization,deep input,corresponding drop,conversion issue,native openccg input,osu system,grammatically complete realization,shared task,output quality,development set,shared task input,grammar-based chart realizer |
Field | DocType | Citations |
Computer science,Grammar,Robustness (computer science),Chart,Artificial intelligence | Conference | 4 |
PageRank | References | Authors |
0.45 | 9 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Rajakrishnan Rajkumar | 1 | 94 | 6.72 |
Dominic Espinosa | 2 | 71 | 3.71 |
Michael White | 3 | 101 | 7.24 |