Paper: On Redundancy” in Selecting Attributes for Generating Referring Expressions

ACL ID C08-2029
Title On Redundancy” in Selecting Attributes for Generating Referring Expressions
Venue International Conference on Computational Linguistics
Session Poster Session
Year 2008
Authors

We seek to develop an efficient algorithm selecting attributes that approximates hu- man selection. In contrast to previous work we sought to combine the strengths of cog- nitive theories and simple learning algo- rithms. We then developed a new algo- rithm for attribute selection based on ob- servations from a corpus, which outper- formed a simple base algorithm by a sig- nificant margin. We then carried out a de- tailed comparison between our algorithm and Reiter & Dale’s “Incremental Algo- rithm”. In terms of achieving a human-like attribute selection, the overall performance of both algorithms is fundamentally equiv- alent, while differing in the handling of re- dundancy in selected attributes. We further investigated this phenomenon and draw some conclusions for further improvem...