Sciweavers

AAAI
2007

A Meta-learning Approach for Selecting between Response Automation Strategies in a Help-desk Domain

13 years 6 months ago
A Meta-learning Approach for Selecting between Response Automation Strategies in a Help-desk Domain
We present a corpus-based approach for the automation of help-desk responses to users’ email requests. Automation is performed on the basis of the similarity between a request and previous requests, which affects both the content included in a response and the strategy used to produce it. The latter is the focus of this paper, which introduces a meta-learning mechanism that selects between different information-gathering strategies, such as document retrieval and multi-document summarization. Our results show that this mechanism outperforms a random strategy-selection policy, and performs competitively with a gold baseline that always selects the best strategy.
Yuval Marom, Ingrid Zukerman, Nathalie Japkowicz
Added 02 Oct 2010
Updated 02 Oct 2010
Type Conference
Year 2007
Where AAAI
Authors Yuval Marom, Ingrid Zukerman, Nathalie Japkowicz
Comments (0)