Speech Communication - Special issue on interactive voice technology for telecommunication applications (IVITA '96)
BoosTexter: A Boosting-based Systemfor Text Categorization
Machine Learning - Special issue on information retrieval
On natural language call routing
Speech Communication - Special issue on interactive voice technology for telecommunication applications
Machine learning in automated text categorization
ACM Computing Surveys (CSUR)
Spoken language classification using hybrid classifier combination
International Journal of Hybrid Intelligent Systems
| Hi-index | 0.00 |
Automatic call routing is one of the most important issues in the call center domain. It can be modeled ---once performed the speech recognition of utterances--- as a text classification task. Nevertheless, in this case, texts are extremely small (just a few words) and there are a great number of narrow call-type classes. In this paper, we propose a text classification method specially suited to work on this scenario. This method considers a new weighting scheme of terms and uses a multiple stage classification approach with the aim of balance the rate of rejected calls (directed to a human operator) and the classification accuracy. The proposed method was evaluated on a Spanish corpus consisting of 24,638 call utterances achieving outstanding results: 95.5% of classification accuracy with a rejection rate of just 8.2%.