Тип публикации: статья из журнала
Год издания: 2016
Аннотация: The paper presents the investigation of collectives of term weighting methods for natural language call routing. The database consists of user utterances recorded in English language from caller interactions with commercial automated agents. Utterances from this database are labelled by experts and divided into 20 classes. Seven diПоказать полностьюfferent unsupervised and supervised term weighting methods were tested and compared with each other for classification with k-NN. Also a novel feature extraction method based on terms belonging to classes was applied. After that different combinations of term weighting methods were formed as collectives and used for meta-classification with rule induction. The numerical experiments have shown that the combination of two best term weighting methods (Term Relevance Ratio and Confident Weights) increases classification effectiveness in comparison with the best individual term weighting method significantly.
Журнал: Lecture Notes in Electrical Engineering
Выпуск журнала: Т. 370
Номера страниц: 99-110
ISSN журнала: 18761100