Multiple topic identification in human/human conversations

@article{Bost2015MultipleTI,
  title={Multiple topic identification in human/human conversations},
  author={Xavier Bost and G. Senay and M. El-B{\`e}ze and R. Mori},
  journal={Comput. Speech Lang.},
  year={2015},
  volume={34},
  pages={18-42}
}
  • Xavier Bost, G. Senay, +1 author R. Mori
  • Published 2015
  • Computer Science
  • Comput. Speech Lang.
  • HighlightsA multiple classification methods for multiple theme hypothesization is proposed.Four methods, one of which is new, are initially used and separately evaluated.A new sequential decision strategy for multiple theme hypothesization is introduced.A new hypothesis refinancing component is presented, based on ASR word lattice.Results show that the strategy makes it possible to obtain reliable service surveys. The paper deals with the automatic analysis of real-life telephone conversations… CONTINUE READING

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 33 REFERENCES
    Multiple topic identification in telephone conversations
    • 1
    • PDF
    Context-sensitive learning methods for text categorization
    • 565
    • PDF
    Latent Dirichlet Allocation
    • 24,105
    • PDF
    Bayesian Unsupervised Topic Segmentation
    • 221
    • PDF
    Optimizing search engines using clickthrough data
    • 4,028
    • PDF
    Exploiting latent semantic information in statistical language modeling
    • 384
    • PDF