Paper: Modeling Wisdom of Crowds Using Latent Mixture of Discriminative Experts

ACL ID P11-2058
Title Modeling Wisdom of Crowds Using Latent Mixture of Discriminative Experts
Venue Annual Meeting of the Association of Computational Linguistics
Session Main Conference
Year 2011
Authors

In many computational linguistic scenarios, training labels are subjectives making it nec- essary to acquire the opinions of multiple an- notators/experts, which is referred to as ”wis- dom of crowds”. In this paper, we propose a new approach for modeling wisdom of crowds based on the Latent Mixture of Discrimina- tive Experts (LMDE) model that can automat- ically learn the prototypical patterns and hid- den dynamic among different experts. Experi- ments show improvement over state-of-the-art approaches on the task of listener backchannel prediction in dyadic conversations.