Belief tracking is a core component of modern spoken dialogue system pipelines. However, most current approaches would have difficulty scaling to larger, more complex dialogue domains. This is due to their dependency on either: a) Spoken Language Understanding models that require large amounts of annotated training data; or b) hand-crafted semantic lexicons that capture the lexical variation in users' language. We propose a novel Neural Belief Tracking (NBT) framework which aims to overcome these problems by building on recent advances in semantic representation learning. The NBT models reason over continuous distributed representations of words, utterances and dialogue context. Our evaluation on two datasets shows that this approach overcomes both limitations, matching the performance of state-of-the-art models that have greater resource requirements.
from cs.AI updates on arXiv.org http://ift.tt/1UgMCmM
via IFTTT
No comments:
Post a Comment