• Publications
  • Influence
Sequence-based Structured Prediction for Semantic Parsing
TLDR
We propose an approach for semantic parsing that uses a recurrent neural network to map a natural language question into a logical form representation of a KB query. Expand
  • 76
  • 4
  • PDF
Log-Linear RNNs: Towards Recurrent Neural Networks with Flexible Prior Knowledge
TLDR
We introduce \emph{LL-RNNs} (Log-Linear RNNs), an extension of Recurrent Neural Networks that replaces the softmax output layer by a log-linear output layer, of which softmax is a special case. Expand
  • 7
  • 2
  • PDF
Symbolic Priors for RNN-based Semantic Parsing
TLDR
We propose to exploit various sources of prior knowledge: the well-formedness of the logical forms is modeled by a weighted context-free grammar; the likelihood that certain entities present in the input utterance are also present in a logical form is modeling by weighted finite-state automata. Expand
  • 4
  • PDF
Grammatical Sequence Prediction for Real-Time Neural Semantic Parsing
TLDR
We develop a generic approach for restricting the predictions of a seq2seq model to grammatically permissible continuations, which leads to a 74% speed-up on an in-house dataset with a large vocabulary, compared to a neural model without grammatical restrictions. Expand
  • 1
  • PDF
Move from Perturbed scheme to exponential weighting average
TLDR
In an online decision problem, one makes decisions often with a pool of decision sequence called experts but without knowledge of the future. Expand
Analysis on the Application of Technical Investigation Officer System
With the advancement of science and technology, courts are facing more and more technical cases, which makes judges in non-professional fields feel powerless. In order to overcome the shortcomings ofExpand
Orthogonality regularizer for question answering
TLDR
Learning embeddings of words and knowledge base elements is a promising approach for open domain question answering. Expand
Neural-Symbolic Learning for Semantic Parsing
TLDR
Our goal in this thesis is to build a system that answers a natural language question (NL) by representing its semantics as a logical form (LF) and then computing the answer by executing the LF over a knowledge base. Expand