Curated KBsFreebase Yago 2 Extracted KBsOpen IE NELL
知识库 Curated KBs(Freebase, Yago 2) Extracted KBs(Open IE, NELL) Entity linking Relation extraction
IEQA what is the name of Justin Bieber brother?
基于深度学习的QA Who first voiced Meg on Family Guy?
基于深度学习的QA Memory networks Attention Mechanism End-to-End …
Data. Sets
End-to-End An End-to-End Model for Question Answering over Knowledge Base with Cross-Attention Combining Global Knowledge Bordes et al. (2014 b) BOW method Bordes et al. (2014 a) subgraph embeddings Yang et al. (2014) SP-based and embeddings Dong et al. (2015) three columns of CNNs Bordes et al. (2015) put KB-QA into memory networks Our approach Bi-LSTM, cross-attention, global KB
Semantic Parsing Framework Executor Grammar Model Parser Learner
Semantic Parsing input utterances => logical forms Model theory and Compositionality Development:Weak supervision and Scaling up Challenge: linguistic(representation and compositionality) Statistical(Learn and Generalization) Computational(Efficiency)
Semantic Parsing Challenges Lexical gap Ambiguity Multilingualism Complex Queries Distributed Knowledge Procedural, Temporal and Spatial Questions Templates
Semantic Parsing Automated Template Generation for Question Answering over Knowledge Graphs Which actress played character [[Amy Squirrel | Amy. Squirrel]]on [[Bad Teacher | Bad. Teacher]]?
Semantic Parsing KBQA: Learning Question Answering over QA Corpora and Knowledge Bases
Semantic Parsing Language interpretation in a reinforcement learning setting, where an agent who presented with an utterance in some context performs some action, and receives a corresponding reward signal. This framework highlights the importance of contextdependence in language interpretation. Neural Symbolic Machines: Learning Semantic Parsers on Freebase with Weak Supervision
Improved Neural Relation Detection for Knowledge Base Question Answerin
Q&A Thank you for your listening.
- Slides: 22