conftrace_

2017 ACL ACL 2017

Improving Semantic Composition with Offset Inference

Abstract

AbstractCount-based distributional semantic models suffer from sparsity due to unobserved but plausible co-occurrences in any text collection. This problem is amplified for models like Anchored Packed Trees (APTs), that take the grammatical type of a co-occurrence into account. We therefore introduce a novel form of distributional inference that exploits the rich type structure in APTs and infers missing data by the same mechanism that is used for semantic composition.

🌉 Interdisciplinary Bridge - Artificial Intelligence and Knowledge & Reasoning and Natural Language Processing

📈 Trend Setter - Semantics

Authors

Thomas Kober, Julie Weeds, Jeremy Reffin, David Weir

Topics

Machine Learning > Core Methods > Representation Learning Machine Learning > Optimization & Theory > Bayesian Inference Natural Language Processing > Resources & Methods > Lexical Semantics Knowledge & Reasoning > Representation > Knowledge Representation Machine Learning > Bayesian & Probabilistic > Probabilistic Modeling Natural Language Processing > Understanding > Semantics

Keywords

bayesian inference probabilistic inference topic modeling hierarchical dirichlet process distributional semantics distributional semantic model semantic composition sparsity problem anchored packed tree distributional inference type structure

More from ACL 2017

All ACL 2017 papers →

STAIR Captions: Constructing a Large-Scale Japanese Image Caption Dataset

“Liar, Liar Pants on Fire”: A New Benchmark Dataset for Fake News Detection

English Multiword Expression-aware Dependency Parsing Including Named Entities

Learning Topic-Sensitive Word Representations

Temporal Word Analogies: Identifying Lexical Replacement with Diachronic Word Embeddings

Methodical Evaluation of Arabic Word Embeddings