• Corpus ID: 195346263

Unsupervised Hypernym Detection by Distributional Inclusion Vector Embedding

  title={Unsupervised Hypernym Detection by Distributional Inclusion Vector Embedding},
  author={Haw-Shiuan Chang and ZiYun Wang and Luke Vilnis and Andrew McCallum},
Modeling hypernymy, such as poodle is-a dog, is an important generalization aid to many NLP tasks, such as entailment, relation extraction, and question answering. Supervised learning from labeled hypernym sources, such as WordNet, limit the coverage of these models, which can be addressed by learning hypernyms from unlabeled text. Existing unsupervised methods either do not scale to large vocabularies or yield unacceptably poor accuracy. This paper introduces distributional inclusion vector… 

Figures and Tables from this paper

Term Definitions Help Hypernymy Detection

This work proposes a new paradigm, HyperDef, for hypernymy detection – expressing word meaning by encoding word definitions, along with context driven representation, which gets state-of-the-art results in multiple benchmarks.

New Embedded Representations and Evaluation Protocols for Inferring Transitive Relations

A new representation of types as hyper-rectangular regions, that generalize and improve on OE, and shows that some current protocols to evaluate transitive relation inference can be misleading, and offers a sound alternative.

Composition of Sentence Embeddings: Lessons from Statistical Relational Learning

This article builds on recent SRL models to address textual relational problems, showing that they are more expressive, and can alleviate issues from simpler compositions, and significantly improve the state of the art in both transferable sentence representation learning and relation prediction.

End-to-End Reinforcement Learning for Automatic Taxonomy Induction

This work presents a novel end-to-end reinforcement learning approach to automatic taxonomy induction from a set of terms that outperforms prior state-of-the-artTaxonomy induction methods up to 19.6% on ancestor F1.

Order Embeddings from Merged Ontologies using Sketching

A simple, low resource method to produce order embeddings from ontologies, which map words to vectors so that order relations on the words, such as hypernymy/hyponymy, are represented in a direct way.

Octet: Online Catalog Taxonomy Enrichment with Self-Supervision

This paper presents a self-supervised end-to-end framework, Octet, for Online Catalog Taxonomy Enrichmen, and proposes to distantly train a sequence labeling model for term extraction and employ graph neural networks to capture the taxonomy structure as well as the query-item-taxonomy interactions for term attachment.



Hierarchical Embeddings for Hypernymy Detection and Directionality

This work presents a novel neural model HyperVec, an unsupervised measure where embeddings are learned in a specific order and capture the hypernym–hyponym distributional hierarchy, which outperforms both state-of-the-art un supervised measures and embedding models on hypernymy detection and directionality, and on predicting graded lexical entailment.

Hypernyms under Siege: Linguistically-motivated Artillery for Hypernymy Detection

Comparison to the state-of-the-art supervised methods shows that while supervised methods generally outperform the unsupervised ones, the former are sensitive to the distribution of training instances, hurting their reliability.

Relations such as Hypernymy: Identifying and Exploiting Hearst Patterns in Distributional Vectors for Lexical Entailment

It is found that the model strongly learns to identify hypernyms using Hearst patterns, which are well known to be predictive of lexical relations, and is presented with a novel model which exploits this behavior as a method of feature extraction in an iterative procedure similar to Principal Component Analysis.

Learning to Distinguish Hypernyms and Co-Hyponyms

This work compares a novel approach based on training a linear Support Vector Machine on pairs of feature vectors with state-of-the-art methods based on distributional similarity and shows that the new supervised approach does better even when there is minimal information about the target words in the training data, giving a 15% reduction in error rate.

Improving Hypernymy Detection with an Integrated Path-based and Distributional Method

An improved path-based algorithm is suggested, in which the dependency paths are encoded using a recurrent neural network, that achieves results comparable to distributional methods.

Lifted Rule Injection for Relation Embeddings

A highly efficient method for incorporating implication rules into distributed representations for automated knowledge base construction and finds that the strong restriction of the entity-tuple embedding space does not hurt the expressiveness of the model and even acts as a regularizer that improves generalization.

HyperLex: A Large-Scale Evaluation of Graded Lexical Entailment

We introduce HyperLex—a data set and evaluation resource that quantifies the extent of the semantic category membership, that is, type-of relation, also known as hyponymy–hypernymy or lexical

EVALution 1.0: an Evolving Semantic Dataset for Training and Evaluation of Distributional Semantic Models

In this paper, we introduce EVALution 1.0, a dataset designed for the training and the evaluation of Distributional Semantic Models (DSMs). This version consists of almost 7.5K tuples, instantiating

Question Classification using Head Words and their Hypernyms

This work proposes head word feature and present two approaches to augment semantic features of such head words using WordNet and proposes a compact yet effective feature set.

Dual Tensor Model for Detecting Asymmetric Lexico-Semantic Relations

This work proposes the Dual Tensor model, a neural architecture with which it explicitly model the asymmetry and capture the translation between unspecialized and specialized word embeddings via a pair of tensors.