natural-language-processing-8

Semantic Composition

3260 papers • 126 benchmarks • 313 datasets

Understanding the meaning of text by composing the meanings of the individual words in the text (Source: https://arxiv.org/pdf/1405.7908.pdf)

(Image credit: Papersgraph)

Benchmarks

These leaderboards are used to track progress in semantic-composition-8

Trend

Dataset

Best Model

Actions

No benchmarks available.

Libraries

i

Use these libraries to find semantic-composition-8 models and implementations

Datasets

Polish CDSCorpus

Cryptics

Subtasks

No subtasks available.

Most implemented papers

The Lifted Matrix-Space Model for Semantic Composition

Samuel R. Bowman, Woojin Chung•Wed Nov 08 2017

The Lifted Matrix-Space model is introduced, which uses a global transformation to map vector word embeddings to matrices, which can be composed via an operation based on matrix-matrix multiplication, and is found to consistently outperforms TreeLSTM, the previous best known composition function for tree-structured models.

6

Content

0

Paper Graph

TOWARDS HIERARCHICAL IMPORTANCE ATTRIBUTION: EXPLAINING COMPOSITIONAL SEMANTICS FOR NEURAL SEQUENCE MODELS

X. Xue, Xiang Ren, Xisen Jin, Zhongyu Wei, Junyi Du•Tue Dec 31 2019

This paper proposes a formal and general way to quantify the importance of each word and phrase and proposes Sampling and Contextual Decomposition (SCD) algorithm and Sampling and Occlusion (SOC) algorithm, which outperform prior hierarchical explanation algorithms.

54 0

Paper Graph

SentiBERT: A Transferable Transformer-Based Architecture for Compositional Sentiment Semantics

Kai-Wei Chang, Da Yin, Tao Meng•Thu May 07 2020

SentiBERT is better than baseline approaches in capturing negation and the contrastive relation and model the compositional sentiment semantics, and can be transferred to other sentiment analysis tasks as well as related tasks, such as emotion classification tasks.

148 0

Paper Graph

SnapMix: Semantically Proportional Mixing for Augmenting Fine-grained Data

D. Tao, Xinchao Wang, Shaoli Huang•Tue Dec 08 2020

Semantically Proportional Mixing (SnapMix) is proposed that exploits class activation map (CAM) to lessen the label noise in augmenting fine-grained data and consistently outperforms existing mixed-based approaches regardless of different datasets or network depths.

131 0

Paper Graph

Domain and Function: A Dual-Space Model of Semantic Relations and Compositions

Peter D. Turney•Mon Apr 30 2012

A dual-space model is introduced that matches the performance of the best previous models for relations and compositions and can model relations, compositions, and other aspects of semantics.

195 0

Paper Graph

Modeling Relation Paths for Representation Learning of Knowledge Bases

Zhiyuan Liu, Yankai Lin, Maosong Sun, Huanbo Luan, S. Rao, Song Liu•Sun May 31 2015

This model considers relation paths as translations between entities for representation learning, and addresses two key challenges: (1) Since not all relation paths are reliable, it design a path-constraint resource allocation algorithm to measure the reliability of relation paths and (2) represents relation paths via semantic composition of relation embeddings.

648 0

Paper Graph

Higher-order logical inference with compositional semantics

K. Mineshima, D. Bekki, Yusuke Miyao, Pascual Martínez-Gómez•Mon Aug 31 2015

The results show that a system based on a reasonably-sized semantic lexicon and a manageable number of non-first-order axioms enables efficient logical inferences, including those concerned with generalized quantifiers and intensional operators, and outperforms the state-of-the-art firstorder inference system.

80 0

Paper Graph

Improving Sparse Word Representations with Distributional Inference for Semantic Composition

Thomas Kober, Julie Weeds, Jeremy Reffin, David J. Weir•Tue Aug 23 2016

This paper shows that distributional inference improves sparse word representations on several word similarity benchmarks and demonstrates that the model is competitive with the state-of-the-art for adjective-noun, noun- noun and verb-object compositions while being fully interpretable.

15 0

Paper Graph

Semantic Compositional Networks for Visual Captioning

L. Deng, Yunchen Pu, Zhe Gan, L. Carin, Jianfeng Gao, Xiaodong He, Chuang Gan, Kenneth Tran•Tue Nov 22 2016

Experimental results show that the proposed method significantly outperforms prior state-of-the-art approaches, across multiple evaluation metrics.

448 0

Paper Graph

Adding a benchmark result helps the community track progress.