Abstract:A deep learning model adaptive to both sentence-level and article-level paraphrase identification is proposed in this paper. It consists of pairwise unit similarity feature and semantic context correlation feature. In this model, sentences are represented by word and phrase embedding while articles are represented by sentence embedding. Those phrase and sentence embedding are learned from parse trees through Weighted Unfolding Recursive Autoencoders (WURAE), an unsupervised learning algorithm. Then, unit similarity matrix is calculated by matching the pairwise lists of embedding. It is used to extract the pairwise unit similarity feature through CNN and k-max pooling layers. In addition, semantic context correlation feature is taken into account, which is captured by the combination of CNN and LSTM. CNN layers learn collocation information between adjacent units while LSTM extracts the long-term dependency feature of the text based on the output of CNN. This model is experimented on a famous English sentence paraphrase corpus, MSRPC, and a Chinese article paraphrase corpus. The results show that the deep semantic feature of text could be extracted based on WURAE, unit similarity and context correlation feature. We release our code of WURAE, deep learning model for paraphrase identification and pre-trained phrase end sentence embedding data for use by the community.

Paraphrase Identification with External Memory and SRL Knowledge

Bidirectional Long Short-Term Memory with Gated Relevance Network for Paraphrase Identification

Convolution Neural Network Based Syntactic and Semantic Aware Paraphrase Identification.

Paraphrase Recognition Via Combination of Neural Classifier and Keywords

Paraphrase Identification Based on ResNeXt

Semantic Role Labeling Using Recursive Neural Network

Semantic Role Labeling with Associated Memory Network

Sentence Representation Method Based on Multi-Layer Semantic Network

Syntax-aware Neural Semantic Role Labeling

A Memory-Based Sentence Split and Rephrase Model with Multi-task Training

A Deep Network Model for Paraphrase Detection in Short Text Messages

Paraphrase Detection with Dependency Embedding

Cross-lingual paraphrase identification

Paraphrase Identification Based on Weighted URAE, Unit Similarity and Context Correlation Feature.

Semantic Role Labeling Integrated with Multilevel Linguistic Cues and Bi-LSTM-CRF

Integrating Linguistic Knowledge to Sentence Paraphrase Generation.

Linguistic Knowledge as Memory for Recurrent Neural Networks.

Paraphrase Thought: Sentence Embedding Module Imitating Human Language Recognition

Chinese Semantic Role Labeling with Bidirectional Recurrent Neural Networks

Abstract Rule Learning for Paraphrase Generation

Paraphrase Generation with Collaboration Between the Forward and the Backward Decoder