The paper introduces a novel approach to matching natural language sentences using convolutional neural networks (CNNs). The authors propose two deep CNN architectures, ARC-I and ARC-II, which combine hierarchical sentence modeling and the capture of rich matching patterns at different levels of abstraction. These models are designed to be generic and do not require prior knowledge of natural language, making them applicable to various matching tasks across different languages. The empirical study demonstrates the effectiveness of the proposed models on three diverse matching tasks: sentence completion, tweet-response matching, and paraphrase identification. The results show that the models outperform competitor methods, highlighting their superior performance in capturing complex linguistic structures and interactions. The paper also discusses the advantages and limitations of the proposed models, providing insights into the balance between representing matching patterns and the internal structures of sentences.The paper introduces a novel approach to matching natural language sentences using convolutional neural networks (CNNs). The authors propose two deep CNN architectures, ARC-I and ARC-II, which combine hierarchical sentence modeling and the capture of rich matching patterns at different levels of abstraction. These models are designed to be generic and do not require prior knowledge of natural language, making them applicable to various matching tasks across different languages. The empirical study demonstrates the effectiveness of the proposed models on three diverse matching tasks: sentence completion, tweet-response matching, and paraphrase identification. The results show that the models outperform competitor methods, highlighting their superior performance in capturing complex linguistic structures and interactions. The paper also discusses the advantages and limitations of the proposed models, providing insights into the balance between representing matching patterns and the internal structures of sentences.