Injecting Relational Structural Representation in Neural Networks for\n Question Similarity

2018/06/20 by Antonio Emmanuele Uva, Uva, Antonio, Daniele Bonadiman +3
Computer Science · #Computation and Language (cs.CL) #FOS: Computer and information sciences #Natural Language Processing Techniques #Text Readability and Simplification #Topic Modeling

paper · pdf · doi:10.48550/arxiv.1806.08009

openalex publication_date 2018/06/20 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Effectively using full syntactic parsing information in Neural Networks (NNs)\nto solve relational tasks, e.g., question similarity, is still an open problem.\nIn this paper, we propose to inject structural representations in NNs by (i)\nlearning an SVM model using Tree Kernels (TKs) on relatively few pairs of\nquestions (few thousands) as gold standard (GS) training data is typically\nscarce, (ii) predicting labels on a very large corpus of question pairs, and\n(iii) pre-training NNs on such large corpus. The results on Quora and SemEval\nquestion similarity datasets show that NNs trained with our approach can learn\nmore accurate models, especially after fine tuning on GS.\n

Related

• Modeling Relational Information in Question-Answer Pairs with Convolutional Neural Networks
• Neural Network Models for Paraphrase Identification, Semantic Textual Similarity, Natural Language Inference, and Question Answering
• Combining word embeddings and convolutional neural networks to detect duplicated questions
• Learning Semantic Textual Similarity from Conversations
• Siamese Neural Networks with Random Forest for detecting duplicate question pairs