Back to Search
Start Over
Building siamese attention-augmented recurrent convolutional neural networks for document similarity scoring.
- Source :
-
Information Sciences . Nov2022, Vol. 615, p90-102. 13p. - Publication Year :
- 2022
-
Abstract
- • We proposed a new architecture - the Siamese attention-augmented recurrent convolutional neural network (S-ARCNN). • We compared the performance of S-ARCNN with eight popular models for measuring document similarity. • Our model outperformed the state-of-the-art Transformer based model (Sentence BERT) by over 5% in F1. • Simply fitting an optimal decision threshold can significantly improve pre-trained BERT model for the new task. • S-ARCNN performed best in longer question pairs (length > = 50 words). Automatically measuring document similarity is imperative in natural language processing, with applications ranging from recommendation to duplicate document detection. State-of-the-art approach in document similarity commonly involves deep neural networks, yet there is little study on how different architectures may be combined. Thus, we introduce the Siamese Attention-augmented Recurrent Convolutional Neural Network (S-ARCNN) that combines multiple neural network architectures. In each subnetwork of S-ARCNN, a document passes through a bidirectional Long Short-Term Memory (bi-LSTM) layer, which sends representations to local and global document modules. A local document module uses convolution, pooling, and attention layers, whereas a global document module uses last states of the bi-LSTM. Both local and global features are concatenated to form a single document representation. Using the Quora Question Pairs dataset, we evaluated S-ARCNN, Siamese convolutional neural networks (S-CNNs), Siamese LSTM, and two BERT models. While S-CNNs (82.02% F1) outperformed S-ARCNN (79.83% F1) overall, S-ARCNN slightly outperformed S-CNN on duplicate question pairs with more than 50 words (39.96% vs. 39.42% accuracy). With the potential advantage of S-ARCNN for processing longer documents, S-ARCNN may help researchers identify collaborators with similar research interests, help editors find potential reviewers, or match resumes with job descriptions. [ABSTRACT FROM AUTHOR]
Details
- Language :
- English
- ISSN :
- 00200255
- Volume :
- 615
- Database :
- Academic Search Index
- Journal :
- Information Sciences
- Publication Type :
- Periodical
- Accession number :
- 160251254
- Full Text :
- https://doi.org/10.1016/j.ins.2022.10.032