NeuSub: A Neural Submodular Approach for Citation Recommendation
Citation recommendation is a task that aims to automatically select suitable references for a working manuscript. This task has become increasingly urgent as the typical pools of candidates continue to grow, in the order of tens or hundreds of thousands or more. While several approaches to citation...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2021-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9576707/ |
| _version_ | 1798033141650161664 |
|---|---|
| author | Binh Thanh Kieu Inigo Jauregi Unanue Son Bao Pham Hieu Xuan Phan Massimo Piccardi |
| author_facet | Binh Thanh Kieu Inigo Jauregi Unanue Son Bao Pham Hieu Xuan Phan Massimo Piccardi |
| author_sort | Binh Thanh Kieu |
| collection | DOAJ |
| description | Citation recommendation is a task that aims to automatically select suitable references for a working manuscript. This task has become increasingly urgent as the typical pools of candidates continue to grow, in the order of tens or hundreds of thousands or more. While several approaches to citation recommendation have been proposed in the literature, they generally seem to lack principled mechanisms to ensure diversity and other global properties among the recommended citations. For this reason, in this paper we propose a novel citation recommendation approach that leverages a submodular scoring function and a deep document representation to achieve an effective trade-off between relevance to the query and diversity of the references. To optimally train the scoring function and the deep representation, we propose a novel training objective based on a structural/multiclass hinge loss and incremental recommendations. The experimental results over three popular citation datasets have showed that the proposed approach has led to remarkable accuracy improvements, with an increase of up to 1.91 pp of MRR and 3.29 pp of F1@100 score with respect to a state-of-the-art citation recommendation system. |
| first_indexed | 2024-04-11T20:24:26Z |
| format | Article |
| id | doaj.art-01ecc9f92f9244a58182820bc5f5c16a |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-04-11T20:24:26Z |
| publishDate | 2021-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-01ecc9f92f9244a58182820bc5f5c16a2022-12-22T04:04:41ZengIEEEIEEE Access2169-35362021-01-01914845914846810.1109/ACCESS.2021.31207279576707NeuSub: A Neural Submodular Approach for Citation RecommendationBinh Thanh Kieu0https://orcid.org/0000-0001-7233-9271Inigo Jauregi Unanue1https://orcid.org/0000-0001-6223-9584Son Bao Pham2Hieu Xuan Phan3https://orcid.org/0000-0002-7640-9190Massimo Piccardi4https://orcid.org/0000-0001-9250-6604Faculty of Engineering and Information Technology, University of Technology Sydney, Broadway, NSW, AustraliaFaculty of Engineering and Information Technology, University of Technology Sydney, Broadway, NSW, AustraliaFaculty of Engineering and Information Technology, University of Technology Sydney, Broadway, NSW, AustraliaFaculty of Information Technology, VNU University of Engineering and Technology, Hanoi, VietnamFaculty of Engineering and Information Technology, University of Technology Sydney, Broadway, NSW, AustraliaCitation recommendation is a task that aims to automatically select suitable references for a working manuscript. This task has become increasingly urgent as the typical pools of candidates continue to grow, in the order of tens or hundreds of thousands or more. While several approaches to citation recommendation have been proposed in the literature, they generally seem to lack principled mechanisms to ensure diversity and other global properties among the recommended citations. For this reason, in this paper we propose a novel citation recommendation approach that leverages a submodular scoring function and a deep document representation to achieve an effective trade-off between relevance to the query and diversity of the references. To optimally train the scoring function and the deep representation, we propose a novel training objective based on a structural/multiclass hinge loss and incremental recommendations. The experimental results over three popular citation datasets have showed that the proposed approach has led to remarkable accuracy improvements, with an increase of up to 1.91 pp of MRR and 3.29 pp of F1@100 score with respect to a state-of-the-art citation recommendation system.https://ieeexplore.ieee.org/document/9576707/Citation recommendationdeep neural networksstructural/multiclass hinge losssubmodular inferencetransformer modelsBERT |
| spellingShingle | Binh Thanh Kieu Inigo Jauregi Unanue Son Bao Pham Hieu Xuan Phan Massimo Piccardi NeuSub: A Neural Submodular Approach for Citation Recommendation IEEE Access Citation recommendation deep neural networks structural/multiclass hinge loss submodular inference transformer models BERT |
| title | NeuSub: A Neural Submodular Approach for Citation Recommendation |
| title_full | NeuSub: A Neural Submodular Approach for Citation Recommendation |
| title_fullStr | NeuSub: A Neural Submodular Approach for Citation Recommendation |
| title_full_unstemmed | NeuSub: A Neural Submodular Approach for Citation Recommendation |
| title_short | NeuSub: A Neural Submodular Approach for Citation Recommendation |
| title_sort | neusub a neural submodular approach for citation recommendation |
| topic | Citation recommendation deep neural networks structural/multiclass hinge loss submodular inference transformer models BERT |
| url | https://ieeexplore.ieee.org/document/9576707/ |
| work_keys_str_mv | AT binhthanhkieu neusubaneuralsubmodularapproachforcitationrecommendation AT inigojauregiunanue neusubaneuralsubmodularapproachforcitationrecommendation AT sonbaopham neusubaneuralsubmodularapproachforcitationrecommendation AT hieuxuanphan neusubaneuralsubmodularapproachforcitationrecommendation AT massimopiccardi neusubaneuralsubmodularapproachforcitationrecommendation |