Unsupervised Event Graph Representation and Similarity Learning on Biomedical Literature
The automatic extraction of biomedical events from the scientific literature has drawn keen interest in the last several years, recognizing complex and semantically rich graphical interactions otherwise buried in texts. However, very few works revolve around learning embeddings or similarity metrics...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-12-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/22/1/3 |
| _version_ | 1797497725442326528 |
|---|---|
| author | Giacomo Frisoni Gianluca Moro Giulio Carlassare Antonella Carbonaro |
| author_facet | Giacomo Frisoni Gianluca Moro Giulio Carlassare Antonella Carbonaro |
| author_sort | Giacomo Frisoni |
| collection | DOAJ |
| description | The automatic extraction of biomedical events from the scientific literature has drawn keen interest in the last several years, recognizing complex and semantically rich graphical interactions otherwise buried in texts. However, very few works revolve around learning embeddings or similarity metrics for event graphs. This gap leaves biological relations unlinked and prevents the application of machine learning techniques to promote discoveries. Taking advantage of recent deep graph kernel solutions and pre-trained language models, we propose Deep Divergence Event Graph Kernels (DDEGK), an unsupervised inductive method to map events into low-dimensional vectors, preserving their structural and semantic similarities. Unlike most other systems, DDEGK operates at a graph level and does not require task-specific labels, feature engineering, or known correspondences between nodes. To this end, our solution compares events against a small set of anchor ones, trains cross-graph attention networks for drawing pairwise alignments (bolstering interpretability), and employs transformer-based models to encode continuous attributes. Extensive experiments have been done on nine biomedical datasets. We show that our learned event representations can be effectively employed in tasks such as graph classification, clustering, and visualization, also facilitating downstream semantic textual similarity. Empirical results demonstrate that DDEGK significantly outperforms other state-of-the-art methods. |
| first_indexed | 2024-03-10T03:23:21Z |
| format | Article |
| id | doaj.art-a72ec709a3fa49a5b30245e3affcd9a9 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-10T03:23:21Z |
| publishDate | 2021-12-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-a72ec709a3fa49a5b30245e3affcd9a92023-11-23T12:15:23ZengMDPI AGSensors1424-82202021-12-01221310.3390/s22010003Unsupervised Event Graph Representation and Similarity Learning on Biomedical LiteratureGiacomo Frisoni0Gianluca Moro1Giulio Carlassare2Antonella Carbonaro3Department of Computer Science and Engineering (DISI), University of Bologna, 40126 Bologna, ItalyDepartment of Computer Science and Engineering (DISI), University of Bologna, 40126 Bologna, ItalyIndependent Researcher, 48018 Faenza, ItalyDepartment of Computer Science and Engineering (DISI), University of Bologna, 40126 Bologna, ItalyThe automatic extraction of biomedical events from the scientific literature has drawn keen interest in the last several years, recognizing complex and semantically rich graphical interactions otherwise buried in texts. However, very few works revolve around learning embeddings or similarity metrics for event graphs. This gap leaves biological relations unlinked and prevents the application of machine learning techniques to promote discoveries. Taking advantage of recent deep graph kernel solutions and pre-trained language models, we propose Deep Divergence Event Graph Kernels (DDEGK), an unsupervised inductive method to map events into low-dimensional vectors, preserving their structural and semantic similarities. Unlike most other systems, DDEGK operates at a graph level and does not require task-specific labels, feature engineering, or known correspondences between nodes. To this end, our solution compares events against a small set of anchor ones, trains cross-graph attention networks for drawing pairwise alignments (bolstering interpretability), and employs transformer-based models to encode continuous attributes. Extensive experiments have been done on nine biomedical datasets. We show that our learned event representations can be effectively employed in tasks such as graph classification, clustering, and visualization, also facilitating downstream semantic textual similarity. Empirical results demonstrate that DDEGK significantly outperforms other state-of-the-art methods.https://www.mdpi.com/1424-8220/22/1/3event embeddinggraph representation learninggraph similarity learningmetric learninggraph kernelsgraph neural networks |
| spellingShingle | Giacomo Frisoni Gianluca Moro Giulio Carlassare Antonella Carbonaro Unsupervised Event Graph Representation and Similarity Learning on Biomedical Literature Sensors event embedding graph representation learning graph similarity learning metric learning graph kernels graph neural networks |
| title | Unsupervised Event Graph Representation and Similarity Learning on Biomedical Literature |
| title_full | Unsupervised Event Graph Representation and Similarity Learning on Biomedical Literature |
| title_fullStr | Unsupervised Event Graph Representation and Similarity Learning on Biomedical Literature |
| title_full_unstemmed | Unsupervised Event Graph Representation and Similarity Learning on Biomedical Literature |
| title_short | Unsupervised Event Graph Representation and Similarity Learning on Biomedical Literature |
| title_sort | unsupervised event graph representation and similarity learning on biomedical literature |
| topic | event embedding graph representation learning graph similarity learning metric learning graph kernels graph neural networks |
| url | https://www.mdpi.com/1424-8220/22/1/3 |
| work_keys_str_mv | AT giacomofrisoni unsupervisedeventgraphrepresentationandsimilaritylearningonbiomedicalliterature AT gianlucamoro unsupervisedeventgraphrepresentationandsimilaritylearningonbiomedicalliterature AT giuliocarlassare unsupervisedeventgraphrepresentationandsimilaritylearningonbiomedicalliterature AT antonellacarbonaro unsupervisedeventgraphrepresentationandsimilaritylearningonbiomedicalliterature |