Skip to Main content Skip to Navigation
Journal articles

The MeSH-Gram Neural Network Model: Extending Word Embedding Vectors with MeSH Concepts for Semantic Similarity.

Saïd Abdeddaim 1 Sylvestre Vimard 1 Lina F. Soualmia 1
1 TIBS - LITIS - Equipe Traitement de l'information en Biologie Santé
LITIS - Laboratoire d'Informatique, de Traitement de l'Information et des Systèmes
Abstract : Eliciting semantic similarity between concepts remains a challenging task. Recent approaches founded on embedding vectors have gained in popularity as they have risen to efficiently capture semantic relationships. The underlying idea is that two words that have close meaning gather similar contexts. In this study, we propose a new neural network model, named MeSH-gram, which relies on a straightforward approach that extends the skip-gram neural network model by considering MeSH (Medical Subject Headings) descriptors instead of words. Trained on publicly available PubMed/MEDLINE corpus, MeSH-gram is evaluated on reference standards manually annotated for semantic similarity. MeSH-gram is first compared to skip-gram with vectors of size 300 and at several windows' contexts. A deeper comparison is performed with twenty existing models. All the obtained results with Spearman's rank correlations between human scores and computed similarities show that MeSH-gram (i) outperforms the skip-gram model and (ii) is comparable to the best methods that need more computation and external resources.
Document type :
Journal articles
Complete list of metadatas

https://hal-normandie-univ.archives-ouvertes.fr/hal-02285377
Contributor : Lina F Soualmia <>
Submitted on : Thursday, September 12, 2019 - 4:31:30 PM
Last modification on : Friday, September 13, 2019 - 1:21:35 AM

Identifiers

Citation

Saïd Abdeddaim, Sylvestre Vimard, Lina F. Soualmia. The MeSH-Gram Neural Network Model: Extending Word Embedding Vectors with MeSH Concepts for Semantic Similarity.. Studies in Health Technology and Informatics, IOS Press, 2019, 264, pp.5-9. ⟨10.3233/SHTI190172⟩. ⟨hal-02285377⟩

Share

Metrics

Record views

21