BioSentVec: creating sentence embeddings for biomedical texts

22 Oct 2018  ยท  Qingyu Chen, Yifan Peng, Zhiyong Lu ยท

Sentence embeddings have become an essential part of today's natural language processing (NLP) systems, especially together advanced deep learning methods. Although pre-trained sentence encoders are available in the general domain, none exists for biomedical texts to date. In this work, we introduce BioSentVec: the first open set of sentence embeddings trained with over 30 million documents from both scholarly articles in PubMed and clinical notes in the MIMIC-III Clinical Database. We evaluate BioSentVec embeddings in two sentence pair similarity tasks in different text genres. Our benchmarking results demonstrate that the BioSentVec embeddings can better capture sentence semantics compared to the other competitive alternatives and achieve state-of-the-art performance in both tasks. We expect BioSentVec to facilitate the research and development in biomedical text mining and to complement the existing resources in biomedical word embeddings. BioSentVec is publicly available at https://github.com/ncbi-nlp/BioSentVec

PDF Abstract

Datasets


Results from the Paper


 Ranked #1 on Sentence Embeddings For Biomedical Texts on MedSTS (using extra training data)

     Get a GitHub badge
Task Dataset Model Metric Name Metric Value Global Rank Uses Extra
Training Data
Result Benchmark
Sentence Embeddings For Biomedical Texts BIOSSES BioSentVec (PubMed + MIMIC-III) Pearson Correlation 0.795 # 7
Sentence Embeddings For Biomedical Texts BIOSSES BioSentVec (MIMIC-III) Pearson Correlation 0.350 # 12
Sentence Embeddings For Biomedical Texts BIOSSES BioSentVec (PubMed) Pearson Correlation 0.817 # 4
Sentence Embeddings For Biomedical Texts BIOSSES Universal Sentence Encoder Pearson Correlation 0.345 # 13
Sentence Embeddings For Biomedical Texts MedSTS BioSentVec (PubMed + MIMIC-III) Pearson Correlation 0.767 # 1
Sentence Embeddings For Biomedical Texts MedSTS BioSentVec (MIMIC-III) Pearson Correlation 0.759 # 2
Sentence Embeddings For Biomedical Texts MedSTS BioSentVec (PubMed) Pearson Correlation 0.750 # 3
Sentence Embeddings For Biomedical Texts MedSTS Universal Sentence Encoder Pearson Correlation 0.714 # 4

Methods


No methods listed for this paper. Add relevant methods here