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.

Query auto completion (QAC) systems are a standard part of search engines in industry, helping users formulate their query. Such systems update their suggestions after the user types each character, predicting the user's intent using various signals - one of the most common being popularity. Recently, deep learning approaches have been proposed for the QAC task, to specifically address the main limitation of previous popularity-based methods: the inability to predict unseen queries. In this work we improve previous methods based on neural language modeling, with the goal of building an end-to-end system. We particularly focus on using real-world data by integrating user information for personalized suggestions when possible. We also make use of time information and study how to increase diversity in the suggestions while studying the impact on scalability. Our empirical results demonstrate a marked improvement on two separate datasets over previous best methods in both accuracy and scalability, making a step towards neural query auto-completion in production search engines.

PubMed ® and other literature databases contain a wealth of information on diseases and their diagnosis/treatment in the form of scientific publications. In order to take advantage of such rich information, several text-mining tools have been developed for automatically detecting mentions of disease names in the PubMed abstracts. The next important step is the normalization of the various disease names to standardized vocabulary entries and medical dictionaries. To this end, we present an automatic approach for mapping disease names in PubMed abstracts to their corresponding concepts in Medical Subject Headings (MeSH ® ) or Online Mendelian Inheritance in Man (OMIM ® ). For developing our algorithm, we merged disease concept annotations from two existing corpora. In addition, we hand annotated a separate test set of decease concepts for our method evaluation. Different from others, we reformulate the disease name normalization task as an information retrieval task where input queries are disease names and search results are disease concepts. As such, our inference method builds on existing Lucene search and further improves it by taking into account the string similarity of query terms to the disease concept name and synonyms. Evaluation results show that our method compares favorably to other state-of-the-art approaches. In conclusion, we find that our approach is a simple and effective way for linking disease names to controlled vocabularies and that the merged disease corpus provides added value for the development of text mining tools for named entity recognition from biomedical text. Data is available at http://www.ncbi.nlm.nih.gov/CBBresearch/Fellows/Dogan/disease.html