Over the last decade, the online search for biological information has progressed rapidly and has become an integral part of any scientific discovery process. Today, it is virtually impossible to conduct R&D in biomedicine without relying on the kind of Web resources developed and maintained by the NCBI. Indeed, each day millions of users search for biological information via NCBIs online Entrez system. However, finding data relevant to a users information need is not always easy in Entrez. Improving our understanding of the growing population of Entrez users, their information needs and the way in which they meet these needs opens opportunities to improve information services and information access provided by NCBI. Among all Entrez databases, PubMed is the most used one and often serves as an entry point for people to access related data in other Entrez databases. Tools to aid searching PubMed are query suggestion, expansion, and spelling correction. Dedicated best match algorithms aid navigational queries by ignoring minor errors and aid informational searches using machine learning to combine relevant signals such as article popularity, publication date and type, and query-document relevance score. Additional valuable aids including identifying related articles and author name disambiguation. PubMed Labs provides a place to trial and improve new search features. It features a clean and mobile-friendly design tailored specifically towards small screen devices and a platform for users to provide feedback guiding future work. While search has usually focused on full documents or references, the value of sentence search is rising. It can identify specific statements rather than whole articles on a general topic. Our new tool, LitSense, provides sentence level search, making sense of biomedical literature at sentence level. A specific use of sentence similarity is to aid the curation efforts in the Conserved Domain Database (CDD). To this end, LitSense has been used to both finds sentences in PubMed articles already used to create CDD summaries and identify new sentences closely related to existing CDD summaries. For using sentences in Deep Learning tasks, BioSentVec is the first sentence encoder specifically for the biomedical domain. It better captures biomedical semantics than general domain encoders. Of course, word embeddings remain the primary method of using Deep Learning in NLP tasks. BioWordVec uses subword information and MeSH to generate biomedical word embeddings that can significantly improve performance. Biomedical terminology often includes important subword information. The semantic information available in ontologies such as MeSH is meaningful. A generic word embedding cannot take advantage of this valuable supplemental information. These machine learning methods benefit from having a large amount of text available. To that end a Web API serves BioC versions of the PMC Open Access Subset and Author Manuscripts. This is a continuously updated complement to our existing FTP service. The documents are available in either JSON or XML and both ASCII and Unicode encodings are available.
Showing the most recent 10 out of 18 publications
