Preface

摘要

Biomedical NLP deals with the processing of healthcare-related text-clinical documents created by physicians and other healthcare providers at the point of care, scientific publications in the areas of biology and medicine, and consumer healthcare text such as social media blogs. Recent years have seen dramatic changes in the types and amount of data available to researchers in this field. Where most research on publications in the past has dealt with the abstracts of journal articles, we now have access to the full texts of journal articles via PubMedCentral. Where research on clinical documents has been hampered by a lack of availability of data, we now have access to large bodies of data through the auspices of the Cincinnati Children's Hospital NLP Challenge, the i2b2 shared tasks (www.i2b2.org), the TREC Electronic Medical Records track, Clinical TempEval series of tasks, the US-funded Strategic Health Advanced Research Projects Axea 4 (www.sharpn.org) and the Shared Annotated Resources (ShARe; https://sites.google.com/site/shareclefehealth/taskdescription; www.clinicalnlpannotations.org) project. Meanwhile, the number of abstracts in PubMed continues to grow exponentially. Text in the form of blogs created by patients discussing various healthcare topics has emerged as another data source, with a new perspective on healthrelated issues. Connecting the information from the three main sources in multiple languages to the scientific community, the healthcare provider, and the healthcare consumer presents new challenges.