Investigation of traditional and deep neural sequence models for biomedical concept recognition
The goal of this dissertation is to improve state-of-the-art biomedical concept recognition. The task has a long history, but performance improvement is still an open research question. The dissertation has three main aims---in the first aim, which is presented in Chapter 3, we investigate traditional biomedical concept recognition systems. The challenges in the task are identified and five post-processing components are introduced to improve precision by removing false positive annotations. In the second aim, which is described in Chapter 4}, a two-stage machine learning system for biomedical ...
(For more, see "View full record.")