Summary
In this chapter, we built the summarization application for medical transcriptions. In the beginning, we listed the challenges in order to generate a good parallel corpus for the summarization task in the medical domain. After that, for our baseline approach, we used the already available Python libraries, such as PyTeaser and Sumy. In the revised approach, we used word frequencies to generate the summary of the medical document. In the best possible approach, we combined the word frequency-based approach and the ranking mechanism in order to generate a summary for medical notes.
In the end, we developed a solution, where we used Amazon's review dataset, which is the parallel corpus for the summarization task, and we built the deep learning-based model for summarization. I would recommend that researchers, community members, and everyone else come forward to build high-quality datasets that can be used for building some great data science applications for the health and medical domains...
