Deep Learning Adaptation with Word Embeddings for Sentiment Analysis on Online Course Reviews

摘要: Online educational platforms are enabling learners to consume a great variety of content and share opinions on their learning experience. The analysis the sentiment behind such collective intelligence represents key element for supporting both instructors institutions shaping offered Combining Word Embedding representations deep architectures has made possible design systems able accurately measure text polarity several contexts. However, application data still appears limited. Therefore, considering over-sensitiveness emerging models context where training is collected, conducting adaptation processes that target e-learning becomes crucial unlock full potential model. In this chapter, we describe approach that, starting from representations, measures textual reviews posted by after attending online courses. Then, demonstrate how Embeddings trained smaller e-learning-specific resources more effective with respect those bigger general-purpose resources. Moreover, show benefits achieved combining instead common machine models. We expect chapter will help stakeholders get clear view shape future research field.