Evaluation of Rough Sets Data Preprocessing on Context-Driven Semantic Analysis with RNN

In the application examples of NLP (natural language learning), the rich semantic information in medical literature can extract characteristic target words through the training of RNN-LSTM (recurrent neural network -long shortterm memory). In the process of extracting these target words, we often encounter some wrong target words which cause RNN to reduce the hit rate and extend the training time. In this paper, we take Diabetes in medical research as an example, the data preprocessing of rough sets, and the word vector tagging for target word can improve the hit efficiency of the target words in the RNN-LSTM training process.