English Education Text Recommendation Technology Based on Word Embedding

Text is a common way for learners to obtain information. But the resources for English teaching texts on the Internet are very large. Because of language barriers, learners cannot select them well in a short time. In order to solve this problem, this paper uses word embedding technology to generate word vectors for each word in the text through the model. Text vectors are expressed in two ways: one is the average vector of all words, and the other is the average vector of keywords calculated by TF-IDF. Text similarity is calculated by calculating the cosine distance between text vectors. The comparison results show that the keyword method is better. It solves the recommendation problem of teaching texts.