Educational Videos Subtitles’ Summarization Using Latent Dirichlet Allocation and Length Enhancement

Nowadays, people use online resources such as educational videos and courses. However, courses are mostly long thus, summarizing them will be valuable. The video contents (visual, audio, subtitles) could analyzed to generate textual summaries, i.e., notes. Videos’ subtitles contain significant information. Therefore, is effective concentrate on the necessary details. Most of existing studies used Term Frequency–Inverse Document Frequency (TF-IDF) Latent Semantic Analysis (LSA) models create lectures’ summaries. This study takes another approach applies Dirichlet Allocation (LDA), which proved its effectiveness in document summarization. Specifically, proposed LDA summarization model follows three phases. first phase aims prepare subtitle file for modelling by performing some preprocessing steps, removing stop words. In second phase, trained keywords list extract important sentences. Whereas third a summary generated based list. summaries were lengthy; length enhancement method has been proposed. For evaluation, authors developed manual “EDUVSUM” dataset. compared with manual-generated outlines using two methods, (i) Recall-Oriented Understudy Gisting Evaluation (ROUGE) (ii) human evaluation. performance LDA-based outperforms TF-IDF LSA. Besides reducing summaries’ length, did improve precision rates. Other domains, news videos, can apply