Semantic Search of Educational Texts

Obtaining the most precise and relevant answer when asking a question can be difficult. Some search engines may only provide specific articles, websites, or educational texts or suggest some text from books, with huge texts or only a portion of the answer. However, for students to discover and comprehend a topic, some tool that can reduce the long texts to small paragraphs based on the searches, which would be easily readable and depict the same meaning, is required. Text summarization, a beneficial and significant component of Natural Language Processing (NLP), can be helpful in this regard. Semantic analysis, which employs computational algorithms, is carried out to derive meaning and context from the question posed. The paper aims to create a model to comprehend a given query and offer a precise question summary. The model utilizes Term Frequency-Inverse Document Frequency (TF-IDF) and the third-generation Generative Pre-trained Transformer (GPT-3). The two models were tested for summarizing a huge paragraph and for getting answers to various questions posed by the user, respectively. It was observed that both models successfully provided the results. The GPT-3 model was trained to offer the top five answers to any question and a summary of the results.