Fast and Efficient Multilingual Unified MOOCs Semantic Search Engine (UMSSE)

Abstract

Massive Open Online Courses (MOOCs) proliferation has created a demand for effective search engines to help learners identify and enrol in courses that meet their needs. However, building a multilingual unified MOOC search engine that can provide comprehensive search results has been challenging due to the many platforms in different languages and the diversity of available courses. This paper proposed and implemented a model for Unified MOOCs Semantic Search Engine (UMSSE). This UMSSE leveraged a combination of text encoder models and an Approximate Nearest Neighbor (ANN) algorithm to improve the speed and accuracy of search results. The model integrated data from the platform and multiple languages, utilizing Natural Language Processing (NLP) and machine learning techniques to understand the meaning of search queries and recommend relevant courses. The performance of the proposed model was evaluated using a dataset of various MOOC course descriptions. The results showed that it outperformed traditional keyword-based search engines regarding performance metrics. Several applied examples further illustrated how the proposed model could improve the speed and effectiveness of MOOCs search and recommend appropriate courses to users.