Online Educational Video Recommendation System Analysis

Online Educational Video Recommendation System Analysis

Parvathi R., Aarushi Siri Agarwal, Urmila Singh
Copyright: © 2023 |Pages: 19
DOI: 10.4018/978-1-7998-9220-5.ch093
(Individual Chapters)
No Current Special Offers


Most online platforms that provide video content, including TEDx, usually use various recommendation systems to gather more viewers. These videos are recommended based on various criteria. They can be either based on the user behavior and history of watched videos or on the basis of generally liked videos. The aim of this work is to conduct an in-depth analysis of the education platform called TEDx. This analysis will help in deriving the current protocols and thresholds this platform follows to curate and recommend videos to new users of the platform. The end goal is to figure out the various correlations between different parameters pertaining to these videos and on this basis to derive concrete illustrative representations of said relations and also to build a framework around these facts to find the exact relation between various videos on the platform.
Chapter Preview


Systems designed based on content-based recommendation explicitly analyze parameters, considering feedback, descriptions of previously rated items, and work on further building a model based on the candidate specification and similarity metrics. Several channels can be used up for generating relevant feedback and specifications required. New interesting items are then recommended on the basis of the structure model developed. Highly positive and accurate recommendations require good hand-engineered features. Semantic analysis (lexicons and ontologies) are used by other members of Content-based RSs for enhanced and accurate item representation. Furthermore (Kempe et al., 2003) discussed one of the most popular algorithms for extracting clusters in a graph is proposed by New-man and Girvan (Newman and Girvan, 2004), which is based on modularity. (Zhou et al., 2016), developed a modularity based graph clustering approach, which can help users discover common interests of other users quite effectively. It obtains an undirected weighted tag-graph for each user. One of other methods is the Content-based video retrieval (CBVR) technique, which has been widely used in content-based video lecture recommendation systems (Zhou et al., 2016). OCR, ASR techniques, and folksonomy are very widely used in annotation tasks for information retrieval for user preference. In most of the related works, the solidity and consistency problems are caused by varying accuracy of different analysis engines (Yang and Meinel, 2014), which has not been thoroughly discussed. (Wingrove 2022; Holland et al., 2022).

Key Terms in this Chapter

K-Means Clustering: Used to partition n observations into k groups, where each observation belongs to cluster with nearest mean.

Principal Component Analysis: A method often used to reduce the dimensions of a large dataset.

Term Frequency: Inverse document frequency used to get important words in a set of documents by calculating frequency of occurrence for unique words in the document Silhouette score method to validate consistency within a cluster of data.

Collaborative Filtering: Uses user ratings and gives more personalized recommendations.

Content-Based Filtering: The use of certain features on the basis of likes, comments, reactions and explicit feedback to escalate recommendation for other items.

Word Cloud: A visual representation of text data and keywords in a particular corpus.

Complete Chapter List

Search this Book: