Analyzing Trends in Computer Science Toipcs on Social Media Using Content Mining with Youtube as Case Study
Loading...
Date
item.page.authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Researchers of this era have shifted their vision from traditional research methodologies to modern research techniques which are used to develop certain knowledge base systems. These knowledge based systems could be used to forecast, predict, identify patterns, detect anomaly, clustering, classification, recommendation, motif discovery, segmentation, trend analysis and more.
newlineThis shift of researchers is due to the fact that the user generated data such as blogs, wikis, tweets, forums, podcasts, chats, posts, advertisements and others are easily available in huge amounts. The researchers are now more interested in exploring the wide possibilities of research in different computer science domains.
newlineThis study may provide assistance to naive learners in identifying the trending topics of the Computer Science domain for their exploration. For the purpose of building such a system, the researcher has crawled 9539 Video Text Tracks (vtt) files of computer science domains videos uploaded during the years 2015 to 2019 from the most popular video search engine YouTube. The videos were crawled along with its meta files. The vtt files were then converted to plain text files as the research was based on content mining of the videos. These plain text files then undergo processing to be classified and clustered into a Computer Science Ontology domain. The keyphrase extraction algorithm RAKE (Rapid Automatic Keyphrase Extraction), Computer Science Ontology (CSO) and Random Forest machine learning algorithm has been used in the study.Once the clusters of plain text files for computer science ontology domains were obtained then the trend analysis has been performed by analyzing Pearson Correlation Coefficient (PCC) of metrics #likeCount #dislikeCount #avgRating with #viewcount. The model was built for trend analysis using these metrics and has been evaluated using R-Squared value. The computer science ontology domains which show positive upward trends are Internet, Computer Networks and Computer Hardware.
newline