Intelligent Mobile App for You-Tube Video Selection

Abstract

The usage of mobile devices has increased dramatically in recent years. These devices serve us in many practical ways and provide us with many services – many of them in real-time and “on demand”. The delivery of streaming audio, streaming video and internet content to these devices has become common place. One of the most popular sources of video/audio content is You-Tube – where billions of videos are uploaded and accessed each day. An increasing challenge is to locate the desired video from among many dozens of possibilities. This paper introduces an intelligent mobile application that utilizes the You-Tube Application Programming Interface (API) in developing a novel algorithm for selecting the most appropriate video associated with a song title and artist. The test case used in this application is the domain of all the top 40 popular songs (as provided by Billboard Inc.) from 1970 to present – resulting in approximately 14,000 possible songs. The application described in this work invokes the You-Tube API based on 3 different criteria – most popular video, most relevant video and a key word search of the video’s comments. These criteria are then merged in a voting algorithm thus providing the best possible video pertaining to a song title and artist. This system, implemented for iOS 9 using XCode and Swift, allows the user to choose from thousands of song titles and provide a “music on demand” system therefore playing the best possible video associated with the song title and artist. Results for the app consist of randomly selecting 50 songs from each year (1867 total) and verifying that the most appropriate video was selected.