Student research opportunities

Tracking viral videos on Youtube

Project Code: CECS_678

This project is available at the following levels:
Honours, Summer Scholar, Masters

Keywords:

youtube, social media

Supervisor:

Dr Lexing Xie

Outline:

Youtube is a massive living lab of video sharing and creation. But since video-making is a time-consuming process, how much of it is new?

This project investigate tools for tracking partial video duplicates. e.g. can you imagine (without watching them) that the two videos below will share something in common?
Such outcomes are excellent pre-requisites to studying social communication and large-scale video sharing on youtube.

Goals of this project

The project involve developing video analysis, feature extraction and matching tools. Project report will include benchmarking its performance against alternatives.

Stretch goals include performing analysis on the dynamics and social context of viral videos, make observations and predictions of the nature of virality.

Requirements/Prerequisites

- skills in one or more scripting languages.
- good analytical skills.

Student Gain

- get familiar with data protocols of web 2.0 data.
- hands-on experiences with video analysis and feature extraction tools.
- skills for bench-marking and speed optimization for large-scale matching.

Background Literature

Design and evaluation of an effective and efficient video copy detection system. A Natsev, M Hill, J R Smith, ICME 2010 Workshop on Visual Content Identification and Search (VCIDS'10).

John R. Kender, Matthew L. Hill, Apostol Natsev, John R. Smith, Lexing Xie: Video genetics: a case study from YouTube. ACM Multimedia 2010: 1253-1258