The ValueTube project aims to build a machine learning algorithm which can ascertain the values held by YouTube videos from their metadata and comment section. To compliment this we shall also build a web interface for this “more human focused” search algorithm/filter, which allows people to search for and view YouTube videos which align with their values. The values model used by our project follows the Schwartz theory of basic values in order to provide a comprehensive range of values, and a fixed point to reference them from. This project is being built as part of the CSIT321 – Project subject at the University of Wollongong, this subject is the capstone for the Bachelor of Computer Science and Bachelor of Information Technology.

Located in this repository is the software aspects of this project including all source code, full documentation on the contents of this repository and instructions on how to setup a development environment as well as the data used to train our machine learning algorithms.

As part of this project the ValueTube team will have to undertake several milestones of work to achieve the scope of this project. To begin with we will have to create a dataset of YouTube videos and their metadata and comments, to ensure the dataset is of a manageable size we shall only analyse 1000 – 2000 videos as our dataset. To do this we will have to write a data gathering tool which shall query the YouTube Data API to get video metadata and comments. Once we have a database of video metadata and comments, we will have to create a training dataset of 200 – 500 videos by manually evaluating the videos. Once we have these two datasets, we can build our machine learning algorithm which uses the video metadata and comments to assign values to each video. While this AI section is being built, we can also create the web interface for this project. To do this we have to design the front-end, for the video search, recommendations and playing of videos (comments can also be shown). We will also have to create the various functionality for the front-end of the web interface including, sign up, sign in, playing video from YouTube’s servers, searching our video database, etc. As an extension on this project we can enable users to sign in with Google and use their Google account data to automatically determine what values a user has so that their search results and recommendations align with the values they want to view.