Brain.Dump

A Sentiment Analysis tool made to translate journal entries into meaningful representations of the emotion within the entries.

By:

John Howes
Morgan Martin
Sam Oates
Luke Ritchie
Oscar Skean
Robert Wheaton

Components

Brain.Dump can be split into three major components: a journaling site, a reinforcement-learned algorithm that acts as an interpreter for analyzing the entries, and a drawing mechanism for turning the algorithm's output into a meaningful image.

The Journal

The Journal will be a website allowing you to write and save a journal entry, and then submit once you have finished writing it. Upon submission it will be sent to the interpreter to be analyzed. Emotionally charged words will be picked up by the Interpreter.

In future iterations of the project, users will be able to save multiple entries, and view their entries in a chronological order.

The Interpreter

The Interpreter is a reinforcement-learned sentiment analysis tool that will read the journal entries and return decimal values for different percentages of certain emotions, which will then be passed on to the Drawer.

The Drawer

The Drawer will be an image creation tool making use of parametric trig equations to generate a spirograph-like image. It can also make use of a multi-axis chart of percentages of different emotions for a less abstract representation.

Future Goals

We envision many future applications for such a device.

The machine learning tool could be reinforced by actual analyzation data from psychologists and psychiatrists, and be used to flag entries potentially indicative of mental illness
Users could view summaries of their journal entries via most used words, most present emotion, graphs of emotional changes, most common themes, etc.
This machine learning instance could be used on more than just braindumps, the braindump journaling platform could expand to blogs, message boards, forums, each of which would have icons indicative of the emotional content within.
There could be multiple styles of artful icons that users could choose between
The main site page could be an image of the summarized emotional content of all recent posts, to create a snapshot of the recent mindset of its users
Users of the site could search through posts and entries via tag words and view user opinions on the site via summarized emotional content of all entries relevant to the tag word(s).
Users could feed the ML instance pre-written content to be analyzed and represented