Braindump is a wiki system but with a little more semantic power, exploring the possibilities stemming from using triples - "facts" with a subject, predicate, and object, to store metadata.

For example, a Wikipedia article on John Doe might say somewhere: "John is the wife of Jane Doe". As someone reads this, a web of entities and relationships (a "graph" in semantic web speak) starts to take shape in their mind: there are two people, John Doe and Jane Doe, and John's wife is Jane. This is all fine, but if you're on a research mission and search Wikipedia for "whose wife is Jane Doe", you get a crazy jumble of results. This is because Wikipedia's search algorithm doesn't draw conclusions about entities and relationships from text the way humans do when they read, it simply returns pages containing the specified words.

Braindump moves forward by storing relationships between entities as triples. To express John and Jane's relationship, for example, you would make a page called "John Doe" and type "wife: Jane Doe". This text, in addition to being displayed, is parsed and stored as a triple, with a subject of "John Doe", a predicate of "wife" and an object of "Jane Doe".

Let's go back to the search, now - instead of crudely finding articles with matching words, Braindump can give you the subject of all triples with an object of "Jane Doe" and a predicate of "wife" - in this case, "John Doe".

For the foreseeable future, Braindump will be a sandbox, a walled garden in which to explore the possibilities stemming from data structured in this more computer-understandable way. To learn more about how these ideas are being applied to the web as a whole, read more at the W3C.

Braindump is written in Python has two main parts: GraphStore, a module that takes care of storing all the data, and an MVC web app, built on CherryPy, Jinja, and GraphStore, which allows you to add and edit data like a wiki.

Check out Install for installation instructions & system requirements.