vasia / gelly-school Goto Github PK

ignore...

Describe Gelly's neighborhood methods and how to use them for computing neighborhood aggregations.
As an example, we can maybe implement a naive version of clustering coefficient or triangle counting.

The first tutorial should explain the graph creation methods in Gelly. We can use a publicly available graph dataset (e.g. SNAP, Konnect) to read a file of edges, create, and initialize a graph in Gelly.
Then, we show how to run a library algorithm on the graph, e.g. SSSP or PageRank, using the run() method.

Explain the vertex-centric model and how this is implemented in Gelly.
As an example we can use the weighted PageRank that @samk3211 has already implemented.

Clicking on http://gellyschool.com takes you to a very spammy page

We should add some highlighting to the code boxes. e.g. if you look into the gelly documentation, keywords and gelly methods have different colors.

The initial page should briefly explain what is Gelly and state the purpose of the website. It should have a link to the Gelly guide, Flink installation instructions a list of the available tutorials.

Here we should introduce

• joinWithVertices, joinWithEdgesOnSource, joinWithEdgesOnTarget
• union, difference

Movie "Se7en" is misspelled as "Se7ev"

A simple one can check whether a graph contains duplicate edges or if all edge weights are positive.

Explain the GAS model and how it's implemented in Gelly. We can use the same example as in #9 to make the comparison between the 2 models.

This will be a larger example that uses both Gelly and the DataSet API for pre- and post-processing, similar to the MusicProfiles examples.
I will try to come up with a nice use-case!

After having created a graph with some public data, we will use some of the Gelly methods to get some metrics and statistics out of it.
For example:

• number of vertices, edges
• node degrees

This one will explain how to transform the graph into another graph, using Gelly's transformation methods.
Some ideas:

• reverse edge directions
• change vertex values type (with mapVertices)
• filter out all edges with a certain value (with filterOnEdges)