This is a simple flask app that serves a KNN graph visualization of a given dataset. The visualization is done with D3.js library.
More info about force directed graph drawing: https://en.wikipedia.org/wiki/Force-directed_graph_drawing
You only need to install flask to be able to run my example.
pip3 install flask
After installing flask, and cloning the repository, cd inside the directory where app.py is and run app.py:
python3 app.py
if you face any problems please refer to flask documentation
After you run app.py you should see the following:
which means that the local server is running, now you should open the browser and open the local server with the specified port and see the following:
I've wrote a very handy script that doesn't make assumptions about your data, you have to provide it with a set of vectors, a set of labels for each vector, a set of titles for each vector, and it will populate the graphFile.json which the D3 library uses to draw the nodes and conections between the nodes in the graph.
in the above script, vecs.csv is a sample of vectors, their labels, and their associated titles of some news articles that i've prepared for this example.
then in the last line i call the function pipe which calculates the distances between the vectors, and outputs the graph connections to graphFile.json which will be used by d3 to render the graph.
pipe is a VERY general tool that doesn't make assumptions about your data, you only provide it a set of vectors and a set of labels, in my example i provided it with data from vecs.csv, but you can provide your own data and labels, and it will calculate the distances and output the result in graphFile.json, and the visulaization is ready to be served by the flask server.
- Mostafa Mahmoud