datasift / vedo-datascience-toolkit

Tools and scripts for working with VEDO

DataSift VEDO Data Science Toolkit

This repository contains useful tools that help you get the best out of DataSift VEDO. When working with the DataSift platform, VEDO allows you add custom metadata to unstructured data to make it much easier to process and action.

DataSift is the leading platform for giving unified, compliant access to data from social networks and news providers. To learn more, please visit our website http://datasift.com.

Linear Classifier Generator

The linear classifier library helps you generate machine learned classifiers to run on VEDO.

vedo-datascience-toolkit's Issues

Write CSDL to output file

Please change the script so that:

• Progress is written to standard out, giving the user an idea of progress
• The script takes in an argument for a file to write the CSDL to

New potential features

When I was classifying data I noticed that there was correlation between certain categories and the number of links or mentions in a given interaction.

It would be nice to get some features implemented that focus on the number of links or mentions present in an interaction if this is at all possible.

Make it more intuitive to add apriori words to config base

At the moment the apriori_features method contains an empty set. This set must contain a number of feature objects or the classifier script will fail.

My suggestion would either to make it more explicit that this set requires feature objects in the documentation/comments or if strings are passed in to the set to create them as PunctuatedWordFeature features with a default path of 'interaction.content'