We need to upload the dataset somewhere publicly available so that others can contribute to extending the dataset as well as improving the classifier. It should be published in a way that respects the containing datasets that we used.