Context: Text from various journals classified as tool or non tool based on availability and whether publication is about a piece of software or not. Content: JSON file containing training data as key value pairs with text key having value of entire text of publication and class key is 0 (non tool) or 1 (tool). Acknowledgments: All data was obtained from Pubmed using its API and complied manually and via Python. Past Research: Experimenting with just using Abstract and Title, sentences with links/urls, full text etc. Inspiration: Trying to tackle the classification problem posted here https://www.kaggle.com/forums/f/15/kaggle-forum/t/23079/classification-of-scientific-publications-as-software-tools/133018#post133018 Steadily increasing the training set to add more publications.