Baseline code for Word Salad: Relating Food Prices and Descriptions, Chahuneau V., Gimpel K., Routledge B.R., Scherlis L., and Smith N.A., EMNLP 2012
Depends on scikit-learn. Ridge regression is used because L1-regularized linear regression is slow.