This study investigates the feasibility of programming Internet of Things (IoT) kitchen devices using semantic parsing of cooking recipes. The researchers developed a unique annotated corpus to transform cooking instructions into machine-understandable commands. They utilized machine learning-based methods, specifically conditional random fields (CRF) and a neural network model, to parse the recipe instructions. While the results show it’s possible to train semantic parsers with the annotations, most natural-language instructions are incomplete, making the transformation into formal meaning representation challenging.
Publication date: 26 Sep 2023
Project Page: https://www.research.manchester.ac.uk/portal/en/publications/towards-enduser-development-for-iot-a-case-study-on-semantic-parsing-of-cooking-recipes-for-programming-kitchen-devices(0d9b0e7b-4ef5-4fa8-9a1c-7b0bb2c4a3a1).html
Paper: https://arxiv.org/pdf/2309.14165
