Recipe-to-Diagram Transformation System

Abstract

Automation graph transformation from natural language has been encountered for years in different research areas. As exemplified by Chinese and English recipes on the web, it is a very complicated for other system to retrieve information such as recommendation system and information filtering system. This project focuses on English and Chinese recipes to extract their linguistic feature’s difference. Nowadays, the Web consist the largest corpora of Chinese and English recipes. Most of them are stored in free format text in natural language. Moreover, examine the information within the text has been discussed for years in different domain. This project focuses on the recipe domain corpus and extracts meaningful information inside it. Also, a workflow like cooking procedure is proposed which is a generalized Recipe Graph Model. It uses a sequence of simple graphical images to represent the whole recipe. It supports simple Mathematical aggregation on the modeled set. This aims for the further information retrieval on the RGM for the recommendation system. More mining algorithm can be applied on the graph to extract information efficiently and precisely. It also supports the comparison between different recipes in different domain such as from ingredient or from constraints.