Analyzing Dynamical Activities of Co-occurrence Patterns for Cooking Ingredients

Due to the increasing popularity of cooking-recipe sharing sites and the success of complex network science, attention has recently been devoted to developing an effective network-based method of analyzing the characteristics of ingredient combinations used in recipes. Unlike previous approaches dealing with static properties, we aim at analyzing the dynamical changes in ingredient pairs jointly used in recipes, and propose an efficient method of extracting the change patterns for co-occurrence activities of ingredients. Based on the extracted change patterns, we build an active network among ingredients at every timestep, and identify active co-occurrence patterns. Moreover, we provide a method of interpreting active co-occurrence patterns in terms of recipes, and present a framework for visually analyzing their dynamical changes. Using real data from a Japanese recipe sharing site, we quantitatively demonstrate the effectiveness of the proposed method for extracting the activity change patterns for ingredient pairs, and uncover the characteristics of the seasonal changes in ingredient pairs jointly used in Japanese recipes by applying the proposed method.