Cuisine Discovery based on Recipe-Ingredient Network and Matrix Factorization

• Main Authors: Cheng-Jui Chang, 張政叡
• Other Authors: Yihuang Kang
• Format: Others
• Language: en_US
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/a32qmb

碩士 === 國立中山大學 === 資訊管理學系研究所 === 106 === This research proposes an approach to find the cuisines, the types of dishes, from the recipes, ingredients and methods of producing dishes. We believe that the cuisines can be distinguished by the culture, the ingredients, and the processing action of a dish. Therefore, we applied three methods, the nsNMF, the regularized nsNMF and network analysis to analyze recipe data. The nsNMF is mostly employed in the field of text mining and implemented the topic modeling, but we used it on the cuisine modeling throw the correlations between recipes and ingredients. On the other hand, another dimension of the cuisines− processing action, was introduced into the modeling to produce the nsNMF with constraint. The network analysis was implemented to process the relationships among ingredients. We employed an algorithm, which is greedy−community in network analysis, to detect how many clusters there was in the ingredients. Finally, we analogized what the difference are between the results of the matrix factorization and the network analysis.