Informatics Publishing Limited

Yun-Jung Lee ¹, Su-Do Kim ², Jang-Pyo Hong ³, Hwan-Gue Cho ⁴, Seong-Min Yoon ⁴

Affiliations
1 Research Institute for Social Criticality, Pusan Nat’l University, Korea
2 School of Economics, Pukyong Nat’l University, Korea
3 Department of Computer Science, Pusan Nat’l University, Korea
4 Department of Economics, Pusan Nat’l University, Korea

Abstract

Background/Objectives: Inter-firm transaction data can be used as basic information to grasp the effectiveness and robustness of country’s economic activity. Methods/Statistical analysis: In this paper, we constructed and analyzed an inter-firm network and inter-industry network using inter-firm transaction data from KED. From the result of analysis, we found that the distribution of transaction frequency between companies tends to follow the power-law distribution. This is because a large number of companies have trading connections few firms such as a conglomerate, implying that most transactions are concentrated in the major companies. And we analyzed the transaction type of each industry using the E-I index. According to the result, the companies which belong to the service-related industries tend to trade with the companies in other industries. Otherwise, the companies which belong to the motor manufacture-related industry and the electronic-related industry do a lot of trade with the companies in same industries. Especially, the network structure of these industries is hierarchized as a tree structure in comparison with other industries. Findings: From the inter-industry transaction network, we also found that the industries in Korea are largely divided into two groups: domestic service industries and export manufacturing. These two sub-networks commonly form a tree structure representing the hierarchical flow of transactions where a transaction flows from leaf nodes to ischolar_main node in the inter-industry network. Application/Improvements: The General Construction industry is the ischolar_main node that is located at the top in the network of domestic service industries. And the electronic and computer-related industries are the ischolar_main nodes in the network of export manufacturing.

Keywords

Inter-firm transaction network, Inter-industry transaction network, Social network analysis, E-I index, Hierarchical network, Tree structure

Full Text

   

Su-Do Kim ¹, Yun-Jung Lee ¹, Hwan-Gue Cho ², Seong-Min Yoon ³

Affiliations
1 Research Institute for Social Criticality, Pusan Nat’l University, KR
2 Department of Computer Science, Pusan Nat’l University, KR
3 Department of Economics, Pusan Nat’l University, KR

Abstract

Background/Objective: New cuisines are being constantly created by people and shared on the internet. Cooking is usually classified and evaluated by subjective factors such as the personal cooking ability. Methods/Statistical Analysis: In this paper, we define the objective method for measuring the recipe complexity using the probabilistic entropy measurement. Through the cooking ingredients and cooking verbs in a recipe, we measure the recipe complexity in terms of the recipe preparation and the recipe procedure. And we calculate the recipe similarity by the entropy of common ingredients and construct the social network of recipes that observes the whole correlation between recipes. Findings: As a result, the most difficult recipe in the cooking preparation is ‘Braised Dongtae Seafood’ that needs 27 ingredients. And the easiest recipe is ‘Boiled Potato’ that only needs 3 ingredients: ‘Potato’, Salt’ and ‘Sugar’. In the cooking procedure, ‘Jeolla-Province Mosi-Songpyeon’ is the most difficult recipe that uses 18 verb types and 28 verbs. The easiest recipe is ‘Healthy Silk-Fowl Soup’ that only uses ‘Boil’ to cook. Through the recipe network, we can measure the distance of the recipes and minimize the effort in ingredient preparation and cooking procedure when we planed the cooking schedule to prepare for many dishes. Application/Improvements: If we plan to cook a schedule, the easy path to cook could be indicated by the shortest path between the recipes in the network.

Keywords

Entropy, Ingredient, Recipe Complexity, Recipe Network.

Full Text

   

Su-Do Kim ¹, Yun-Jung Lee ², Sang Hoon Kang ³, Hwan-Gue Cho ⁴, Seong-Min Yoon ⁵

Affiliations
1 Research Institute for Computer and Information Communication, Pusan National University, KR
2 Research Institute for Social Criticality, Pusan National University, KR
3 Department of Business Administration, Pusan National University, KR
4 Department of Computer Science, Pusan National University, KR
5 Department of Economics, Pusan National University, KR

Abstract

Food recipes, from traditional recipes to fusion recipes, are easily uploaded and shared online. Recipes consist of a set of ingredients, the cooking procedure, cooking time, etc. It is not easy to classify recipes in terms of the taste of cooked foods, the cuisine styles, or the characteristics of foods. In this paper, we construct the recipe similarity network by adding edges if two different recipes share common ingredients. For this, we newly define the similarity measure among recipes using the probabilistic entropy measures over ingredients. And we construct the ingredient relation network that shows the correlations of ingredients in the recipes. We show these networks can be applied to show the hierarchical structure of 683 recipes and 375 ingredients and the similar recipes are well clustered according to the entropy measure.

Keywords

Complex Network, Ingredient, Ingredient Entropy, Recipe Entropy, Recipe, Recipe Similarity.

Full Text