• Journal of Korean Society of Transportation
• /
• v.23 no.2
• /
• pp.117-130
• /
• 2005

In both deteministic user Optimal Traffic Assignment Model (UOTAM) and stochastic UOTAM, travel time, which is a major ccriterion for traffic loading over transportation network, is defined by the sum of link travel time and turn delay at intersections. In this assignment method, drivers actual route perception processes and choice behaviors, which can become main explanatory factors, are not sufficiently considered: therefore may result in biased traffic loading. Even though there have been some efforts in Stochastic UOTAM for reflecting drivers' route perception cost by assuming cumulative distribution function of link travel time, it has not been fundamental fruitions, but some trials based on the unreasonable assumptions of Probit model of truncated travel time distribution function and Logit model of independency of inter-link congestion. The critical reason why deterministic UOTAM have not been able to reflect route perception cost is that the route perception cost has each different value according to each origin, destination, and path connection the origin and destination. Therefore in order to find the optimum route between OD pair, route enumeration problem that all routes connecting an OD pair must be compared is encountered, and it is the critical reason causing computational failure because uncountable number of path may be enumerated as the scale of transportation network become bigger. The purpose of this study is to propose a method to enable UOTAM to reflect route perception cost without route enumeration between an O-D pair. For this purpose, this study defines a link as a least definition of path. Thus since each link can be treated as a path, in two links searching process of the link label based optimum path algorithm, the route enumeration between OD pair can be reduced the scale of finding optimum path to all links. The computational burden of this method is no more than link label based optimum path algorithm. Each different perception cost is embedded as a quantitative value generated by comparing the sub-path from the origin to the searching link and the searched link.

• Journal of Korean Society of Disaster and Security
• /
• v.12 no.2
• /
• pp.73-82
• /
• 2019

Recently, as the occurrence frequency of sudden floods due to climate change increased, the flood damage on riverside social infrastructures was extended so that there has been a threat of overflow. Therefore, a rapid prediction of potential flooding in riverside social infrastructure is necessary for administrators. However, most current flood forecasting models including hydraulic model have limitations which are the high accuracy of numerical results but longer simulation time. To alleviate such limitation, data driven models using artificial neural network have been widely used. However, there is a limitation that the existing models can not consider the time-series parameters. In this study the water surface elevation of the Hangang River bridge was predicted using the NARX model considering the time-series parameter. And the results of the ANN and RNN models are compared with the NARX model to determine the suitability of NARX model. Using the 10-year hydrological data from 2009 to 2018, 70% of the hydrological data were used for learning and 15% was used for testing and evaluation respectively. As a result of predicting the water surface elevation after 3 hours from the Hangang River bridge in 2018, the ANN, RNN and NARX models for RMSE were 0.20 m, 0.11 m, and 0.09 m, respectively, and 0.12 m, 0.06 m, and 0.05 m for MAE, and 1.56 m, 0.55 m and 0.10 m for peak errors respectively. By analyzing the error of the prediction results considering the time-series parameters, the NARX model is most suitable for predicting water surface elevation. This is because the NARX model can learn the trend of the time series data and also can derive the accurate prediction value even in the high water surface elevation prediction by using the hyperbolic tangent and Rectified Linear Unit function as an activation function. However, the NARX model has a limit to generate a vanishing gradient as the sequence length becomes longer. In the future, the accuracy of the water surface elevation prediction will be examined by using the LSTM model.

• Asia-Pacific Journal of Business Venturing and Entrepreneurship
• /
• v.14 no.2
• /
• pp.47-59
• /
• 2019

Various studies have been carried out on the subject of entrepreneurship, which is required to create new businesses and organizations during the early process of startups based on innovative technologies and ideas. At the same time, the concept of organizational entrepreneurial orientation, which explains how to manage enterprises in the process of pioneering new products and markets, is drawing more and more attention for the purpose of continuously creating and maintaining a competitive edge of startups. This study focused on the relationship between entrepreneurial orientation and startup performance and the role of technological orientation and social capital. An empirical research was conducted on 144 different startup companies residing in startup supporting institutions. To evaluate the suitability of the research model, a PLS-based structural equation model was used. The research results are as follows: First, the entrepreneurial orientation of startups was found to have a positive effect on startup performance. Second, it was shown that entrepreneurial orientation had a positive effect on all three dimensions of social capital and technological orientation. Third, it has been shown that technological orientation and the cognitive dimension of social capital mediates the relationship between entrepreneurial orientation and startup performance. Through this, it was confirmed that entrepreneurial orientation directly affects startup performance, and it even influences the growth of startups by increasing technological superiority and social capital which is inherent in the network. Also, the research identified the need for additional research on the relationship between the strengthening of technological orientation and strategical orientation in startups. This study is expected to expand the discussion about social capital in the field of startup related research by affirming the role and importance of the cognitive system embedded in the network as well as the connectivity of networks, which has been already emphasized in previous startup related studies. Finally, the results of this study were reflected to present new practical implications.

• (The)Study of the Eastern Classic
• /
• no.70
• /
• pp.303-345
• /
• 2018

In this article, how the Bangmok(榜目) documents, which are essentially lists of successful passers for the civil competitive examination system of the $Chos{\breve{o}}n$ dynasty, when rendered into digitalized formats, could serve as source of information, which would not only lets us know the $Chos{\breve{o}}n$ individuals' social backgrounds and bloodlines but also enables us to understand the intricate nature that the Yangban network had, will be discussed. In digitalized humanity studies, the Bangmok materials, literally a list of leading elites of the $Chos{\breve{o}}n$ period, constitute a very interesting and important source of information. Based upon these materials, we can see how the society -as well as the Yangban community- was like. Currently, all data inside these Bangmok lists are rendered in XML(eXtensible Makrup Language) format and are being served through DBMS(Database Management System), so anyone who would want to examine the statistics could freely do so. Also, by connecting the data in these Bangmok materials with data from genealogy records, we could identify an individual's marital relationship, home town, and political affiliation, and therefore create a complex narrative that would be effective in describing that individual's life in particular. This is a graphic database, which shows-when Bangmok data is punched in-successful passers as individual nodes, and displays blood and marital relations in a very visible way. Clicking upon the nodes would provide you with access to all kinds of relationships formed among more than 90 thousand successful passers, and even the overall marital network, once the genealogical data is input. In Korea, since 2005 and through now, the task of digitalizing data from the Civil exam Bangmok(Mun-gwa Bangmok), Military exam Bangmok (Mu-gwa Bangmok), the "Sa-ma" Bangmok and "Jab-gwa" Bangmok materials, has been completed. They can be accessed through a website(http://people.aks.ac.kr/index.aks) which has information on numerous famous past Korean individuals. With this kind of source of information, we are now able to extract professional Jung-in figures from these lists. However, meaningful and practical studies using this data are yet to be announced. This article would like to remind everyone that this information should be used as a window through which we could see not only the lives of individuals, but also the society.

• Journal of Intelligence and Information Systems
• /
• v.28 no.2
• /
• pp.147-170
• /
• 2022

Although the number of patent which is one of the core outputs of technological innovation continues to increase, the number of low-value patents also hugely increased. Therefore, efficient evaluation of patents has become important. Estimation of patent lifespan which represents private value of a patent, has been studied for a long time, but in most cases it relied on a linear model. Even if machine learning methods were used, interpretation or explanation of the relationship between explanatory variables and patent lifespan was insufficient. In this study, patent lifespan (number of renewals) is predicted based on the idea that patent lifespan represents the value of the patent. For the research, 4,033,414 patents applied between 1996 and 2017 and finally granted were collected from USPTO (US Patent and Trademark Office). To predict the patent lifespan, we use variables that can reflect the characteristics of the patent, the patent owner's characteristics, and the inventor's characteristics. We build four different models (Ridge Regression, Random Forest, Feed Forward Neural Network, Gradient Boosting Models) and perform hyperparameter tuning through 5-fold Cross Validation. Then, the performance of the generated models are evaluated, and the relative importance of predictors is also presented. In addition, based on the Gradient Boosting Model which have excellent performance, Accumulated Local Effects Plot is presented to visualize the relationship between predictors and patent lifespan. Finally, we apply Kernal SHAP (SHapley Additive exPlanations) to present the evaluation reason of individual patents, and discuss applicability to the patent evaluation system. This study has academic significance in that it cumulatively contributes to the existing patent life estimation research and supplements the limitations of existing patent life estimation studies based on linearity. It is academically meaningful that this study contributes cumulatively to the existing studies which estimate patent lifespan, and that it supplements the limitations of linear models. Also, it is practically meaningful to suggest a method for deriving the evaluation basis for individual patent value and examine the applicability to patent evaluation systems.

• Journal of Intelligence and Information Systems
• /
• v.27 no.1
• /
• pp.83-102
• /
• 2021

The government recently announced various policies for developing big-data and artificial intelligence fields to provide a great opportunity to the public with respect to disclosure of high-quality data within public institutions. KSURE(Korea Trade Insurance Corporation) is a major public institution for financial policy in Korea, and thus the company is strongly committed to backing export companies with various systems. Nevertheless, there are still fewer cases of realized business model based on big-data analyses. In this situation, this paper aims to develop a new business model which can be applied to an ex-ante prediction for the likelihood of the insurance accident of credit guarantee. We utilize internal data from KSURE which supports export companies in Korea and apply machine learning models. Then, we conduct performance comparison among the predictive models including Logistic Regression, Random Forest, XGBoost, LightGBM, and DNN(Deep Neural Network). For decades, many researchers have tried to find better models which can help to predict bankruptcy since the ex-ante prediction is crucial for corporate managers, investors, creditors, and other stakeholders. The development of the prediction for financial distress or bankruptcy was originated from Smith(1930), Fitzpatrick(1932), or Merwin(1942). One of the most famous models is the Altman's Z-score model(Altman, 1968) which was based on the multiple discriminant analysis. This model is widely used in both research and practice by this time. The author suggests the score model that utilizes five key financial ratios to predict the probability of bankruptcy in the next two years. Ohlson(1980) introduces logit model to complement some limitations of previous models. Furthermore, Elmer and Borowski(1988) develop and examine a rule-based, automated system which conducts the financial analysis of savings and loans. Since the 1980s, researchers in Korea have started to examine analyses on the prediction of financial distress or bankruptcy. Kim(1987) analyzes financial ratios and develops the prediction model. Also, Han et al.(1995, 1996, 1997, 2003, 2005, 2006) construct the prediction model using various techniques including artificial neural network. Yang(1996) introduces multiple discriminant analysis and logit model. Besides, Kim and Kim(2001) utilize artificial neural network techniques for ex-ante prediction of insolvent enterprises. After that, many scholars have been trying to predict financial distress or bankruptcy more precisely based on diverse models such as Random Forest or SVM. One major distinction of our research from the previous research is that we focus on examining the predicted probability of default for each sample case, not only on investigating the classification accuracy of each model for the entire sample. Most predictive models in this paper show that the level of the accuracy of classification is about 70% based on the entire sample. To be specific, LightGBM model shows the highest accuracy of 71.1% and Logit model indicates the lowest accuracy of 69%. However, we confirm that there are open to multiple interpretations. In the context of the business, we have to put more emphasis on efforts to minimize type 2 error which causes more harmful operating losses for the guaranty company. Thus, we also compare the classification accuracy by splitting predicted probability of the default into ten equal intervals. When we examine the classification accuracy for each interval, Logit model has the highest accuracy of 100% for 0~10% of the predicted probability of the default, however, Logit model has a relatively lower accuracy of 61.5% for 90~100% of the predicted probability of the default. On the other hand, Random Forest, XGBoost, LightGBM, and DNN indicate more desirable results since they indicate a higher level of accuracy for both 0~10% and 90~100% of the predicted probability of the default but have a lower level of accuracy around 50% of the predicted probability of the default. When it comes to the distribution of samples for each predicted probability of the default, both LightGBM and XGBoost models have a relatively large number of samples for both 0~10% and 90~100% of the predicted probability of the default. Although Random Forest model has an advantage with regard to the perspective of classification accuracy with small number of cases, LightGBM or XGBoost could become a more desirable model since they classify large number of cases into the two extreme intervals of the predicted probability of the default, even allowing for their relatively low classification accuracy. Considering the importance of type 2 error and total prediction accuracy, XGBoost and DNN show superior performance. Next, Random Forest and LightGBM show good results, but logistic regression shows the worst performance. However, each predictive model has a comparative advantage in terms of various evaluation standards. For instance, Random Forest model shows almost 100% accuracy for samples which are expected to have a high level of the probability of default. Collectively, we can construct more comprehensive ensemble models which contain multiple classification machine learning models and conduct majority voting for maximizing its overall performance.

• Journal of Intelligence and Information Systems
• /
• v.17 no.4
• /
• pp.143-155
• /
• 2011

Lately electronic tagging policy for the sexual offenders was introduced in order to reduce and prevent sexual offences. However, most sexual offences against children happening these days are committed by the tagged offenders whose identities have been released. So, for the crime prevention, we need measures with which we could minimize the suffers more promptly and actively. This paper suggests a new system to relieve the sexual abuse related anxiety of the children and solve the problems that electronic bracelet has. Existing bracelets are only worn by serious criminals, and it's only for risk management and positioning, there is no way to protect the children who are the potential victims of sexual abuse and there actually happened some cases. So we suggest also letting the students(children) wear the LBS(Location Based Service) and USN(Ubiquitous Sensor Network) technology based electronic bracelets to monitor and figure out dangerous situations intelligently, so that we could prevent sexual offences against children beforehand, and while a crime is happening, we could judge the situation of the crime intelligently and take swift action to minimize the suffer. And by checking students' attendance and position, guardians could know where their children are in real time and could protect the children from not only sexual offences but also violent crimes against children like kidnapping. The overall system is like follows : RFID Tag for children monitors the approach of offenders. While an offender's RFID tag is approaching, it will transmit the situation and position as the first warning message to the control center and the guardians. When the offender is going far away, it turns to monitoring mode, and if the tag of the child or the offender is taken off or the child and offender stay at one position for 3~5 minutes or longer, then it will consider this as a dangerous situation, then transmit the emergency situations and position as the second warning message to the control center and the guardians, and ask for the dispatch of police to prevent the crime at the initial stage. The RFID module of criminals' electronic bracelets is RFID TAG, and the RFID module for the children is RFID receiver(reader), so wherever the offenders are, if an offender is at a place within 20m from a child, RFID module for children will transmit the situation every certain periods to the control center by the automatic response of the receiver. As for the positioning module, outdoors GPS or mobile communications module(CELL module)is used and UWB, WI-FI based module is used indoors. The sensor is set under the purpose of making it possible to measure the position coordinates even indoors, so that one could send his real time situation and position to the server of central control center. By using the RFID electronic roll book system of educational institutions and safety system installed at home, children's position and situation can be checked. When the child leaves for school, attendance can be checked through the electronic roll book, and when school is over the information is sent to the guardians. And using RFID access control turnstiles installed at the apartment or entrance of the house, the arrival of the children could be checked and the information is transmitted to the guardians. If the student is absent or didn't arrive at home, the information of the child is sent to the central control center from the electronic roll book or access control turnstiles, and look for the position of the child's electronic bracelet using GPS or mobile communications module, then send the information to the guardians and teacher so that they could report to the police immediately if necessary. Central management and control system is built under the purpose of monitoring dangerous situations and guardians' checking. It saves the warning and pattern data to figure out the areas with dangerous situation, and could help introduce crime prevention systems like CCTV with the highest priority. And by DB establishment personal data could be saved, the frequency of first and second warnings made, the terminal ID of the specific child and offender, warning made position, situation (like approaching, taken off of the electronic bracelet, same position for a certain time) and so on could be recorded, and the data is going to be used for preventing crimes. Even though we've already introduced electronic tagging to prevent recurrence of child sexual offences, but the crimes continuously occur. So I suggest this system to prevent crimes beforehand concerning the children's safety. If we make electronic bracelets easy to use and carry, and set the price reasonably so that many children can use, then lots of criminals could be prevented and we can protect the children easily. By preventing criminals before happening, it is going to be a helpful system for our safe life.

• Journal of Intelligence and Information Systems
• /
• v.23 no.2
• /
• pp.71-88
• /
• 2017

Language models were originally developed for speech recognition and language processing. Using a set of example sentences, a language model predicts the next word or character based on sequential input data. N-gram models have been widely used but this model cannot model the correlation between the input units efficiently since it is a probabilistic model which are based on the frequency of each unit in the training set. Recently, as the deep learning algorithm has been developed, a recurrent neural network (RNN) model and a long short-term memory (LSTM) model have been widely used for the neural language model (Ahn, 2016; Kim et al., 2016; Lee et al., 2016). These models can reflect dependency between the objects that are entered sequentially into the model (Gers and Schmidhuber, 2001; Mikolov et al., 2010; Sundermeyer et al., 2012). In order to learning the neural language model, texts need to be decomposed into words or morphemes. Since, however, a training set of sentences includes a huge number of words or morphemes in general, the size of dictionary is very large and so it increases model complexity. In addition, word-level or morpheme-level models are able to generate vocabularies only which are contained in the training set. Furthermore, with highly morphological languages such as Turkish, Hungarian, Russian, Finnish or Korean, morpheme analyzers have more chance to cause errors in decomposition process (Lankinen et al., 2016). Therefore, this paper proposes a phoneme-level language model for Korean language based on LSTM models. A phoneme such as a vowel or a consonant is the smallest unit that comprises Korean texts. We construct the language model using three or four LSTM layers. Each model was trained using Stochastic Gradient Algorithm and more advanced optimization algorithms such as Adagrad, RMSprop, Adadelta, Adam, Adamax, and Nadam. Simulation study was done with Old Testament texts using a deep learning package Keras based the Theano. After pre-processing the texts, the dataset included 74 of unique characters including vowels, consonants, and punctuation marks. Then we constructed an input vector with 20 consecutive characters and an output with a following 21st character. Finally, total 1,023,411 sets of input-output vectors were included in the dataset and we divided them into training, validation, testsets with proportion 70:15:15. All the simulation were conducted on a system equipped with an Intel Xeon CPU (16 cores) and a NVIDIA GeForce GTX 1080 GPU. We compared the loss function evaluated for the validation set, the perplexity evaluated for the test set, and the time to be taken for training each model. As a result, all the optimization algorithms but the stochastic gradient algorithm showed similar validation loss and perplexity, which are clearly superior to those of the stochastic gradient algorithm. The stochastic gradient algorithm took the longest time to be trained for both 3- and 4-LSTM models. On average, the 4-LSTM layer model took 69% longer training time than the 3-LSTM layer model. However, the validation loss and perplexity were not improved significantly or became even worse for specific conditions. On the other hand, when comparing the automatically generated sentences, the 4-LSTM layer model tended to generate the sentences which are closer to the natural language than the 3-LSTM model. Although there were slight differences in the completeness of the generated sentences between the models, the sentence generation performance was quite satisfactory in any simulation conditions: they generated only legitimate Korean letters and the use of postposition and the conjugation of verbs were almost perfect in the sense of grammar. The results of this study are expected to be widely used for the processing of Korean language in the field of language processing and speech recognition, which are the basis of artificial intelligence systems.

• Journal of Intelligence and Information Systems
• /
• v.24 no.2
• /
• pp.59-83
• /
• 2018

With the increasing importance of sentiment analysis to grasp the needs of customers and the public, various types of deep learning models have been actively applied to English texts. In the sentiment analysis of English texts by deep learning, natural language sentences included in training and test datasets are usually converted into sequences of word vectors before being entered into the deep learning models. In this case, word vectors generally refer to vector representations of words obtained through splitting a sentence by space characters. There are several ways to derive word vectors, one of which is Word2Vec used for producing the 300 dimensional Google word vectors from about 100 billion words of Google News data. They have been widely used in the studies of sentiment analysis of reviews from various fields such as restaurants, movies, laptops, cameras, etc. Unlike English, morpheme plays an essential role in sentiment analysis and sentence structure analysis in Korean, which is a typical agglutinative language with developed postpositions and endings. A morpheme can be defined as the smallest meaningful unit of a language, and a word consists of one or more morphemes. For example, for a word '예쁘고', the morphemes are '예쁘(= adjective)' and '고(=connective ending)'. Reflecting the significance of Korean morphemes, it seems reasonable to adopt the morphemes as a basic unit in Korean sentiment analysis. Therefore, in this study, we use 'morpheme vector' as an input to a deep learning model rather than 'word vector' which is mainly used in English text. The morpheme vector refers to a vector representation for the morpheme and can be derived by applying an existent word vector derivation mechanism to the sentences divided into constituent morphemes. By the way, here come some questions as follows. What is the desirable range of POS(Part-Of-Speech) tags when deriving morpheme vectors for improving the classification accuracy of a deep learning model? Is it proper to apply a typical word vector model which primarily relies on the form of words to Korean with a high homonym ratio? Will the text preprocessing such as correcting spelling or spacing errors affect the classification accuracy, especially when drawing morpheme vectors from Korean product reviews with a lot of grammatical mistakes and variations? We seek to find empirical answers to these fundamental issues, which may be encountered first when applying various deep learning models to Korean texts. As a starting point, we summarized these issues as three central research questions as follows. First, which is better effective, to use morpheme vectors from grammatically correct texts of other domain than the analysis target, or to use morpheme vectors from considerably ungrammatical texts of the same domain, as the initial input of a deep learning model? Second, what is an appropriate morpheme vector derivation method for Korean regarding the range of POS tags, homonym, text preprocessing, minimum frequency? Third, can we get a satisfactory level of classification accuracy when applying deep learning to Korean sentiment analysis? As an approach to these research questions, we generate various types of morpheme vectors reflecting the research questions and then compare the classification accuracy through a non-static CNN(Convolutional Neural Network) model taking in the morpheme vectors. As for training and test datasets, Naver Shopping's 17,260 cosmetics product reviews are used. To derive morpheme vectors, we use data from the same domain as the target one and data from other domain; Naver shopping's about 2 million cosmetics product reviews and 520,000 Naver News data arguably corresponding to Google's News data. The six primary sets of morpheme vectors constructed in this study differ in terms of the following three criteria. First, they come from two types of data source; Naver news of high grammatical correctness and Naver shopping's cosmetics product reviews of low grammatical correctness. Second, they are distinguished in the degree of data preprocessing, namely, only splitting sentences or up to additional spelling and spacing corrections after sentence separation. Third, they vary concerning the form of input fed into a word vector model; whether the morphemes themselves are entered into a word vector model or with their POS tags attached. The morpheme vectors further vary depending on the consideration range of POS tags, the minimum frequency of morphemes included, and the random initialization range. All morpheme vectors are derived through CBOW(Continuous Bag-Of-Words) model with the context window 5 and the vector dimension 300. It seems that utilizing the same domain text even with a lower degree of grammatical correctness, performing spelling and spacing corrections as well as sentence splitting, and incorporating morphemes of any POS tags including incomprehensible category lead to the better classification accuracy. The POS tag attachment, which is devised for the high proportion of homonyms in Korean, and the minimum frequency standard for the morpheme to be included seem not to have any definite influence on the classification accuracy.

• Journal of the Korean Geographical Society
• /
• v.29 no.2
• /
• pp.166-182
• /
• 1994

Kyonggi Province in the late Chosun dynasty was a center of superior government offices including 'Han' River water-road transportation and was located in the middle of an 'X'-shaped arterial road network. Because of these reasons, Kyonggi Province had a faster inflow of commodities, informations and technics compared with the other province. At this period of time, every local 'Eup' (name of administrative district) had not been affected by their above administrative districts and had their own autonomy. For this reason, every 'Eup' could be developed as a town, even if its size was small when it had sufficient internal growing conditions. Moreover, the markets ('Si-Jon') in big towns and periodical markets which were spread over the Kyonggi Province played role of commercial functions of town. And because military bases for the defence of the royal capital in Kyonggi Province also took parts of a non-agricultural city role, Xyonggi Provinc had much more possibilities of growing as a town rather than the other provinces. The towns of the late Chosun Dynasty were, except the capital and superior administrative districts which were governed by the 'You-Su', small towns which had only about 3, 000-5, 000 people. Most of the town dewellers were local officials, nobles, merchants, craftmen and slaves. And the farmers who lived near town became a pseudo-towner through suburb agriculture. Among these people, the merchants were leaders of townization. The downtowns were affected by the landform and traffic roads. The most fundamental function of towns were administrative. The opcial's grade, which was dispatched to the local administrative district ('Kun' or 'Hyun'), was decided by the size of population and agricultural land of each county. Large county which was governed by a high ranking opcial had more possibilities to develop as a large town. Because they supervised other opcials of lower rank and obtained more land and population for the town. The phonomena of farm abandonment after the Japanese Invasion of Korea in 1592-1598 stimulated the development of towns for commercial function. The commercial functions of towns were evident in the Si-Jon or Nan-Jon (names of markets) in the big cities such as Hansung and Kaesung, meanffwhile in the local areas it was emerged in the shape of periodical market networks as allied with near markets (which were called as Jang-Si) or permanent markets which were grown up from periodical markets. These facts of commercial development induced the birth of commercial town. Kyonggi Province showed the weak points of its defense system during both wars (Japanese Invasion in 1592 and Manchu's Invasion in 1636). The government reinforced its defense system by adding 4 'You-Su-Bus' and several military bases. Each local districts ('Eup'), where Geo-Jins were established, were stimulated to be a town while Jin-Kwan system were, adjusted and enforced. Among Dok-Jins(name of solitary military bases), Youngjongjin was grown up as a large garrison town which only played a role of defense. The number of towns that took roles of non-agricultural functions in Kyonggi Province was 52. Among these towns, 29 were developed as big towns which had above 3, 000 people and most of these towns were located on the northwest-southeast axes of 'X'-shaped arterial trafic network in the Chosn Dynasty, This fact points out that the traffic road is one of the important causes of the development of towns. When we make hierarchy of the towns of Kyonggi Province according to its population and how many functions it had, we can make it as 6 grades. The virst grade town 'Hansung' was the biggest central town of administration, commerce and defdnse. The 2nd grade town includes 'Kaesung' which had historical inertia that it had been the capital of the Koryo Dynesty. The 3rd grade towns include some 'You- Su-Bus' such as Soowon, Kanghwa, Kwangju and also include Mapo, Yongsan and from this we can imagine that the commercial development in the late Chosun Dynasty extremely affected the townization. The 4th-6th grade towns had smiliar population but it can be discriminated by how many town functions it had. So the 4th grade towns were the core of administration, commerce and defense function. 5th grade towns had administrative functions and one of commercial and defense functions. 6th grade towns had only one of these functions. When we research and town conditions of each grades as the ratio of non-agricultural population, we can find out that the towns from the 1st grade to 4th grade show difference by degree of townization but from the 4th grade to 6th grade towns do not show big difference in general.