• Tuberculosis and Respiratory Diseases
• /
• v.45 no.2
• /
• pp.322-332
• /
• 1998

Background: Accurate staging is important to determine treatment modalities and to predict prognosis for the patients with lung cancer. The simple two-stage system of the Veteran's Administration Lung Cancer study Group has been used for staging of small cell lung cancer(SCLC) because treatment usually consists of chemotherapy with or without radiotherapy. However, this system does not accurately reflect segregation of patients into homogenous prognostic groups. Therefore, a variety of new staging system have been proposed as more intensive treatments including either intensive radiotherapy or surgery enter clinical trials. We evaluate the prognostic importance of TNM staging, which has the advantage of providing a uniform detailed classification of tumor spread, in patients with SCLC. Methods: The medical records of 166 patients diagnosed with SCLC between January 1989 and December 1996 were reviewed retrospectively. The influence of TNM stage on survival was analyzed in 147 patients, among 166 patients, who had complete TNM staging data. Results: Three patients were classified in stage I / II, 15 in stage III a, 78 in stage IIIb and 48 in stage IV. Survival rate at 1 and 2 years for these patients were as follows: stage I / II, 75% and 37.5% ; stage IIIa, 46.7% and 25.0% ; stage III b, 34.3% and 11.3% ; and stage IV, 2.6% and 0%. The 2-year survival rates for 84 patients who received chemotherapy(more than 2 cycles) with or without radiotherapy were as follows: stage I / II, 37.5% ; stage rna, 31.3% ; stage IIIb 13.5% ; and stage IV 0%. Overall outcome according to TNM staging was significantly different whether or not received treatment. However, there was no significant difference between stage IIIa and stage IIIb though median survival and 2-year survival rate were higher in stage IIIa than stage IIIb. Conclusion: These results suggest that the TNM staging system may be helpful for predicting the prognosis of patients with SCLC.

• Journal of Intelligence and Information Systems
• /
• v.26 no.3
• /
• pp.149-169
• /
• 2020

"The Urban Renewal New Deal project", one of the government's major national projects, is about developing underdeveloped areas by investing 50 trillion won in 100 locations on the first year and 500 over the next four years. This project is drawing keen attention from the media and local governments. However, the project model which fails to reflect the original characteristics of the area as it divides project area into five categories: "Our Neighborhood Restoration, Housing Maintenance Support Type, General Neighborhood Type, Central Urban Type, and Economic Base Type," According to keywords for successful urban regeneration in Korea, "resident participation," "regional specialization," "ministerial cooperation" and "public-private cooperation", when local governments propose urban regeneration projects to the government, they can see that it is most important to accurately understand the characteristics of the city and push ahead with the projects in a way that suits the characteristics of the city with the help of local residents and private companies. In addition, considering the gentrification problem, which is one of the side effects of urban regeneration projects, it is important to select and implement urban regeneration types suitable for the characteristics of the area. In order to supplement the limitations of the 'Urban Regeneration New Deal Project' methodology, this study aims to propose a system that recommends urban regeneration types suitable for urban regeneration sites by utilizing various machine learning algorithms, referring to the urban regeneration types of the '2025 Seoul Metropolitan Government Urban Regeneration Strategy Plan' promoted based on regional characteristics. There are four types of urban regeneration in Seoul: "Low-use Low-Level Development, Abandonment, Deteriorated Housing, and Specialization of Historical and Cultural Resources" (Shon and Park, 2017). In order to identify regional characteristics, approximately 100,000 text data were collected for 22 regions where the project was carried out for a total of four types of urban regeneration. Using the collected data, we drew key keywords for each region according to the type of urban regeneration and conducted topic modeling to explore whether there were differences between types. As a result, it was confirmed that a number of topics related to real estate and economy appeared in old residential areas, and in the case of declining and underdeveloped areas, topics reflecting the characteristics of areas where industrial activities were active in the past appeared. In the case of the historical and cultural resource area, since it is an area that contains traces of the past, many keywords related to the government appeared. Therefore, it was possible to confirm political topics and cultural topics resulting from various events. Finally, in the case of low-use and under-developed areas, many topics on real estate and accessibility are emerging, so accessibility is good. It mainly had the characteristics of a region where development is planned or is likely to be developed. Furthermore, a model was implemented that proposes urban regeneration types tailored to regional characteristics for regions other than Seoul. Machine learning technology was used to implement the model, and training data and test data were randomly extracted at an 8:2 ratio and used. In order to compare the performance between various models, the input variables are set in two ways: Count Vector and TF-IDF Vector, and as Classifier, there are 5 types of SVM (Support Vector Machine), Decision Tree, Random Forest, Logistic Regression, and Gradient Boosting. By applying it, performance comparison for a total of 10 models was conducted. The model with the highest performance was the Gradient Boosting method using TF-IDF Vector input data, and the accuracy was 97%. Therefore, the recommendation system proposed in this study is expected to recommend urban regeneration types based on the regional characteristics of new business sites in the process of carrying out urban regeneration projects."

• Journal of Intelligence and Information Systems
• /
• v.25 no.1
• /
• pp.163-177
• /
• 2019

As smartphones are getting widely used, human activity recognition (HAR) tasks for recognizing personal activities of smartphone users with multimodal data have been actively studied recently. The research area is expanding from the recognition of the simple body movement of an individual user to the recognition of low-level behavior and high-level behavior. However, HAR tasks for recognizing interaction behavior with other people, such as whether the user is accompanying or communicating with someone else, have gotten less attention so far. And previous research for recognizing interaction behavior has usually depended on audio, Bluetooth, and Wi-Fi sensors, which are vulnerable to privacy issues and require much time to collect enough data. Whereas physical sensors including accelerometer, magnetic field and gyroscope sensors are less vulnerable to privacy issues and can collect a large amount of data within a short time. In this paper, a method for detecting accompanying status based on deep learning model by only using multimodal physical sensor data, such as an accelerometer, magnetic field and gyroscope, was proposed. The accompanying status was defined as a redefinition of a part of the user interaction behavior, including whether the user is accompanying with an acquaintance at a close distance and the user is actively communicating with the acquaintance. A framework based on convolutional neural networks (CNN) and long short-term memory (LSTM) recurrent networks for classifying accompanying and conversation was proposed. First, a data preprocessing method which consists of time synchronization of multimodal data from different physical sensors, data normalization and sequence data generation was introduced. We applied the nearest interpolation to synchronize the time of collected data from different sensors. Normalization was performed for each x, y, z axis value of the sensor data, and the sequence data was generated according to the sliding window method. Then, the sequence data became the input for CNN, where feature maps representing local dependencies of the original sequence are extracted. The CNN consisted of 3 convolutional layers and did not have a pooling layer to maintain the temporal information of the sequence data. Next, LSTM recurrent networks received the feature maps, learned long-term dependencies from them and extracted features. The LSTM recurrent networks consisted of two layers, each with 128 cells. Finally, the extracted features were used for classification by softmax classifier. The loss function of the model was cross entropy function and the weights of the model were randomly initialized on a normal distribution with an average of 0 and a standard deviation of 0.1. The model was trained using adaptive moment estimation (ADAM) optimization algorithm and the mini batch size was set to 128. We applied dropout to input values of the LSTM recurrent networks to prevent overfitting. The initial learning rate was set to 0.001, and it decreased exponentially by 0.99 at the end of each epoch training. An Android smartphone application was developed and released to collect data. We collected smartphone data for a total of 18 subjects. Using the data, the model classified accompanying and conversation by 98.74% and 98.83% accuracy each. Both the F1 score and accuracy of the model were higher than the F1 score and accuracy of the majority vote classifier, support vector machine, and deep recurrent neural network. In the future research, we will focus on more rigorous multimodal sensor data synchronization methods that minimize the time stamp differences. In addition, we will further study transfer learning method that enables transfer of trained models tailored to the training data to the evaluation data that follows a different distribution. It is expected that a model capable of exhibiting robust recognition performance against changes in data that is not considered in the model learning stage will be obtained.

• Journal of Sasang Constitutional Medicine
• /
• v.11 no.2
• /
• pp.119-150
• /
• 1999

In spite of recent remarkable recent development in both western and oriental medical sciences, there is still only a shallow understanding of individual differences for various prognoses of incurable diseases and immunopathy diseases. Nevertheless, the care, cure and prevention methods of Sasang Constitutional Medicine are broadly used as an effective treatment of incurable diseases like immunopathy diseases and stress-related diseases and diseases due to aging. In this sense, the establishment of classification norms is urgent and essential for the worldwide application of Sasang Constitutional Medicine(SCM). This study began with the confirmation process of whether Sasang Constitutional types exist in Americans. To accomodate for cultural differences, the distinguishing tool was readjusted so that Sasang Constitutional Types in Americans could be determined. Hence, the selected tool is the new QSCCII+, which is a newly revised English version of the QSCCII. QSCCII was made and standardized by Dept. of SCM in Kyung Hee Medical Center and Dr. Kim7). The evaluation methods of the old version were improved in the new QSCCII+ through necessary statistical manipulation. The original QSCCII was officially authorized by the Korean Society of Sasang Constitutional Medicine as the only computerized version of Sasang diagnostics. This study is the first attempt to design a new diagnostic tool for the classification of Sasang Constitutional types in North Americans with the revision of QSCCII. The subjects of this study were selected from the cooperative people among the students and staffs of the University of Bridgeport and the patients who visited the Clinic in the Health Science Center. This study takes for about 1 year from 1998. 8 to 1999. 8 The conclusions of the study can be summarized as follows: 1. Sasang constitutional types also exist in Americans. It can also naturally be inferred that Sasang Constitutional types exist in all human beings, for there are many different human races in America. 2. There are more So-Yang In's than any other types in American white people. This result confirms the hypothesis that there also exist Sasang Constitutional types in westerners. 3. The result of repetitive tests suggests that the new QSCCII+ is an effective diagnostic tool for westerners when we consider the constant diagnostic results of the QSCCII+. 4. Sasang Constitutional types exit in the sample group regardless of racial difference. 5. The question items that were not often checked by Americans need to be modified into more understandable expressions. 6. The standardization of diagnosis for Americans should be established by use of the QSCCII+ 7. It can be guessed that there are many Tae-yang In's among the 71 persons who could not be clearly classified by the QSCCII+. Due to the scarcity of Tae-yang-In in general, it is important to improve upon the discernability of the QSCC II+. 8. The results of the Sasang Constitutional distribution in North Americans are as follows: The percentage of So-yang In distribution in the sample group is 36.25%(87persons), that of Tae-eum In is 13.75%(33persons), and that of So-eum In is 20.41%(49persons).

• Journal of Intelligence and Information Systems
• /
• v.22 no.3
• /
• pp.143-163
• /
• 2016

The demographics of Internet users are the most basic and important sources for target marketing or personalized advertisements on the digital marketing channels which include email, mobile, and social media. However, it gradually has become difficult to collect the demographics of Internet users because their activities are anonymous in many cases. Although the marketing department is able to get the demographics using online or offline surveys, these approaches are very expensive, long processes, and likely to include false statements. Clickstream data is the recording an Internet user leaves behind while visiting websites. As the user clicks anywhere in the webpage, the activity is logged in semi-structured website log files. Such data allows us to see what pages users visited, how long they stayed there, how often they visited, when they usually visited, which site they prefer, what keywords they used to find the site, whether they purchased any, and so forth. For such a reason, some researchers tried to guess the demographics of Internet users by using their clickstream data. They derived various independent variables likely to be correlated to the demographics. The variables include search keyword, frequency and intensity for time, day and month, variety of websites visited, text information for web pages visited, etc. The demographic attributes to predict are also diverse according to the paper, and cover gender, age, job, location, income, education, marital status, presence of children. A variety of data mining methods, such as LSA, SVM, decision tree, neural network, logistic regression, and k-nearest neighbors, were used for prediction model building. However, this research has not yet identified which data mining method is appropriate to predict each demographic variable. Moreover, it is required to review independent variables studied so far and combine them as needed, and evaluate them for building the best prediction model. The objective of this study is to choose clickstream attributes mostly likely to be correlated to the demographics from the results of previous research, and then to identify which data mining method is fitting to predict each demographic attribute. Among the demographic attributes, this paper focus on predicting gender, age, marital status, residence, and job. And from the results of previous research, 64 clickstream attributes are applied to predict the demographic attributes. The overall process of predictive model building is compose of 4 steps. In the first step, we create user profiles which include 64 clickstream attributes and 5 demographic attributes. The second step performs the dimension reduction of clickstream variables to solve the curse of dimensionality and overfitting problem. We utilize three approaches which are based on decision tree, PCA, and cluster analysis. We build alternative predictive models for each demographic variable in the third step. SVM, neural network, and logistic regression are used for modeling. The last step evaluates the alternative models in view of model accuracy and selects the best model. For the experiments, we used clickstream data which represents 5 demographics and 16,962,705 online activities for 5,000 Internet users. IBM SPSS Modeler 17.0 was used for our prediction process, and the 5-fold cross validation was conducted to enhance the reliability of our experiments. As the experimental results, we can verify that there are a specific data mining method well-suited for each demographic variable. For example, age prediction is best performed when using the decision tree based dimension reduction and neural network whereas the prediction of gender and marital status is the most accurate by applying SVM without dimension reduction. We conclude that the online behaviors of the Internet users, captured from the clickstream data analysis, could be well used to predict their demographics, thereby being utilized to the digital marketing.

• Radiation Oncology Journal
• /
• v.19 no.2
• /
• pp.118-126
• /
• 2001

Purpose : In this study, it was investigated whether dose response relation existed or not in local radiotherapy for primary hepatocellular carcinoma. Materials and Methods : From January 1992 to March 2000, 158 patients were included in present study. Exclusion criteria included the presence of extrahepatic metastasis, liver cirrhosis of Child's class C, tumors occupying more than two thirds of the entire liver, and performance status on the ECOG scale of more than 3. Radiotherapy was given to the field including tumor with generous margin using 6, 10-MV X-ray. Mean tumor dose was $48.2{\pm}7.9\;Gy$ in daily 1.8 Gy fractions. Tumor response was based on diagnostic radiologic examinations such as CT scan, MR imaging, hepatic artery angiography at $4\~8$ weeks following completion of treatment. Statistical analysis was done to investigate the existence of dose response relationship of local radiotherapy when it was applied to the treatment of primary hepatocellular carcinoma. Results : An objective response was observed in 106 of 158 patients, giving a response rate of $67.1\%$. Statistical analysis revealed that total dose was the most significant factor in relation to tumor response when local radiotherapy was applied to the treatment of primary hepatocellular carcinoma. Only $29.2\%$ showed objective response in patients treated with dose less than 40 Gy, while $68.6\%\;and\;77.1\%$ showed major response in patients with $40\~50\;Gy$ and more than 50 Gy, respectively. Child-Pugh classification was significant factor in the development of ascites, overt radiation induced liver disease and gastroenteritis. Radiation dose was an important factor for development of radiation induced gastroduodenal ulcer. Conclusion : Present study showed the existence of dose response relationship in local radiotherapy for primary hepatocellular carcinoma. Only radiotherapy dose was a significant factor to predict the objective response. Further study is required to predict the maximal tolerance dose in consideration of liver function and non-irradiated liver volume.

• Korean Journal of Soil Science and Fertilizer
• /
• v.43 no.2
• /
• pp.217-223
• /
• 2010

This study was conducted to reclassify Dongsong series based on the second edition of Soil Taxonomy and to discuss the formation of Dongsong series distributed on the lava plain at Cheolweon in Korea. Morphological properties of typifying pedon of Dongsong series were investigated, and physico-chemical properties were analyzed according to Soil Survey Laboratory Methods Manual. The typifying pedon of Dongsong series has brown (7.5YR 4/2) silty clay loam Ap horizon (0-16 cm), brown (7.5YR 4/2) silty clay loam BA horizon (16-22 cm), brown (7.5YR 4/2) silty clay Bt1 horizon (22-50 cm), reddish brown (5YR 5/4) silty clay Bt2 horizon (50-92 cm), and brown (7.5YR 4/3) silty clay loam Bt3 horizon (92-120 cm). It occurs on lava plain derived from baslt materials. The typifying pedon has higher bulk density than 0.90 Mg $m^{-3}$. That can not be classified as Andisol. But it has an argillic horizon from a depth of 22 to more than 120 cm, and a base saturation (sum of cations) of less than 35% at 125 cm below the upper boundary of the argillic horizon. It can be classified as Ultisol, not as Andisol or Alfisol. It has aquic conditions for some time in normal years in one or more horizons within 50 cm of the mineral soil surface, redoximorphic features between a depth of 25 cm, and a depth of 40 cm from the mineral soil surface, and redox concentrations, and 50%or more redox depletions with chroma of 2 or less in the matrix within the upper 12.5 cm of the argillic horizon. Therefore it can be classified as Aquult. It has episaturation, and keys out as Epiaquult. It has 50% or more chroma of 3 or more in one or more horizons between a depth of 25 cm from the mineral soil surface, and a depth of 75 cm. It can be classified as Aeric Aquult. Dongsong series have 35%or more clay at the particle-size control section, and have mesic soil temperature regime. Therefore they can be classified as fine, mesic family of Aeric Epiaquults, not as fine, mesic family of Typic Epiaqualfs. The Quarternary volcanic activities occurred in Jeju Island, Ulrung Island, Baekryeong Island, Cheolweon area, and Mt. Paekdu et al. in the Korean Penninsula. Most of them belong to the central eruption type, but Cheolweon area may be of the fissure eruption type. Dongsong series occur on Cheolweon lava plains derived from basaltic materials. Most soils distributed in Jeju Island, and derived from mainly pyroclastics are developed as Andisols. But Dongsong series distributed in Cheolweon lava plains which have a relatively dry climate and derived from basaltic materials are developed as Ultisols.

• Radiation Oncology Journal
• /
• v.14 no.2
• /
• pp.95-103
• /
• 1996

Purpose : A retrospective analysis for patients with oropharyngeal carcinoma who were treated with radiation was performed to assess the results of treatment and patterns of failure, and to identify the factors that might influence survival. materials and methods : From March 1985 through June 1993, 53 patients with oropharyngeal carcinoma were treated with either radiation therapy alone or combination of neoadjuvant chemotherapy and radiation therapy at the Department of Radiation Oncology, Kyungpook National University Hospital. Patients' ages ranged from 31 to 73 years with a median age of 54 years. There were 47 men and 6 women, Forty-two Patients ($79.2\%$) had squamous cell carcinoma, 10 patients ($18.9\%$) had undifferentiated carcinoma and 1 patient ($19\%$) had adenoid cystic carcinoma. There were 2 patients with stage I, 12 patients with stage II, 12 Patients with stage III and 27 patients with stage IV. According to the TNM classification, patients were distributed as follows: T1 7, T2 28, T3 10, T4 7, TX 1, and N0 17, Nl 13, N2 21, N3 2. The primary tumor sites were tonsillar region in 36 patients ($67.9\%$), base of the tongue in 12 patients ($22.6\%$), and soft palate in 5 patients ($9.4\%$). Twenty-five patients were treated with radiation therapy alone and twenty-eight Patients were treated with one to three courses of chemotherapy followed by radiation therapy. Chemotherapeutic regimens used were either CF (cisplatin and 5-fluorouracil) or CVB (cisplatin, vincristine and bleomycin). Radiation therapy was delivered 180-200 cGy daily, five times a week using 6 MV X-ray with or without 8-10 MeV electron beams A tumor dose ranged from 4500 cGy to 7740 cGy with a median dose of 7100 cGy. The follow-up time ranged from 4 months to 99 months with a median of 21 months. Results : Thirty-seven patients ($69.8\%$) achieved a CR (complete response) and PR (partial response) in 16 patients ($30.2\%$) after radiation therapy. The overall survival rates were $47\%$ at 2 years and $42\%$ at 3 years, respectively. The median survival time was 23 months. Overall stage (p=0.02) and response to radiation therapy (p=0.004) were significant prognostic factors for overall survival. The 2-year disease-free survival rate was $45.5\%$. T-stage (p=0.03), N-stage (p=0.04) and overall stage (P=0.04) were significant prognostic factors for disease-free survival. Age, sex, histology, primary site of the tumor, radiation dose, combination of chemotherapy were not significantly associated with disease-free survival. Among evaluable 32 Patients with CR to radiation therapy, 12 patients were considered to have failed Among these, 8 patients failed locoregionally and 4 Patients failed distantly. Conclusion : T-stage, N-stage and overall stage were significant prognostic factors for disease-free survival in the treatment of oropharyngeal cancer Since locoregional failure was the predominant pattern of relapse, potential methods to improve locoregional control with radiation therapy should be attempted. More controlled clinical, trials should be completed before acceptance of chemotherapy as a part of treatment of oropharyngeal carcinoma.

• Tuberculosis and Respiratory Diseases
• /
• v.46 no.2
• /
• pp.204-214
• /
• 1999

Background: Resting pulmonary function tests(PFTs) are routinely used in the evaluation of pulmonary impairment/disability. But the significance of the cardiopulmonary exercise test(CPX) in the evaluation of pulmonary impairment is controvertible. Many experts believe that dyspnea, though a necessary part of the assessment, is not a reliable predictor of impairment. Nevertheless, oxygen requirements of an organism at rest are different from at activity or exercising, and a clear relationship between resting PFTs and exercise tolerance has not been established in patients with chronic pulmonary disease. As well, the relationship between resting PFTs and dyspnea is complex. To investigate the relationship of dyspnea, resting PFTs, and CPX, we evaluated the patients of stabilized chronic pulmonary disease with clinical dyspnea rating(baseline dyspnea index, BDI), resting PFTs, and CPX. Method: The 50 patients were divided into two groups: non-severe and severe group on basis of results of resting PFTs(by criteria of ATS), CPX(by criteria of ATS or Ortega), and dyspnea rating(by focal score of BDI). Groups were compared with respect to pulmonary function, indices of CPX, and dyspnea rating. Results: 1. According to the criteria of pulmonary impairment with resting PFTs, $VO_2$max, and focal score of BDI were significantly low in the severe group(p<0.01). According to the criteria of $VO_2$max(ml/kg/min) and $VO_2$max(%), the parameters of resting PFTs, except $FEV_1$ were not significantly different between non-severe and severe(p>0.05). According to focal score($FEV_1$(%), FVC(%), MW(%), $FEV_1/FVC$, and $VO_2$max were significantly lower in the severe group(p<0.01). However, in the more severe dyspneic group(focal score<5), only $VO_2$max(ml/kg/min) and $VO_2$max(%) were low(p<0.01). $FEV_1$(%) was correlated with $VO_2$max(%)(r=0.52;p<0.01), but not predictive of exercise performance. The focal score had the correlation with max WR(%) (r=0.55;p<0.01). Sensitivity and specificity analysis were utilized to compare the different criteria used to evaluate the severity of pulmonary impairment, revealed that the classification would be different according to the criteria used. And focal score for dyspnea showed similar sensitivity and specificity. Conclusion : According to these result, resting PFTs were not superior to rating of dyspnea in prediction of exercise performance in patients with chronic pulmonary diseases and less correlative with focal score for dyspnea than $VO_2$max and max WR. Therefore, if not contraindicated, CPX would be considered to evaluate the severity of pulmonary impairment in patients with chronic pulmonary diseases, including with severe resting PFTs. Current criteria used to evaluate the severity of impairment were insufficient in considering the degree of dyspnea, so new criteria, including the severity of dyspnea, may be necessary.

• Asia pacific journal of information systems
• /
• v.21 no.1
• /
• pp.103-122
• /
• 2011

Recently, numerous documents have been made available electronically. Internet search engines and digital libraries commonly return query results containing hundreds or even thousands of documents. In this situation, it is virtually impossible for users to examine complete documents to determine whether they might be useful for them. For this reason, some on-line documents are accompanied by a list of keywords specified by the authors in an effort to guide the users by facilitating the filtering process. In this way, a set of keywords is often considered a condensed version of the whole document and therefore plays an important role for document retrieval, Web page retrieval, document clustering, summarization, text mining, and so on. Since many academic journals ask the authors to provide a list of five or six keywords on the first page of an article, keywords are most familiar in the context of journal articles. However, many other types of documents could not benefit from the use of keywords, including Web pages, email messages, news reports, magazine articles, and business papers. Although the potential benefit is large, the implementation itself is the obstacle; manually assigning keywords to all documents is a daunting task, or even impractical in that it is extremely tedious and time-consuming requiring a certain level of domain knowledge. Therefore, it is highly desirable to automate the keyword generation process. There are mainly two approaches to achieving this aim: keyword assignment approach and keyword extraction approach. Both approaches use machine learning methods and require, for training purposes, a set of documents with keywords already attached. In the former approach, there is a given set of vocabulary, and the aim is to match them to the texts. In other words, the keywords assignment approach seeks to select the words from a controlled vocabulary that best describes a document. Although this approach is domain dependent and is not easy to transfer and expand, it can generate implicit keywords that do not appear in a document. On the other hand, in the latter approach, the aim is to extract keywords with respect to their relevance in the text without prior vocabulary. In this approach, automatic keyword generation is treated as a classification task, and keywords are commonly extracted based on supervised learning techniques. Thus, keyword extraction algorithms classify candidate keywords in a document into positive or negative examples. Several systems such as Extractor and Kea were developed using keyword extraction approach. Most indicative words in a document are selected as keywords for that document and as a result, keywords extraction is limited to terms that appear in the document. Therefore, keywords extraction cannot generate implicit keywords that are not included in a document. According to the experiment results of Turney, about 64% to 90% of keywords assigned by the authors can be found in the full text of an article. Inversely, it also means that 10% to 36% of the keywords assigned by the authors do not appear in the article, which cannot be generated through keyword extraction algorithms. Our preliminary experiment result also shows that 37% of keywords assigned by the authors are not included in the full text. This is the reason why we have decided to adopt the keyword assignment approach. In this paper, we propose a new approach for automatic keyword assignment namely IVSM(Inverse Vector Space Model). The model is based on a vector space model. which is a conventional information retrieval model that represents documents and queries by vectors in a multidimensional space. IVSM generates an appropriate keyword set for a specific document by measuring the distance between the document and the keyword sets. The keyword assignment process of IVSM is as follows: (1) calculating the vector length of each keyword set based on each keyword weight; (2) preprocessing and parsing a target document that does not have keywords; (3) calculating the vector length of the target document based on the term frequency; (4) measuring the cosine similarity between each keyword set and the target document; and (5) generating keywords that have high similarity scores. Two keyword generation systems were implemented applying IVSM: IVSM system for Web-based community service and stand-alone IVSM system. Firstly, the IVSM system is implemented in a community service for sharing knowledge and opinions on current trends such as fashion, movies, social problems, and health information. The stand-alone IVSM system is dedicated to generating keywords for academic papers, and, indeed, it has been tested through a number of academic papers including those published by the Korean Association of Shipping and Logistics, the Korea Research Academy of Distribution Information, the Korea Logistics Society, the Korea Logistics Research Association, and the Korea Port Economic Association. We measured the performance of IVSM by the number of matches between the IVSM-generated keywords and the author-assigned keywords. According to our experiment, the precisions of IVSM applied to Web-based community service and academic journals were 0.75 and 0.71, respectively. The performance of both systems is much better than that of baseline systems that generate keywords based on simple probability. Also, IVSM shows comparable performance to Extractor that is a representative system of keyword extraction approach developed by Turney. As electronic documents increase, we expect that IVSM proposed in this paper can be applied to many electronic documents in Web-based community and digital library.