• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.39 no.3
• /
• pp.90-99
• /
• 2016

Forecasting of box office performance after a film release is very important, from the viewpoint of increase profitability by reducing the production cost and the marketing cost. Analysis of psychological factors such as word-of-mouth and expert assessment is essential, but hard to perform due to the difficulties of data collection. Information technology such as web crawling and text mining can help to overcome this situation. For effective text mining, categorization of objects is required. In this perspective, the objective of this study is to provide a framework for classifying films according to their characteristics. Data including psychological factors are collected from Web sites using the web crawling. A clustering analysis is conducted to classify films and a series of one-way ANOVA analysis are conducted to statistically verify the differences of characteristics among groups. The result of the cluster analysis based on the review and revenues shows that the films can be categorized into four distinct groups and the differences of characteristics are statistically significant. The first group is high sales of the box office and the number of clicks on reviews is higher than other groups. The characteristic of the second group is similar with the 1st group, while the length of review is longer and the box office sales are not good. The third group's audiences prefer to documentaries and animations and the number of comments and interests are significantly lower than other groups. The last group prefer to criminal, thriller and suspense genre. Correspondence analysis is also conducted to match the groups and intrinsic characteristics of films such as genre, movie rating and nation.

• Journal of the Korean Society for information Management
• /
• v.35 no.2
• /
• pp.63-88
• /
• 2018

This research evaluated differences of classification performance for feature selection methods using LDA topic model and Doc2Vec which is based on word embedding using deep learning, feature corpus sizes and classification algorithms. In addition to find the feature corpus with high performance of classification, an experiment was conducted using feature corpus was composed differently according to the location of the document and by adjusting the size of the feature corpus. Conclusionally, in the experiments using deep learning evaluate training frequency and specifically considered information for context inference. This study constructed biomedical document dataset, Disease-35083 which consisted biomedical scholarly documents provided by PMC and categorized by the disease category. Throughout the study this research verifies which type and size of feature corpus produces the highest performance and, also suggests some feature corpus which carry an extensibility to specific feature by displaying efficiency during the training time. Additionally, this research compares the differences between deep learning and existing method and suggests an appropriate method by classification environment.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.14 no.7
• /
• pp.895-901
• /
• 2004

This paper describes a two-phase intelligent method for filtering spam mail based on textual information and hyperlinks. Scince the body of spam mail has little text information, it provides insufficient hints to distinguish spam mails from legitimate mails. To resolve this problem, we follows hyperlinks contained in the email body, fetches contents of a remote webpage, and extracts hints (i.e., features) from original email body and fetched webpages. We divided hints into two kinds of information: definite information (sender`s information and definite spam keyword lists) and less definite textual information (words or phrases, and particular features of email). In filtering spam mails, definite information is used first, and then less definite textual information is applied. In our experiment, the method of fetching web pages achieved an improvement of F-measure by 9.4% over the method of using on original email header and body only.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.18 no.10
• /
• pp.215-226
• /
• 2017

This study analyzed trends in agricultural and food marketing studies from 1984 to 2015 using text-mining techniques. Text-mining is a part of Big-data analysis, which is an effective tool to objectively process large amounts of information based on categorization and trend analysis. In the present study, frequency analysis, topic analysis and association rules were conducted. Titles of agricultural and food marketing studies in four journals and reports were used for placing the analysis. The results showed that 1,126 total theses related to agricultural and food marketing could be categorized into six subjects. There were significant changes in research trends before and after the 2000s. While research before 2000s focused on farm and wholesale level marketing, research after the 2000s mainly covered consumption, (processed)food, exports and imports. Local food and school meals are new subjects that are increasingly being studied. Issues regarding agricultural supply and demand were the only subjects investigated in policy research studies. Interest in agricultural supply and demand was lost after the 2000s. A number of studies after the 2010s analyzed consumption, primarily consumption trends and consumer behavior.

• Journal of Information Technology Applications and Management
• /
• v.21 no.4_spc
• /
• pp.415-429
• /
• 2014

With the advent of smart devices, users are able to connect to each other through the Internet without the constraints of time and space. Because the Internet has become increasingly important to users in their everyday lives, reliance on it has grown. As a result, the number of web sites constantly increases and the competition between these sites becomes more intense. Even those sites that operate successfully struggle to establish new strategies for customer retention and customer development in order to survive. Many companies use various customer information in order to establish marketing strategies based on customer group segmentation A method commonly used to determine the customer groups of individual sites is to infer customer characteristics based on the customers' demographic information. However, such information cannot sufficiently represent the real characteristics of customers. For example, users who have similar demographic characteristics could nonetheless have different interests and, therefore, different buying needs. Hence, in this study, customers' interests are first identified through an analysis of their Internet news inquiry records. This information is then integrated in order to identify each web category. The study then analyzes the possibilities for the practical use of the proposed methodology through its application to actual Internet news inquiry records and web site browsing histories.

• Journal of the Korean Society for information Management
• /
• v.23 no.1 s.59
• /
• pp.279-299
• /
• 2006

This study utilizes various approaches of machine learning in the process of automatically assigning descriptors to journal articles. The effectiveness of feature selection and the size of training set were examined, after selecting core journals in the field of information science and organizing test collection from the articles of the past 11 years. Regarding feature selection, after reducing the feature set using $x^2$ statistics(CHI) and criteria that prefer high-frequency features(COS, GSS, JAC), the trained Support Vector Machines(SVM) performed the best. With respect to the size of the training set, it significantly influenced the performance of Support Vector Machines(SVM) and Voted Perceptron(VTP). However, it had little effect on Naive Bayes(NB).

• Journal of the Korean Society for information Management
• /
• v.27 no.3
• /
• pp.265-281
• /
• 2010

The proposed method in this study is the Maximal Concept-Strength Recognition Method(MCR). In case that we don't know which database is the most suitable for automatic-classification when new database is imported, MCR method can support to select the most similar database among many databases in the legacy system. For experiments, we constructed four heterogeneous scholarly databases and measured the best performance with MCR method. In result, we retrieved the exact database expected and the precision value of MCR based automatic-classification was close to the best performance.

• Journal of the Korean Society for information Management
• /
• v.29 no.2
• /
• pp.225-246
• /
• 2012

This study purported to investigate the possibility of automatic descriptor assignment using the reclassification of author keywords in domestic scholarly databases. In the first stage, we selected optimal classifiers and parameters for the reclassification by comparing the characteristics of machine learning classifiers. In the next stage, learning the author keywords that were assigned to the selected articles on readings, the author keywords were automatically added to another set of relevant articles. We examined whether the author keyword reclassifications had the effect of vocabulary control just as descriptors collocate the documents on the same topic. The results showed the author keyword reclassification had the capability of the automatic descriptor assignment.

• Journal of Korean Library and Information Science Society
• /
• v.34 no.2
• /
• pp.233-253
• /
• 2003

Knowledge organization is the way we humans understand the world. There are two types of information organization mechanisms studied in information retrieval: namely classification md clustering. Classification organizes entities by pigeonholing them into predefined categories, whereas clustering organizes information by grouping similar or related entities together. The system of the Internet information resources extracts a keyword from the words which appear in the web document and draws up a reverse file. Term clustering based on grouping related terms, however, did not prove overly successful and was mostly abandoned in cases of documents used different languages each other or door-way-pages composed of only an anchor text. This study examines infometric analysis and clustering possibility of web documents based on co-link topology of web pages.

• Journal of KIISE
• /
• v.44 no.11
• /
• pp.1156-1164
• /
• 2017

Existing studies on automatic bug triage were mostly used the method of designing the prediction system based on the machine learning algorithm. Therefore, it can be said that applying a high-performance machine learning model is the core of the performance of the automatic bug triage system. In the related research, machine learning models that have high performance are mainly used, such as SVM and Naïve Bayes. In this paper, we apply Deep Learning, which has recently shown good performance in the field of machine learning, to automatic bug triage and evaluate its performance. Experimental results show that the Deep Learning based Bug Triage system achieves 48% accuracy in active developer experiments, un improvement of up to 69% over than conventional machine learning techniques.