Hyuk-Chul Kwon

Using Cluster-Based Sampling to Select Initial Training Set for Active Learning in Text Classification

We propose a method of selecting initial training examples for active learning so that it can reach high performance faster with fewer further queries. Our method divides the unlabeled examples into clusters of similar ones and then selects from each cluster the most representative example which is the one closest to the cluster’s centroid. These representative examples are labeled by the user and become the members of the initial training set. We also promote inclusion of what we call model examples in the initial training set. Although the model examples which are in fact the centroids of the clusters are not real examples, their contribution to enhancement of classification accuracy is significant because they represent a group of similar examples so well. Experiments with various text data sets have shown that the active learner starting from the initial training set selected by our method reaches higher accuracy faster than that starting from randomly generated initial training set.

Discovery Architecture for the Tracing of Products in the EPCglobal Network

The EPCglobal Network is a network providing a shared view of the disposition of EPC-bearing objects between EPCglobal subscribers, within a relevant business context. In the EPCglobal Network, product data is distributed to several EPCISes(EPC Information Services) via movement of the product. The ONS (Object Naming Service) and the EPCISDS (EPCIS Discovery Service) are used to identify the distributed data for tracing the product. However, there is neither a standard scenario for access to the services when tracing the product nor a defined standard for EPCISDS. This paper suggests a novel architecture for tracing product data with EPCISDS and ONS. Our architecture provides a concrete substructure for product tracing services.

A scalable hybrid approach for extracting head components from Web tables

We have established a preprocessing method for determining the meaningfulness of a table to allow for information extraction from tables on the Internet. A table offers a preeminent clue in text mining because it contains meaningful data displayed in rows and columns. However, tables are used on the Internet for both knowledge structuring and document design. Therefore, we were interested in determining whether or not a table has meaningfulness that is related to the structural information provided at the abstraction level of the table head. Accordingly, we: 1) investigated the types of tables present in HTML documents, 2) established the features that distinguished meaningful tables from others, 3) constructed a training data set using the established features after having filtered any obvious decorative tables, and 4) constructed a classification model using a decision tree. Based on these features, we set up heuristics for table head extraction from meaningful tables, and obtained an F-measure of 95.0 percent in distinguishing meaningful tables from decorative tables and an accuracy of 82.1 percent in extracting the table head from the meaningful tables.

Cyber Forensics Ontology for Cyber Criminal Investigation

We developed Cyber Forensics Ontology for the criminal investigation in cyber space. Cyber crime is classified into cyber terror and general cyber crime, and those two classes are connected with each other. The investigation of cyber terror requires high technology, system environment and experts, and general cyber crime is connected with general crime by evidence from digital data and cyber space. Accordingly, it is difficult to determine relational crime types and collect evidence. Therefore, we considered the classifications of cyber crime, the collection of evidence in cyber space and the application of laws to cyber crime. In order to efficiently investigate cyber crime, it is necessary to integrate those concepts for each cyber crime-case. Thus, we constructed a cyber forensics domain ontology for criminal investigation in cyber space, according to the categories of cyber crime, laws, evidence and information of criminals. This ontology can be used in the process of investigating of cyber crime-cases, and for data mining of cyber crime; classification, clustering, association and detection of crime types, crime cases, evidences and criminals.

Extended Relief Algorithms in Instance-Based Feature Filtering

This paper presents extended Relief algorithms and their use in instance-based feature filtering for document feature selection. The Relief algorithms are general and successful feature estimators that detect conditional dependencies of features between instances, and are applied in the preprocessing step for document classification and regression. Since the introduction the Relief algorithm, many kinds of extended Relief algorithms have been suggested as solutions to problems of redundancy, irrelevant and noisy features as well as Relief algorithms limitations in two-class and multi-class datasets. In this paper, we introduce additional problems including the negative influence of computation similarities and weights caused by the small number of features in an instance, the absence of nearest Hits or nearest Misses for some instances using Relief algorithms, and other of problems. We suggest new extended Relief algorithms to solve those problems, having in the course of our research, and experimented on the estimation of the quality of features from instances, and classified datasets, and having compared the results of the new extended Relief algorithms. Indeed in the experimental results, the new extended Relief algorithms showed better performances for all of the datasets than did the Relief algorithms

Lyrics-Based Emotion Classification Using Feature Selection by Partial Syntactic Analysis

Songs feel emotionally different to listeners depending on their lyrical contents, even when melodies are similar. Accordingly, when using features related to melody, like tempo, rhythm, tune, and musical note, it is difficult to classify emotions accurately through the existing music emotion classification methods. This paper therefore proposes a method for lyrics-based emotion classification using feature selection by partial syntactic analysis. Based on the existing emotion ontology, four kinds of syntactic analysis rules were applied to extract emotion features from lyrics. The precision and recall rates of the emotion feature extraction were 73% and 70%, respectively. The extracted emotion features along with the NB, HMM, and SVM machine learning methods were used, showing a maximum accuracy rate of 58.8%.

Ontology-based Approach to Intelligent Ubiquitous Tourist Information System

This paper presents an ontology-based approach to an intelligent ubiquitous tourist information system. With the recent advances in Internet and ubiquitous technologies, there is increasing use of intelligent tourist information services via the web and mobile systems. Recently, those services have provided integrated heterogeneous travel information, recommended tourist attractions tailored to user profiles and travelers’ preferences, as well as services for pedestrian travelers using mobile systems. In parallel to these developments, many studies on the ontological approach to intelligent tourist information services have been introduced. However, there have been only a few studies undertaken from the perspective of both tour services and travelers’ preferences using ontology. Thus, in this paper, we propose a tourist domain ontology that consists of concepts for tourist contents and locations, and a tour service application ontology for various intelligent tour services. In addition, according to those ontologies, we designed an Ontology-based Intelligent Ubiquitous Tourist Information System (OiUTIS) for an interactive tourist information service tailored to both tour services and travelers in ubiquitous environments. Application Ontology; Touirst Information System; Ubiquitous Tourist System; Intelligent Tour Service

Intelligent integration of information on the Internet for travellers on demand

The paper focuses on the real-time integration of travel information from data that can be accessed on the Internet. In most cases, the necessary information does not exist in a form users really want. Information from various sources on the Internet needs to be integrated intelligently to produce the information in a form users demand. In the proposed system, Internet robots access homepages on the Internet, and the information is analyzed and integrated by an intelligent agent. To integrate the useful information, the profile of each user is needed, as is the situational information, such as where the user is located and what situation the user is in. The system provides a menu that can help the traveller to deal with difficulties and allows the user to click on one of the pre-defined menu options. The system infers the information the user needs from the user profile and situational information, and intelligently integrates information that the user might need. In the case of a traffic accident, for example, the system gives information such as the phone numbers of the nearest hospital, local insurance company, police and so on. The agent also can make automated phone calls on behalf of the traveller.

Semantic representation of Korean numeral classifier and its ontology building for HLT applications

The complexity of Korean numeral classifiers demands semantic as well as computational approaches that employ natural language processing (NLP) techniques. The classifier is a universal linguistic device, having the two functions of quantifying and classifying nouns in noun phrase constructions. Many linguistic studies have focused on the fact that numeral classifiers afford decisive clues to categorizing nouns. However, few studies have dealt with the semantic categorization of classifiers and their semantic relations to the nouns they quantify and categorize in building ontologies. In this article, we propose the semantic recategorization of the Korean numeral classifiers in the context of classifier ontology based on large corpora and KorLex Noun 1.5 (Korean wordnet; Korean Lexical Semantic Network), considering its high applicability in the NLP domain. In particular, the classifier can be effectively used to predict the semantic characteristics of nouns and to process them appropriately in NLP. The major challenge is to make such semantic classification and the attendant NLP techniques efficient. Accordingly, a Korean numeral classifier ontology (KorLexClas 1.0), including semantic hierarchies and relations to nouns, was constructed.

Stochastic Korean Word-Spacing with Smoothing Using Korean Spelling Checker

Word-spacing errors, one of the most frequent errors in Korean, produce ambiguities in the lexical interpretation of parts of speech or render sentences including them incomprehensible. Resolving those errors is thus crucial in Korean language processing application domains. In this paper, we propose a stochastic Korean word-spacing system with smoothing using Korean Spelling Checker, which is equally robust for both inner data and external data. In order to cope with various problems of word-spacing, this study (a) presents a simple stochastic word-spacing system with only two parameters using the odds favoring the inner-spacing of a given syllable bigram as well as relative word ftequencies, and (b) endeavors to (i) remove noise from the training data and (ii) diminish training data-dependency by dynamically creating a candidate word with a longest-radix-selecting algorithm. The system thus becomes robust against unseen words and offers a similar performance for both inner data and external data: it obtained a 98.47% and a 97.78% precision in word-unit correction for the inner test data and the balanced external test data, respectively.