• Journal of Intelligence and Information Systems
• /
• v.17 no.1
• /
• pp.111-125
• /
• 2011

Clustering is a process of grouping similar or relevant documents into a cluster and assigning a meaningful concept to the cluster. By this process, clustering facilitates fast and correct search for the relevant documents by narrowing down the range of searching only to the collection of documents belonging to related clusters. For effective clustering, techniques are required for identifying similar documents and grouping them into a cluster, and discovering a concept that is most relevant to the cluster. One of the problems often appearing in this context is the detection of a complex concept that overlaps with several simple concepts at the same hierarchical level. Previous clustering methods were unable to identify and represent a complex concept that belongs to several different clusters at the same level in the concept hierarchy, and also could not validate the semantic hierarchical relationship between a complex concept and each of simple concepts. In order to solve these problems, this paper proposes a new clustering method that identifies and represents complex concepts efficiently. We developed the Hierarchical Overlapping Clustering (HOC) algorithm that modified the traditional Agglomerative Hierarchical Clustering algorithm to allow overlapped clusters at the same level in the concept hierarchy. The HOC algorithm represents the clustering result not by a tree but by a lattice to detect complex concepts. We developed a system that employs the HOC algorithm to carry out the goal of complex concept detection. This system operates in three phases; 1) the preprocessing of documents, 2) the clustering using the HOC algorithm, and 3) the validation of semantic hierarchical relationships among the concepts in the lattice obtained as a result of clustering. The preprocessing phase represents the documents as x-y coordinate values in a 2-dimensional space by considering the weights of terms appearing in the documents. First, it goes through some refinement process by applying stopwords removal and stemming to extract index terms. Then, each index term is assigned a TF-IDF weight value and the x-y coordinate value for each document is determined by combining the TF-IDF values of the terms in it. The clustering phase uses the HOC algorithm in which the similarity between the documents is calculated by applying the Euclidean distance method. Initially, a cluster is generated for each document by grouping those documents that are closest to it. Then, the distance between any two clusters is measured, grouping the closest clusters as a new cluster. This process is repeated until the root cluster is generated. In the validation phase, the feature selection method is applied to validate the appropriateness of the cluster concepts built by the HOC algorithm to see if they have meaningful hierarchical relationships. Feature selection is a method of extracting key features from a document by identifying and assigning weight values to important and representative terms in the document. In order to correctly select key features, a method is needed to determine how each term contributes to the class of the document. Among several methods achieving this goal, this paper adopted the $x^2$�� statistics, which measures the dependency degree of a term t to a class c, and represents the relationship between t and c by a numerical value. To demonstrate the effectiveness of the HOC algorithm, a series of performance evaluation is carried out by using a well-known Reuter-21578 news collection. The result of performance evaluation showed that the HOC algorithm greatly contributes to detecting and producing complex concepts by generating the concept hierarchy in a lattice structure.

• Journal of Korean Society for Quality Management
• /
• v.24 no.1
• /
• pp.126-136
• /
• 1996

In multi-item production system, GT(Group Technology) is used effectively in order to cluster various parts into groups. GT is based on clustering parts which have similar features, and these features are classified into two properties, namely crisp(quantitative) feature and fuzzy(qualitative) feature. Especially, many difficult problems are often faced that have to evaluate the properties of parts with the crisp and fuzzy feature together. As the basis of determining the similarity of inter-parts, in this method, one aggregate value is calculated on each part. However, because the above aggregate value is only gained from simple additive weighted sum, there is one problem in this method that has been handled the combination effect of inter-parts. For these reasons, in this paper, a proposed method is suggested for representing combination effect in order to cluster parts that have crisp and fuzzy properties into groups using ${\lambda}$-fuzzy measure and fuzzy integral.

• Proceedings of the IEEK Conference
• /
• 2002.06c
• /
• pp.99-102
• /
• 2002

K-means clustering algorithm is a very popular clustering technique, which is used in the field of information retrieval. In this paper, We deal with the problem of K-means Algorithm from the view of creating the centroids and suggest a method reflecting document feature and considering the context of each document to determine the new centroids during the process of forming new centroids. For experiment, We used the automatic document summarizer to summarize the Reuter21578 newslire test dataset and achieved 20% improved results to the recall metrics.

• Proceedings of the Korea Contents Association Conference
• /
• 2019.05a
• /
• pp.179-180
• /
• 2019

Nowadays, clustering algorithm is considered as a promising solution for lacking human-labeled and massive data of social media sites in numerous machine learning tasks. Many researchers propose disaster event detection systems have ability to determine special local events, such as missing people, public transport damage by clustering similar tweets and hashtags together. In this paper, we try to extend tweet hashtag feature definition by applying word embedding. The experimental results are described that word embedding achieve better performance than the reference method.

• ETRI Journal
• /
• v.37 no.2
• /
• pp.283-294
• /
• 2015

A model that precisely forecasts how much wind power is generated is critical for making decisions on power generation and infrastructure updates. Existing studies have estimated wind power from wind speed using forecasting models such as ANFIS, SMO, k-NN, and ANN. This study applies a projected clustering technique to identify wind power patterns of wind turbines; profiles the resulting characteristics; and defines hourly and daily power patterns using wind power data collected over a year-long period. A wind power pattern prediction stage uses a time interval feature that is essential for producing representative patterns through a projected clustering technique along with the existing temperature and wind direction from the classifier input. During this stage, this feature is applied to the wind speed, which is the most significant input of a forecasting model. As the test results show, nine hourly power patterns and seven daily power patterns are produced with respect to the Korean wind turbines used in this study. As a result of forecasting the hourly and daily power patterns using the temperature, wind direction, and time interval features for the wind speed, the ANFIS and SMO models show an excellent performance.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.12 no.4
• /
• pp.263-269
• /
• 2012

The ability to characterize feature objects from radar readings is often limited by simply looking at their still frame reflectivity, differential reflectivity and differential phase data. In many cases, time-series study of these objects' reflectivity profile is required to properly characterize features objects of interest. This paper introduces a novel technique to automatically track multiple 3D radar structures in C,S-band in real-time using Doppler radar and profile their characteristic reflectivity distribution in time series. The extraction of reflectivity profile from different radar cluster structures is done in three stages: 1. static frame (zone-linkage) clustering, 2. dynamic frame (evolution-linkage) clustering and 3. characterization of clusters through time series profile of reflectivity distribution. The two clustering schemes proposed here are applied on composite multi-layers CAPPI (Constant Altitude Plan Position Indicator) radar data which covers altitude range of 0.25 to 10 km and an area spanning over hundreds of thousands $km^2$. Discrete numerical simulations show the validity of the proposed technique and that fast and accurate profiling of time series reflectivity distribution for deformable 3D radar structures is achievable.

• Journal of KIISE:Databases
• /
• v.28 no.4
• /
• pp.655-664
• /
• 2001

The continuous data such as video streams and voice analog signals can be modeled as multidimensional data sequences(MDS's) in the feature space, In this paper, we investigate the clustering technique for multidimensional data sequence, Each sequence is represented by a small number by hyper rectangular clusters for subsequent storage and similarity search processing. We present a linear clustering algorithm that guarantees a predefined level of clustering quality and show its effectiveness via experiments on various video data sets.

• ETRI Journal
• /
• v.36 no.3
• /
• pp.333-342
• /
• 2014

Most hyper-ellipsoidal clustering (HEC) approaches use the Mahalanobis distance as a distance metric. It has been proven that HEC, under this condition, cannot be realized since the cost function of partitional clustering is a constant. We demonstrate that HEC with a modified Gaussian kernel metric can be interpreted as a problem of finding condensed ellipsoidal clusters (with respect to the volumes and densities of the clusters) and propose a practical HEC algorithm that is able to efficiently handle clusters that are ellipsoidal in shape and that are of different size and density. We then try to refine the HEC algorithm by utilizing ellipsoids defined on the kernel feature space to deal with more complex-shaped clusters. The proposed methods lead to a significant improvement in the clustering results over K-means algorithm, fuzzy C-means algorithm, GMM-EM algorithm, and HEC algorithm based on minimum-volume ellipsoids using Mahalanobis distance.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.8 no.2
• /
• pp.116-120
• /
• 2008

Clustering is a method for unsupervised learning. General clustering tools have been depended on statistical methods and machine learning algorithms. One of the popular clustering algorithms based on machine learning is the self organizing map(SOM). SOM is a neural networks model for clustering. SOM and extended SOM have been used in diverse classification and clustering fields such as data mining. But, SOM has had a problem determining optimal number of clusters. In this paper, we propose an improvement of SOM using gap statistic and probability distribution. The gap statistic was introduced to estimate the number of clusters in a dataset. We use gap statistic for settling the problem of SOM. Also, in our research, weights of feature nodes are updated by probability distribution. After complete updating according to prior and posterior distributions, the weights of SOM have probability distributions for optima clustering. To verify improved performance of our work, we make experiments compared with other learning algorithms using simulation data sets.

• Journal of KIISE:Software and Applications
• /
• v.35 no.9
• /
• pp.572-587
• /
• 2008

The Self-organizing Feature Map(SOFM) that is one of unsupervised neural networks is a very powerful tool for data clustering and visualization in high-dimensional data sets. Although the SOFM has been applied in many engineering problems, it needs to cluster similar weights into one class on the trained SOFM as a post-processing, which is manually performed in many cases. The traditional clustering algorithms, such as t-means, on the trained SOFM however do not yield satisfactory results, especially when clusters have arbitrary shapes. This paper proposes automatic clustering on trained SOFM, which can deal with arbitrary cluster shapes and be globally optimized by graph cuts. When using the graph cuts, the graph must have two additional vertices, called terminals, and weights between the terminals and vertices of the graph are generally set based on data manually obtained by users. The Proposed method automatically sets the weights based on mode-seeking on a distance matrix. Experimental results demonstrated the effectiveness of the proposed method in texture segmentation. In the experimental results, the proposed method improved precision rates compared with previous traditional clustering algorithm, as the method can deal with arbitrary cluster shapes based on the graph-theoretic clustering.