Ming-Chuan Hung

An efficient Fuzzy C-Means clustering algorithm

The Fuzzy C-Means (FCM) algorithm is commonly used for clustering. The performance of the FCM algorithm depends on the selection of the initial cluster center and/or the initial membership value. If a good initial cluster center that is close to the actual final cluster center can be found, the FCM algorithm will converge very quickly and the processing time can be drastically reduced. The authors propose a novel algorithm for efficient clustering. This algorithm is a modified FCM called the psFCM algorithm, which significantly reduces the computation time required to partition a dataset into desired clusters. We find the actual cluster center by using a simplified set of the original complete dataset. It refines the initial value of the FCM algorithm to speed up the convergence time. Our experiments show that the proposed psFCM algorithm is on average four times faster than the original FCM algorithm. We also demonstrate that the quality of the proposed psFCM algorithm is the same as the FCM algorithm.

Efficient approaches for materialized views selection in a data warehouse

View materialization is an effective method to increase query efficiency in a data warehouse and improve OLAP query performance. However, one encounters the problem of space insufficiency if all possible views are materialized in advance. Reducing query time by means of selecting a proper set of materialized views with a lower cost is crucial for efficient data warehousing. In addition, the costs of data warehouse creation, query, and maintenance have to be taken into account while views are materialized. In this paper, we propose efficient algorithms to select a proper set of materialized views, constrained by storage and cost considerations, to help speed up the entire data warehousing process. We derive a cost model for data warehouse query and maintenance as well as efficient view selection algorithms that effectively exploit the gain and loss metrics. The main contribution of our paper is to speed up the selection process of materialized views. Concurrently, this will greatly reduce the overall cost of data warehouse query and maintenance.

Efficient Utilization of Materialized Views in a Data Warehouse

View Materialization is an effective method to increase query efficiency in a data warehouse. However, one encounters the problem of space insufficiency if all possible views are materialized in advance. Reducing query time by means of selecting a proper set of materialized views with a lower cost is crucial for efficient data warehousing. In addition, the costs of data warehouse creation, query, and maintenance have to be taken into account while views are materialized. The purpose of this research is to select a proper set of materialized views under the storage and cost constraints and to help speedup the entire data warehousing process. We propose a cost model for data warehouse query and maintenance along with an efficient view selection algorithm, which uses the gain and loss indices. The main contribution of our paper is to speedup the selection process of materialized views. The second one is to reduce the total cost of data warehouse query and maintenance.

Data Migration from Grid to Cloud Computing

The advance of information and communication technologies over the last decade has improved the quality of healthcare and medical services tremendously. Especially, the people living in the countryside or remote areas benefit the most from telemedicine and emergency services. Our Health Grid is one of the test beds which provide various health-related Web services associated with mobile devices and physiological data acquisition and analysis instruments. As the number of new applications being developed increases rapidly, the ever-growing volume of collected data and real-time demand of analysis result have driven the architectural migration from Grid to Cloud Computing much sooner than we expected. Our challenge is to make the transition cost effective. This paper describes the data access migration from a relational database to Apaches HBase - one of the cloud databases. Our contribution is to minimize the required change of software for data access. Since the SQL commands of the relational database cannot be used in HBase, various mechanisms for translation and mapping between two sides must be developed. In addition, the services provided by the Web programs in Health Grid are written in various kinds of Web language while HBase does not support the access authority to these Web languages. To reduce the effort of modifying the source code for accessing HBase, we propose the use of Web services as the communication interface between various Web programs and necessary facilities to execute SQL commands in HBase. Although this is a hard engineering work, our results show that the proposed approaches are feasible and cost effective for the development teams at academic institutes. With this preliminary study, our next step is to improve our methods to take advantage of the efficient functions of HBase in processing the large amount of data.

An Intelligent Recommendation Model with a Case Study on u-Tour Taiwan of Historical Momuments and Cultural Heritage

Although there are many recommendation systems in use, they all face various challenges including the integration of diverse source data, improvement of prediction precision and meeting the user’s satisfaction. In order to increase success rate of satisfied recommendations as well as the applications in different domain fields, we propose an Intelligent Recommendation Model and conduct a case study on the historical monuments and cultural heritage of u-Tour Taiwan to show the feasibility of our model. In this research we use a hybrid approach to combine effective techniques such as popularization-based, community filtering, demographic profiling, and expertise-based in accordance with the type of users and the amount of available data to adjust weight values. We also use association rules of data mining technique to find potential patterns in the web access log, while clustering is used to assign users into different groups suitable for them. The incremental approach of our method can calculate the ranking value of content to be more precise. Finally, Adobe Flex is used to present the recommendation result of Taiwan’s 300 years of rich historical monuments and cultural heritage that provides more effective and efficient user interaction with less effort. Making full use of the valuable digital information of historical sites with our model, we hope to revitalize contemporary cultural and historical meaning that can bring people a brighter future and colorful life.

Fast and Effective Generation of Candidate-Sequences for Sequential Pattern Mining

The existing sequential pattern mining algorithms fall into two categories. One is the candidate-generation-and-test approach such as GSP, and the other is the pattern-growth approach such as PrefixSpan. Both GSP and PrefixSpan require setting the minimum support before their execution. We propose a new approach, called Fast and Effective Generation of Candidate-sequences (FEGC), to mine sequential patterns without predetermining the minimum support threshold. The main contribution is to scan all transactions in the database once and generate all the subsequences with their support counters. The experiments show that our algorithm performs well in various datasets.

A novel mining algorithm for periodic clustering sequential patterns

In knowledge discovery, data mining of time series data has many important applications. Especially, sequential patterns and periodic patterns, which evolved from the association rule, have been applied in many useful practices. This paper presents another useful concept, the periodic clustering sequential (PCS) pattern, which uses clustering to mine valuable information from temporal or serially ordered data in a period of time. For example, one can cluster patients according to symptoms of the illness under study, but this may just result in several clusters with specific symptoms for analyzing the distribution of patients. Adding time series analysis to the above investigation, we can examine the distribution of patients over the same or different seasons. For policymakers, the PCS pattern is more useful than traditional clustering result and provides a more effective support of decision-making.

Implementation of an Intelligent HLA-Compliant Application Layer Gateway for Real-Time Flight Simulation

This paper describes the implementation of an intelligent application layer gateway for the traditional training simulator to connect with other simulators using the high level architecture (HLA). The HLA is the latest open distributed interactive simulation standard developed by the defense modeling and simulation office of the US Department of Defense. To reap the full benefits of simulation interoperability and reusability in the near-term, it is important to quickly transit the legacy training simulators to compliant with the HLA. However, the programming philosophy between traditional training simulator and HLA-compliant simulator is different. To shorten the system rebuild time and avoid altering the original functions of the simulator, the proposed application layer gateway handles all the runtime infrastructure services within the HLA and communicates with the simulator by using the shared memory. To verify the feasibility of the proposed intelligent gateway and its performance, a real-time flight platform was developed. The platform includes a real-time flight model, a software control panel and an image generation system. The experimental results showed that our approach is workable and practical.