Feature selection is an important method for improving the efficiency and accuracy of text categorization algorithms by removing redundant and irrelevant terms from the corpus.Extensive researches have been done to improve the performance of individual feature selection methods, but not much on their combinations.In this paper, we propose a method of combining multiple feature selection methods by using the combinatorial fusion analysis (CFA). A rank-score function and its graph, called rank-score graph,are adopted to measure the diversity of different feature selection methods.We have shown that a combination of multiple feature selection methods can outperform a single method only if each individual feature selection method has unique scoring behavior and relatively high performance. Moreover, it is shown that the rank-score function and rank-score graph are useful for the selection of a combination of feature selection methods.

In this paper we present an enhanced version of our NRF spectra classifier based on local-global graphs (LG-graphs) and provide a comparison with other possible alternative detection schemes. Experimental results verify our claims that the proposed nuclear resonance fluorescence (NRF) signature detection methodology is favorable over other widely employed conventional feature extraction and signal representation methods. The LG-graph methodology is based on the representation of the signal's peaks as triangle-like shapes and then on extracting significant geometrical features from these triangles to derive a concise representation of the peaks that correspond to specific NRF signatures of interest. These features are used to enable the matching of the materials of interest in a new unknown NRF spectrum.

This paper presents a privacy-preserving attribute distribution mechanism for an identity federation framework. Attribute-based access control (ABAC) is a highly flexible and scalable access control scheme which can deal with diverse security requirements in grid environment. However, in ABAC the user attributes delivered by the Identity Providers for authorization decisions may cause some privacy violation. We developed an attribute release control scheme that can improve the privacy protection capability of the current security systems in grids. The Shibboleth identity provider, GridShib, and Globus toolkit are used for our implementation.

XML has emerged as a de facto standard for encoding and sharing data between various applications. As the XML content grows, the ability to handle schemaless XML documents becomes more critical as most XML documents do not have schema or Document Type Definitions (DTDs). In addition, XML content and XML tools are often required to be combined in effective ways for better performance and higher flexibility. In this paper, we propose XML Integrated Environment (XIE) which is a general-purpose service-oriented architecture for processing XML documents in a scalable and efficient fashion. XIE supports a new software service model that provides a proper abstraction to describe a service and divide it into four components: structure, connection, interface and logic. We also propose and implement XIE Service Language (XIESL) that can capture the creation and maintenance of the XML processes and the data flow which is specified by the user and then orchestrate the interactions between different XIE services. XIESL manages the complexity of XML processing by implementing an XML processing pipeline that enables better management, control, interpretation and presentation of the XML data even for non-professional users.

In this paper, we propose a semantic-based access control method for grid data resources in the open grid services architecture - data access and integration (OGSA-DAI). The OGSA-DAI is an efficient grid-enabled middleware implementation of interfaces and services to access and control data resources (such as files, relational databases and XML databases). However, the identity-based access control in the OGSA-DAI causes substantial overhead for the resource providers in virtual organizations (VOs), because the access control information of individual users has to be maintained by each resource provider. In addition, access control policies need to be specified and managed across multiple VOs. To solve these problems, we propose the use of semantic-based access control policies in data grids. We use the Web ontology language (OWL) standard to represent the ontology of an organization's resources and users. Recently, eXtensible Access Control Markup Language (XACML) has been increasingly used for the representation of access control policies in grid environments. We propose the use of semantics in conjunction with the XACML standard for better interoperability and reduced administration overhead.

This paper presents an enhanced design of the community authorization service (CAS) which supports centralized, fine-grain access control by managing the memberships, service types, resource objects and security policies of a virtual organization (VO). The current CAS provides fundamental solutions regarding user privacy, authentication and authorization, but it has some limitations due to its centralized management of the security policies of a VO, in terms of scalability, flexibility and interoperability. We enhanced the CAS to support diverse security requirements within a dynamic grid computing environment by enabling the CAS server to publish a proxy certificate embedding additional attributes of users. It allows the service providers to support customized services by analyzing the attributes of users and security policies.

Feature selection is an important method for improving the efficiency and accuracy of text categorization algorithms by removing redundant and irrelevant terms from the corpus. In this paper, we propose a new supervised feature selection method, named CHIR, which is based on the chi2statistic and new statistical data that can measure the positive term-category dependency. We also propose a new text clustering algorithm, named text clustering with feature selection (TCFS). TCFS can incorporate CHIR to identify relevant features (i.e., terms) iteratively, and the clustering becomes a learning process. We compared TCFS and the K-means clustering algorithm in combination with different feature selection methods for various real data sets. Our experimental results show that TCFS with CHIR has better clustering accuracy in terms of the F-measure and the purity.

In this paper, we evaluated the efficacy of mined association rules between words for measuring the similarity between documents to enhance the text retrieval. In our experiments, for each document relevant to a query, we formed a group of documents having at least one common frequent set of words with the answer document. Then we measured the precision of the documents in the same group as an answer set to the corresponding query. This experiment was performed using a corpus of the Text retrieval conference (TREC) and search results. Our experimental results show that the frequent sets of words mined from our test database are useful in ranking query result sets to improve the precision of retrieval.

Work-centered software systems function as inherent work-aiding systems. Based on the design concept for a work-centered support system (WCSS), these software systems support user tasks and goals through both direct and indirect aiding methods within the interface client. In order to ensure the coherent development and delivery of work-centered software products, WCSS visual requirements must be specified to capture the cognitive aspects of the user interface design. A work-centered specification language based on the user interface markup language (UIML) is an effective solution to bridging this gap between cognitive systems engineering and software engineering. In this paper, we propose a new visual requirements specification language that can capture and describe work-centered visual requirements within a semi-formal syntax. The proposed language can also be easily integrated into a UML object model via the use of UML's extensibility features. A specification language for visual requirements could be employed by cognitive engineers and design teams to help convey requirements in a comprehensible format that is suitable for a software engineer. Such a solution provides coherency in the software modeling process of developing work-centered software systems

In this paper, we propose a role-based access control (RBAC) method for grid database services in open grid services architecture-data access and integration (OGSA-DAI). OGSA-DAI is an efficient grid-enabled middleware implementation of interfaces and services to access and control data sources and sinks. However, in OGSA-DAI, access control causes substantial administration overhead for resource providers in virtual organizations (VOs) because each of them has to manage a role-map file containing authorization information for individual grid users. To solve this problem, we used the community authorization service (CAS) provided by the globus toolkit to support the RBAC within the OGSA-DAI framework. The CAS grants the membership on VO roles to users. The resource providers then need to maintain only the mapping information from VO roles to local database roles in the role-map files, so that the number of entries in the role-map file is reduced dramatically. Furthermore, the resource providers control the granting of access privileges to the local roles. Thus, our access control method provides increased manageability for a large number of users and reduces day-to-day administration tasks of the resource providers, while they maintain the ultimate authority over their resources. Performance analysis shows that our method adds very little overhead to the existing security infrastructure of OGSA-DAI

In this paper, we propose a new text clustering algorithm, named Clustering based on Frequent Word Sequences (CFWS). A word sequence is frequent if it occurs in more than certain percentage of the documents in the text database. In the past, the vector space model was commonly used for information retrieval, but it treats documents as bags of words, ignoring the sequential pattern of word occurrences in the documents. However, the meaning of natural languages strongly depends on the word sequences, and the frequent word sequences can provide compact and valuable information about the text database. Bisecting k-means and FIHC algorithms are evaluated on the performance of text clustering, and are compared with the proposed CFWS algorithm. It has been shown that CFWS has much better performance.

In this paper, we propose a new algorithm, named MSPX, which mines maximal sequential patterns by using multiple samples to effectively exclude infrequent candidates. MSPX begins with a bottom-up search. But at each pass, instead of processing all candidates, it always tries to find most of the infrequent ones effectively by counting only the potentially infrequent candidates against the whole database. After removing verified infrequent candidates, the remaining candidates are used to generate new candidates. Finally, with a top-down search, all the maximal frequent sequences can be identified efficiently. Sampling technique is used at each pass to distinguish potentially infrequent candidates. How to increase the minimum support level for the mining of samples to estimate if candidates could be infrequent is analyzed theoretically. Due to the supersequence frequence based pruning, MSPX reduces much more search space than other algorithms. Unlike the traditional single-sample methods proposed for mining frequent itemsets, MSPX uses multiple samples. Thus, it can avoid or alleviate some problems inherent in the single-sample methods. Our experiments show MSPX has very good performance and better scalability than other algorithms. Moreover, even though MSPX uses sampling, the variance of its performance is very small in multiple runs for the same task.

In this paper, we propose a new algorithm, named Distributed Max-Miner (DMM), for mining maximal frequent itemsets from databases. A frequent itemset is maximal if none of its supersets is frequent. DMM requires very low communication and synchronization overhead in distributed computing systems. DMM has the local mining phase and the global mining phase. During the local mining phase, each node mines the local database to discover the local maximal frequent itemsets, then they form a set of maximal candidate itemsets for the top-down search in the subsequent global mining phase. A new prefix-tree data structure is developed to facilitate the storage and counting of the global candidate itemsets of different sizes. This global mining phase using the prefix-tree can work with any local mining algorithm. We implemented DMM on a cluster of workstations and evaluated its performance for various cases. DMM demonstrates better performance than other sequential and parallel algorithms, and its performance is quite scalable, even when there are large maximal frequent itemsets (i.e., long patterns) in databases.

We propose an efficient scalable algorithm for mining Maximal Sequential Patterns using Sampling (MSPS). The MSPS algorithm reduces much more search space than other algorithms because both the subsequence infrequency based pruning and the supersequence frequency based pruning are applied. In MSPS, sampling technique is used to identify long frequent sequences earlier, instead of enumerating all their subsequences. We propose how to adjust the user-specified minimum support level for mining a sample of the database to achieve better performance. This method makes sampling more efficient when the minimum support is small. A signature technique is utilized for the subsequence infrequency based pruning when the seed set of frequent sequences for the candidate generation is too big to be loaded into memory. A prefix tree structure is developed to count the candidate sequences of different sizes during the database scanning, and it also facilitates the customer sequence trimming. Our experiments showed MSPS has very good performance and better scalability than other algorithms.

Summary form only given. We propose a new algorithm named Parallel Multipass with Inverted Hashing and Pruning (PMIHP) for mining association rules between words in text databases. The characteristics of text databases are quite different from those of retail transaction databases, and existing mining algorithms cannot handle text databases efficiently because of the large number of itemsets (i.e., sets of words) that need to be counted. The new PMIHP algorithm is a parallel version of our multipass with inverted hashing and pruning (MIHP) algorithm, which was shown to be quite efficient than other existing algorithms in the context of mining text databases. The PMIHP algorithm reduces the overhead of communication between miners running on different processors because they are mining local databases asynchronously and prune the global candidates by using the inverted hashing and pruning technique.

In this paper, we propose a parallel algorithm for mining maximal frequent itemsets from databases. A frequent itemset is maximal if none of its supersets is frequent. The new parallel algorithm is named parallel max-miner (PMM), and it is a parallel version of the sequential max-miner algorithm by R.J. Bayardo (1998). Most of existing mining algorithms discover the frequent k-itemsets on the kth pass over the databases, and then generate the candidate (k + 1)-itemsets for the next pass. Compared to those level-wise algorithms, PMM looks ahead at each pass and prunes more candidate itemsets by checking the frequencies of their supersets. We implemented PMM on a cluster of workstations, and evaluated its performance for various cases. PMM demonstrated better performance than other sequential and parallel algorithms, and its performance is quite scalable, even when there are large maximal frequent itemsets (i.e. long patterns) in databases.

In this paper, we propose a new algorithm named multipass with inverted hashing and pruning (MIHP) for mining association rules between words in text databases. The characteristics of text databases are quite different from those of retail transaction databases, and existing mining algorithms cannot handle text databases efficiently because of the large number of itemsets (i.e., words) that need to be counted. Two well-known mining algorithms, the apriori algorithm and the direct hashing and pruning (DHP) algorithm, are evaluated in the context of mining text databases, and are compared with the proposed MIHP algorithm. It has been shown that the MIHP algorithm performs better for large text databases.

We present an adaptive version of the parallel distributive join (DJ) algorithm that we proposed in (Chung and Yang, 1996). The adaptive parallel DJ algorithm can handle the data skew in operand relations efficiently. We implemented the original and adaptive parallel DJ algorithms on a network of Alpha workstations using the Parallel Virtual Machine (PVM). We analyzed the performance of the algorithms, and compared it with that of the parallel Hybrid-Hash (HH) join algorithms. Our results show that the parallel DJ algorithms perform comparably with the parallel HH join algorithms over the entire range of the number of processors used and for different join selectivities. A significant advantage of the parallel DJ algorithms is that they can easily support non-equijoin operations.

In this paper, we propose two new algorithms for mining association rules between words in text databases. The characteristics of text databases are quite different from those of retail transaction databases, and existing mining algorithms cannot handle text databases efficiently because of the large number of itemsets (i.e., words) that need to be counted. Two well-known mining algorithms, Apriori algorithm and Direct Hashing and Pruning (DHP) algorithm, are evaluated in the context of mining text databases, and are compared with the new proposed algorithms named Multipass-Apriori (M-Apriori) and Multipass-DHP (M-DHP). It has been shown that the proposed algorithms have better performance for large text databases.

In this paper, we analyze the performance of the parallel distributive join algorithm that we proposed previously (1996). We implemented the algorithm on an Intel Paragon machine and analyzed the effect of the number of processors and the join selectivity on the performance of the algorithm. We also compared the performance of the distributed join (DJ) algorithm with that of the hybrid-hash (HH) join algorithm. Our results show that the DJ performs comparably with the HH over the entire range of number of processors used and different join selectivities. A big advantage of the parallel DJ algorithm over the RH join algorithm is that it can easily support non-equijoin operations. The results can also be used to estimate the performance of file I/O intensive applications to be implemented on the Intel Paragon machine.

We propose a method for integrating a preexisting conventional database system with a multimedia server in a multidatabase environment. In the multidatabase environment, changes to the preexisting database system are not allowed because such changes are too expensive. For the integration, high-level semantic description of multimedia data is modeled using the enhanced entity-relationship (EER) model to support content-based retrieval of multimedia data. The EER design is translated into a schema of the preexisting database system and then the translated schema is integrated with the preexisting database schema. The content description can be used to locate pertinent multimedia data, and the identifiers are used to access the multimedia data stored in the multimedia server. However, with only a simple schema representation of the semantic description of multimedia data, high levels of recall and precision of queries may not be obtained because conventional database systems provide only exact matching answers to the query. Thus, we extended the conventional query processing mechanism by providing a modified cooperative query answering mechanism.

This paper presents a parallel distributive join algorithm for cube-connected multiprocessors. The performance analysis shows that the proposed algorithm has an almost linear speedup over the sequential distributive join algorithm as the number of processors increases, and its performance is comparable to that of the parallel hybrid-hash join algorithm. A big advantage of the proposed algorithm over hash-based join algorithms is that it does not have the bucket overflow problem caused by nonuniform hashing of the smaller operand relation. Moreover, the proposed algorithm can easily support the nonequijoin operation, which is very hard to implement by using hash-based join algorithms.

We propose a multidatabase transaction management scheme using the flexible transaction model. In our approach, global subtransactions can be committed unilaterally at local sites before the global commit. Since such executions may not guarantee the traditional atomicity requirement of global transactions, a new correctness criterion is proposed. We also present a semantics-based concurrency control scheme and a commit protocol which ensure the proposed correctness criterion without requiring local DBMSs to provide prepare-to-commit operations.

In this paper, a chained-directory cache coherence protocol for a shared-memory multiprocessor sys tem is proposed, which uses linked lists to store the informa tion for the coherence maintenance. The proposed scheme uses small memory space and has high scalability, and it requires less memory accesses by distributing the coherence maintenance to caches. We evaluated the performance of the proposed scheme, and compared it with those of two other directory-based schemes; the singly-linked list and the fullmap schemes. The analysis shows that the proposed scheme provides better processor utilization and results in less memory contention.

Concatenated code word (CCW) surrogate files are useful as indexes for very large knowledge bases to support logic programming inference mechanisms because of their small size and simple maintenance requirements. A parallel back-end database machine is proposed to speed up relation operations based on the CCW surrogate files. The basic idea of the machine is to reduce the amount of fact data to be transferred from the secondary storage systems to satisfy a query by performing relational operations on the CCW surrogate files first. The database machine consists of a number of surrogate file processors (SFPs) and extensional database processors (EDBPs) operating in SIMD mode. The performance of the system for parallel relational operations is evaluated for various cases.>