This paper presents a new clustering analysis approach based on data samples with multiple labels. It especially deals with the case where each label has no antagonistic label and the absence of a label for a data does not necessarily imply that this data cannot have said label, e.g. the substances in mineral exploration, the keywords of the Web pages, . . . The proposed approach relies on two analyses that are conduced in a parallel way: cluster analysis and label analysis. The cluster analysis aims at selecting the most interesting or relevant clusters. The label analysis aims both at classifying the labels into specific categories such as implicit, explicit, noisy and novel and into more general embedding categories that are relevant and irrelevant. The proposed analysis methods are based on the use of two main informations: the similarity between the data given by the clustering algorithm and the distribution of the labels in the model after a projection of these labels on the classification model. Moreover, these methods make use of original quality measures for performing both labels and cluster analyses. An experimentation in the domain of documentary data highlights the accuracy of the proposed approach.
Database replication is a way to increase system performance and fault-tolerance of a given system. The price to pay is the effort needed to guarantee data consistency, and this is not an easy task. In this paper, we introduce a description of two 1-Copy-Serializable (1CS) eager update everywhere replication protocols. The preliminary results of their implementation in a middleware architecture are also presented. The advantage of these replication protocols is that they do not need to re-implement features that are provided by the underlying database. The first one does not rely on strong group communication primitives; distributed deadlock is avoided by a deadlock prevention schema based on transaction priorities (whose information is totally local at each node). The second one manages replica consistency by the total order message delivery featured by Group Communication Systems (GCSs).
Indexing time series data is an interesting problem that has attracted much interest in the research community for the last decade. Traditional indexing methods organize the data space using different metrics. However, searching high-dimensional spaces using a hierarchical index is not always efficient because a large portion of the index might need to be accessed during search. We have revisited this problem of matching subsequences in light of new technological advances. In particular, we have paid close attention to the increasing ratio of CPU to disk performance. We recognize this problem is heavily bound by IO operations and address this issue in a twofold manner. First, we propose the use of quantization to generate small and homogeneous representations of time series. Quantization provides tight upper- and lower-bounds on the measure of similarity to a query sequence. This allows us to drastically reduce the number of false alarms during search. Second, we organize the quantized representation of data in a linear array that can be efficiently read from disk. By reducing the number of false alarms and by sequentially reading the index, we are able to significantly reduce the IO cost of query processing. In consequence, we improve the overall search performance by up to a factor of 3 with respect to state of the art techniques for subsequence matching.
Recursive queries are required in many object-oriented database applications. Among them we can mention Bill-Of-Material (BOM), stock market dependencies, various kinds of networks (transportation, telecommunication, etc.), processing semi-structured data (XML, RDF), and so on. The support for recursive queries in current query languages is limited. In this paper we present recursive query processing capabilities for object-oriented environments. They are a part of the Stack-Based Query Language (SBQL). SBQL offers powerful and flexible recursive querying capabilities due to the fact that recursive processing operators are fully orthogonal to other capabilities of this language. This paper discusses novel SBQL constructs, such as transitive closures, fixed point equations and recursive procedures/views. They are implemented within the OODBMS ODRA devoted to Web and grid applications.
In general, there can be many reusable intermediate results due to the overlapped windows and periodic execution intervals among Multiple Continuous Queries (MCQ) on data streams. In this regard, we propose an efficient greedy algorithm for a global query plan construction, called GAGPC. GAGPC first decides an execution cycle and finds the maximal Set(s) of Related execution Points (SRP). Next, GAGPC constructs a global execution plan to make MCQ share common join-fragments with the highest benefit in each SRP. The algorithm suggests that the best plan of the same continuous queries may be different according to not only the existence of common expressions, but the size of overlapped windows related to them. It also reflects to reuse not only the whole but partial intermediate results unlike previous work. Finally, we show experimental results for the validation of GAGPC.
Data stored in databases constantly evolves as the application requirements change over time. Such evolution is a challenge, as existing applications must continue to work with the evolved data. XML is fast becoming the data model of choice for many applications as the XML data can be evolved easily and it supports heterogeneity. Various ways of storing XML data in tables or columns have been studied and explored in the past. However, they either do not address the evolving or heterogeneous nature of XML, or they do not take advantage of that in query optimization. UNION-ALL Views offer a logical layer to combine XML stored in various forms. This form of views - XML views - unlike relational views, offer logical structural partitioning of the XML data. In this paper we examine how such views can be used to solve the heterogeneity and evolving nature of XML and the unique properties of such views that allow better optimization.
Vertical partitioning is an effective way of improving performance in the database systems where a significant percentage of query processing time is spent on the full scans of relational tables. Vertical partitioning provides even more compelling performance gains when it is combined with the controlled replication of data in the environments where the processing of queries dominates the data manipulations. This paper proposes a new algorithm that finds a suboptimal vertical partitioning of relational tables under a constraint that certain level of redundancies is acceptable in a database. The algorithm is based on a new cost model, which precisely estimates I/O throughput as the total number of physical read/write database operations required to implement a given workload. The solution described in the paper transforms a schema of relational database into a partitioned one and decides which components of the original schema should be replicated as the separate partitions. The experiments conducted in this research and reported in the paper confirm the correctness of the cost model used by the vertical partitioning algorithm and demonstrate the expected performance gains from the partitioning.
In this paper, we propose an implementation method of an integrated associative search for mathematical expressions. A large amount of information resources for technology and science field have been distributed in wide area networks. In order to realize a media data search for technology and science field corresponding to semantic contents, realization of the associative search method for mathematical expressions is important. This method consists of a similarity expression search function and a semantic associative search function based on the mathematical model of meaning. The similarity expression search realizes a similarity retrieval by user's queries given in some mathematical expressions or symbols. The semantic associative search realizes an associative retrieval by user's queries given in some words for mathematical expressions characterized by metadata represented in words corresponding to their features. This method can realize a multifaceted associative search for mathematical expressions by user's queries which consist of words and mathematical expressions. Applications of this work include some retrieval systems for technical report, academic paper, and libraries for numerical analysis.
Object-oriented methods seem to handle well the complexity associated with developing spatio-temporal software systems. In recent years, a growing number of software systems are being developed within the object oriented paradigm. This paper presents the extension and integration of existing methods to provide an object oriented methodology that facilitates the development of spatio-temporal software systems. The Unified Modeling Language (UML) is extended with spatial and temporal semantics to allow the effective modeling of a spatio temporal database during the analysis phase of software development. ArcInfo UML, that is, the Unified Modeling Language extended by ArcInfo structures, is used to facilitate the modeling of a system during the design phase. ArcInfo is used as the implementation platform of the spatio-temporal database. A UML CASE tool (Microsoft Visio) will be used during analysis, design and ArcInfo database schema generation. Development will be lead by the Spatio-temporal UML process, that is, the Booch process extended to address spatial and spatiotemporal concerns.
In order to facilitate efficient query processing, the information contained in data warehouses is typically stored as a set of materialized views. Deciding which views to materialize represent a challenge in order to minimize view maintenance and query processing costs. Some existing approaches are applicable only for small problems, which are far from reality. In this paper we introduce a new approach for materialized view selection using Parallel Simulated Annealing (PSA) that selects views from an input Multiple View Processing Plan (MVPP). With PSA, we are able to perform view selection on MVPPs having hundreds of queries and thousands of views. Also, in our experimental study we show that our method provides a significant improvement in the quality of the obtained set of materialized views over existing heuristic and sequential simulated annealing algorithms.
Due to recent advances in sensor technology greater quantities of sensor data are being generated and circulated. In these circumstances, sensor data stream processing and management technologies have become important research areas. In the development of mechanical and electrical systems, sensor stream data are a potential medium for sharing information among the engineers who are engaged in the various phases in the system development and operation. This paper proposes an annotation method for a sensor data stream that links the information generated in the development and operational phases of a system. The key techniques of the proposed method are sensor pattern construction using hidden Markov models (HMMs) and an annotation method based on the HMMs constructed. We applied the proposed method to the sensor data stream of a supersmall artificial satellite and showed that the proposed method achieved approximately 95% annotation accuracy for long fragments of the sensor data stream.
MatBase is a Knowledge and Database Management System based on the Elementary Mathematical Data Model (EMDM). Its relational import subsystem provides users with the ability to incorporate databases from IBM DB/2, Oracle, or MS SQL Server and Access RDBMS. During import, besides copying table schemes and data, corresponding EMDM scheme and entity-relationship diagram (E-RD) are generated too and a design checking is also performed, including keys, references integrity, and existence constraints, as well as optional E-RD closed loops signalization. Checking is done according to MatBase higher relational standards: for example, every table should have an autonumber primary key, at least another (semantic) key and a not-null column, all keys should be minimally one-to-one, whereas all foreign keys should reference primary keys. Whenever this is possible, design flaws are automatically corrected; when not, they are only logged into MatBase system tables. The paper focuses on the two linear translation algorithms from the relational data model (RDM) to EMDM and ERDM that are reverse conceptual engineering tools reconstructing object-oriented mathematical semantics and visual structuring respectively from relational syntax. As an example, these algorithms are applied to a RDM domain-key normal form meta-catalog database.
In recent years, many non-expert user applications have been developed to query Geographic Information Systems (GIS) that are used to browse and view data about space and time thus naming them spatio-temporal databases. Our research aims at designing and developing an International Visual Query Language (IVQL) that allows users to query databases based on their location. The databases are installed on a GIS server computer. The queries are sent to the server from a mobile phone through the Short Messages System (SMS). With the emerging Globalization of user interfaces, IVQL is meant to have a global and international user interface that could be understood by all users worldwide who are from different countries with different cultures and languages. We propose a user interface consisting of smiley icons that are used to represent and build an international query language. Smiley icons enable the users to access data and build queries easily and in a user-friendly way. The visual query language, its user interface and its model are explained. The query formulation is illustrated using a sample GIS system for tourists. IVQL can be applied to fields such as Management Information Systems and tourism. We then conclude about our future work.
We present a scalable distributed database system called SD-SQL Server. Its original feature is dynamic and transparent repartitioning of growing tables, avoiding the cumbersome manual repartitioning that characterize current technology. SD-SQL Server re-partitions a table when an insert overflows existing segments. With the comfort of a single node SQL Server user, the SD-SQL Server user has larger tables or gets a faster response time through the dynamic parallelism. We present the architecture of our system, and its user/application interface.
Snapshot Isolation (SI) protocol is a database transaction processing algorithm used by some of commercial database systems to manage the concurrent executions of database transactions. SI protocol is a special case of multi-version algorithm. It avoids many of the anomalies typical for the concurrent processing of database transactions. Unfortunately, SI protocol does not guarantee correct serialization of database transactions under certain conditions. A recent work [3] proposed a formal solution, which characterizes the correctness of transactions running under SI protocol. However, the protocol is inefficient when it comes to processing long transactions. In this paper, we show that the limitations imposed on the structures of long transactions improve performance of SI protocol. A different way to characterize the serializability of schedule under SI dynamically is proposed and proved.
Recently, data exchange between different information sources has increased its importance, and many tools to help data exchange have been proposed. However, there has been no established method to evaluate the effectiveness of such tools. If you would like to evaluate the query execution performance of an RDBMS, we have various benchmarks, such as TPC-C, for that purpose. Similarly, it would be convenient if we had benchmarks for evaluating the effectiveness of data exchange tools. Development of such benchmarks is not trivial, because the factors to affect the effectiveness are completely different from those to affect the DBMS performance. Therefore, we cannot develop benchmarks using the same framework for the DBMS performance. This paper proposes a novel framework to develop benchmarks to evaluate the effectiveness of tools to help data exchange.
This paper addresses the problem of retrieving query results by fuzzy association. Many works have studied applications that associate multiple objects by "fuzzy" (or "soft") conditions. For example, in heterogeneous data integration, similar name constants referring to the same entity are joined by the fuzzy notion of similarity, as they can differ in subtle ways (e.g., "Jane Smith" and "J. Smith"). Similarly, in spatial queries, two objects are joined by the fuzzy measure of the distance between the two. However, these works are specific solely to the specific target scenarios. This paper thus models representative access methods and formalizes this problem of supporting fuzzy associations in general, to unify existing applications of varying scenarios and even unstudied applications. Further, we propose optimal algorithms that uniformly apply to varying access methods across applications.
This paper introduces a new method for gathering unstructured workflow definition data in order to model it in relational databases. In addition to introducing an open database model, a solution has been developed to facilitate the process of defining the workflow data for process designers. This solution consists of the Process Definition Language (PDL), pre-processing of unstructured workflow definition to be converted to PDL, and an interpreter to convert PDL lines to SQL statements. This solution is able to fill the basic database tables in order to drive a general workflow management system (WFMS). PDL is very close to natural language in order to be used by non-IT people, but structured enough to be interpreted.
