Marlene Goncalves

SQLf3: an extension of SQLf with SQL3 features

Fuzzy Logic has been proposed for information representation and manipulation in databases, being the current tendency to study mechanisms of flexible querying to traditional databases. SQLf has been one of the efforts in this tendency. On the other hand, there are many different dialects of SQL. There are two major standards: ANSI (American National Standards Institute) SQL and an updated standard adopted in 1992, called SQL-92 or SQL2. There is also an emerging standard called SQL3 that extends SQL2 with many new features such as recursion, triggers, and objects. Then, there are versions of SQL produced by the principal vendors of database management systems. This paper looks for to continue the contribution in the area flexible querying to relational databases. Basically this contribution embraces the definition of the flexible querying language SQLf3 according to those characteristics contemplated in the norms SQL3 that can be susceptible to a fuzzy treatment.

Top-k skyline: a unified approach

The WWW has become a huge repository of information. For almost any knowledge domain there may exist thousands of available sources and billions of data instances. Many of these sources may publish irrelevant data. User-preference approaches have been defined to retrieve relevant data based on similarity, relevance or preference criteria specified by the user. Although many declarative languages can express user-preferences, considering this information during query optimization and evaluation remains as open problem. SQLf, Top-k and Skyline are three extensions of SQL to specify user-preferences. The first two filter irrelevant answers following a score-based paradigm. On the other hand, the latter produces relevant non-dominated answers using an order-based paradigm. The main objective of our work is to propose a unified approach that combines paradigms based on order and score. We propose physical operators for SQLf considering Skyline and Top-k features. Properties of those will be considered during query optimization and evaluation. We describe a Hybrid-Naive operator for producing only answers in the Pareto Curve with best score values. We have conducted initial experimental studies to compare the Hybrid operator, Skyline and SQLf.

SQLf flexible querying language extension by means of the norm SQL2

Traditional information storage and recovery systems are based on classic logic and set theory, limiting the user when he wants to express preferences. Fuzzy logic has been proposed for information representation and manipulation in databases, being the current tendency to study mechanisms of flexible querying to traditional databases. SQLf has been one of the efforts in this tendency. On the other hand, there are many different dialects of SQL. There are two major standards: ANSI (American National Standards Institute) SQL and an updated standard adopted in 1992, called SQL-92 or SQL2. This paper looks for to continue the contribution in the area flexible querying to relational databases. Basically this contribution embraces the definition of the flexible querying language SQLf2 according to those characteristics contemplated in the norms SQL2 that can be susceptible to a fuzzy treatment.

Fuzzy dominance skyline queries

Skyline is an important and recent proposal for expressing user preferences. While no one best row exists, Skyline discards rows which are worse on all criteria than some other and retrieves non-dominated or the best ones that match user preferences. Nevertheless, some dominated rows could be interesting to user requirement, but they will be rejected by Skyline. Dominated rows could be discriminated (or ranked) by means of user preferences, but Skyline only discards dominated ones and it does not discriminate them. SQLf is a proposal for preferences queries based on fuzzy logic that allows to discriminate rows and includes user-defined terms, such as fuzzy comparison operators. In this work, we propose to flexibilize Skyline queries using fuzzy comparison operators in order to retrieve interesting dominated rows. We also introduce an evaluation mechanism for these queries and our initial experimental study shows that this mechanism has a reasonable performance.

Reaching the Top of the Skyline: An Efficient Indexed Algorithm for Top-k Skyline Queries

Criteria that induce a Skyline naturally represent users preference conditions useful to discard irrelevant data in large datasets. However, in the presence of high-dimensional Skyline spaces, the size of the Skyline can still be very large, making unfeasible for users to process this set of points. To identify the best points among the Skyline, the Top-k Skyline approach has been proposed. Top-k Skyline uses discriminatory criteria to induce a total order of the points that comprise the Skyline, and recognizes the best or top-k objects based on these criteria. Different algorithms have been defined to compute the top-k objects among the Skyline; while existing solutions are able to produce the Top-k Skyline, they may be very costly. First, state-of-the-art Top-k Skyline solutions require the computation of the whole Skyline; second, they execute probes of the multicriteria function over the whole Skyline points. Thus, if k is much smaller than the cardinality of the Skyline, these solutions may be very inefficient because a large number of non-necessary probes may be evaluated. In this paper, we propose the TKSI, an efficient solution for the Top-k Skyline that overcomes existing solutions drawbacks. The TKSI is an index-based algorithm that is able to compute only the subset of the Skyline that will be required to produce the top-k objects; thus, the TKSI is able to minimize the number of non-necessary probes. We have empirically studied the quality of TKSI, and we report initial experimental results that show the TKSI is able to speed up the computation of the Top-k Skyline in at least 50% percent w.r.t. the state-of-the-art solutions, when k is smaller than the size of the Skyline.

Evaluating top-k skyline queries over relational databases

Two main languages have been defined to allow users to express their preference criteria: Top-k and Skyline. Top-k ranks the top k tuples in terms of a user-defined score function while Skyline identifies non-dominated tuples, i.e. such tuples that does not exists a better one in all user criteria. A third language, Top-k Skyline, integrates them. One of the drawbacks of relational engines is that they do not understand the notion of preferences. However, some solutions for Skyline and Top-k queries have been integrated into relational engines. The solutions implemented outside the core query engine have lost the advantages of true integration with other basic database query types. To the best of our knowledge, none of the existing engines supports Top-k Skyline queries. In this work, we propose two evaluation algorithms for Top-k Skyline which were implemented in PostgreSQL, and we report initial experimental results that show their properties.

A New Upgrade to SQLf: Towards a Standard in Fuzzy Databases

One remarkable fuzzy set based extension to SQL is SQLf. It allows fuzzy conditions in any place where SQL allows a Boolean one. At present time SQLf development is not up to date with SQL standard. Latest SQLf revision was with the SQL:1999 standard.New features of SQL:2003 standard have not been yet extended to the fuzzy case in SQLf. In this paper, we make a contribution towards a standard in Fuzzy Databases, upgrading SQLf according SQL:2003standard, proposing these new fuzzy features: fuzzy table storage,fuzzy multiset data type and fuzzy merge statement.

Preferred skyline: a hybrid approach between SQLf and skyline

The World Wide Web (WWW) is a great repository of data and it may reference thousands of data sources for almost any knowledge domain. Users frequently access sources to query information and may be interested only in the top k answers that meet their preferences. Although, many declarative languages have been defined to express WWW queries, the problem of specifying user preferences and considering this information during query optimization and evaluation remains open. Most used query languages, such as SQL and XQUERY, do not allow specifying general conditions on user preferences. For example, using the SQL ORDER BY clause one can express the order in which the answer will appear, but the user must be aware of filtering the tuples that satisfy their preferences. Skyline and SQLf are two extensions of SQL defined to express general user preferences. Skyline offers physical operators to construct a Pareto Curve of the non-dominated answers. Skyline may return answers with bad values for some criteria and does not discriminate between the non-dominated answers. On the other hand, SQLf gives the best answers in terms of user preferences, but it may return dominated answers. Finally, the skyline operator evaluation time is higher than SQLf. We proposed a hybrid approach, Preferred Skyline, integrating skyline and SQLf to produce only answers in the Pareto Curve with best satisfaction degrees. We report initial experimental results on execution time and answer precision. They show that Preferred Skyline consumes less time than Skyline and its precision is higher than SQLf.

Efficiently selecting the best web services

Emerging technologies and linking data initiatives have motivated the publication of a large number of datasets, and provide the basis for publishing Web services and tools to manage the available data. This wealth of resources opens a world of possibilities to satisfy user requests. However, Web services may have similar functionality and assess different performance; therefore, it is required to identify among the Web services that satisfy a user request, the ones with the best quality. In this paper we propose a hybrid approach that combines reasoning tasks with ranking techniques to aim at the selection of the Web services that best implement a user request. Web service functionalities are described in terms of input and output attributes annotated with existing ontologies, non-functionality is represented as Quality of Services (QoS) parameters, and user requests correspond to conjunctive queries whose sub-goals impose restrictions on the functionality and quality of the services to be selected. The ontology annotations are used in different reasoning tasks to infer service implicit properties and to augment the size of the service search space. Furthermore, QoS parameters are considered by a ranking metric to classify the services according to how well they meet a user non-functional condition. We assume that all the QoS parameters of the non-functional condition are equally important, and apply the Top-k Skyline approach to select the k services that best meet this condition. Our proposal relies on a two-fold solution which fires a deductive-based engine that performs different reasoning tasks to discover the services that satisfy the requested functionality, and an efficient implementation of the Top-k Skyline approach to compute the top-k services that meet the majority of the QoS constraints. Our Top-k Skyline solution exploits the properties of the Skyline Frequency metric and identifies the top-k services by just analyzing a subset of the services that meet the nonfunctional condition. We report on the effects of the proposed reasoning tasks, the quality of the top-k services selected by the ranking metric, and the performance of the proposed ranking techniques. Our results suggest that the number of services can be augmented by up two orders of magnitude. In addition, our ranking techniques are able to identify services that have the best values in at least half of the QoS parameters, while the performance is improved.

FOPA: A Final Object Pruning Algorithm to Efficiently Produce Skyline Points

We consider the problem of locating the best points in large multidimensional datasets. The goal is to efficiently generate all the points that meet a multi-objective query on data distributed in Vertically Partitioned Tables (VPTs). To compute the skyline on large VPTs, costly joins and comparisons may need to be executed, negatively impacting on the query execution time. We propose a new algorithm named FOPA (Final Object Pruning Algorithm) which is able to efficiently produce the whole set of skyline points and scales up to large datasets. FOPA relies on ordered VPTs, information on the values seen so far, and indices on the VPTs, to prune the space of dominated points and identify the skyline for large datasets in less time than state-of-the-art approaches. Empirically, we study the performance and scalability of FOPA in synthetic data and compare FOPA with existing approaches; our results suggest that FOPA outperforms existing solutions by up to two orders of magnitude.