StreamPref: Uma Linguagem de Consulta para Dados em Fluxo com Suporte a Preferências

Abstract

The work described herein explored the evaluation of queries over data streams (called continuous queries) with support to preferences. No related work had employed the implicit temporal information in the data streams through queries containing temporal preferences. The use of temporal preferences is interesting because it allows the user to express how the data at a given moment influences the preferences at another time moment. The main goal of the work described herein was to create a theoretical and practical framework that allows continuous queries containing temporal conditional preferences. In order to achieve this goal, we created the StreamPref query language as an extension of the Continuous Query Language (CQL). First, we explored the develomentp of a strategy to optimize the execution of queries containing conditional preferences in traditional databases. Then, we designed a new incremental algorithm that is able to evaluate continuous queries containing conditional preferences. Finally, we proposed the StreamPref language with new operators capable of processing continuous queries with temporal conditional preferences. We also developed an efficient algorithm to evaluate the new operators and a new temporal preference model for the comparison of sequence of tuples. We demonstrated that the new operators have equivalent operations in CQL showing that the StreamPref language does not increase the expression power of the CQL language. However, the new operators have specific algorithms that enable a faster processing of StreamPref queries than their CQL counterparts. Through extensive experiments, it has been proven that our algorithms outperforms the state-of-the-art algorithms. Moreover, we also executed experiments to demonstrate how different combinations of StreamPref operators affect the comparisons of sequences and, by consequence, the response to the user.