Temporal Conditional Preference Queries on Streams

Abstract

Preference queries on data streams have been proved very useful for many application areas. Despite of the existence of research studies dedicated to this issue, they lack to support the use of an important implicit information of data streams, the temporal preferences. In this paper we define new operators and an algorithm for the efficient evaluation of temporal conditional preference queries on data streams. We also demonstrate how the proposed operators can be translated to the Continuous Query Language (CQL). The experiments performed show that our proposed operators have considerably superior performance when compared to the equivalent operations in CQL.

Finding Top-k Sequences over Data Streams according to Temporal Conditional Preferences

“…Moreover, the new operator has algorithms that are specially tailored to process queries more efficiently than their CQL counterparts. The CQL equivalences for the SEQ and BESTSEQ operators were already demonstrated in our previous work [Ribeiro et al 2017b]. Equations (1a)-(1d) show how we can use the BESTSEQ operator to evaluate the TOPKSEQ operator.…”

“…This section presents the fundamental concepts regarding the preference model and the operators of the StreamPref query language [Ribeiro et al 2017a, Ribeiro et al 2017b]. Section 2.1 describes the preference model and Section 2.2 presents the existing operators.…”

“…The StreamPref language is an extension of the Continuous Query Language (CQL). The StreamPref operators do not increase the expression power of the CQL [Ribeiro et al 2017b]. However, the equivalences are not trivial since the StremPref operators is equivalent to complex operations using several CQL operators and intermediary relations.…”

“…The GetBestSeq routine is basically an algorithm to evaluate the BESTSEQ operator. This algorithm can use a naive approach [Ribeiro et al 2017a] or an incremental approach [Ribeiro et al 2017b]. The naive approach must compare all sequences at every instant as addressed by the algorithm NaiveBestSeq (see Algorithm 2).…”

“…We have been exploring this research topic in some preliminary works. We proposed a new formalism for reasoning with temporal conditional preferences [Ribeiro et al 2017a] and used this formalism to define the first version of the StreamPref query language which allows querying data streams with temporal conditional preferences [Ribeiro et al 2017b].…”