Answering queries over incomplete data stream histories

Distributed data streams are an important topic of current research. In such a setting, data values will be missed, e.g. due to network errors. This paper aims to allow this incompleteness to be detected and overcome with either the user not being affected or the effects of the incompleteness being reported to the user.

A model for representing the incomplete information has been developed that captures the information that is known about the missing data. Techniques for query answering involving certain and possible answer sets have been extended so that queries over incomplete data stream histories can be answered.

It is possible to detect when a distributed data stream is missing one or more values. When such data values are missing there will be some information that is known about the data and this is stored in an appropriate format. Even when the available data are incomplete, it is possible in some circumstances to answer a query completely. When this is not possible, additional meta‐data can be returned to inform the user of the effects of the incompleteness.

The techniques and models proposed in this paper have only been partially implemented.

The proposed system is general and can be applied wherever there is a need to query the history of distributed data streams. The work in this paper enables the system to answer queries when there are missing values in the data.

This paper presents a general model of how to detect, represent, and answer historical queries over incomplete distributed data streams.