Approximated Summarization of Data Provenance

Abstract

Many modern applications involve collecting large amounts of data from multiple sources, and then aggregating and manipulating it in intricate ways. The complexity of such applications, combined with the size of the collected data, makes it difficult to understand the application logic and how information was derived. Data provenance has been proven helpful in this respect in different contexts; however, maintaining and presenting the full and exact provenance may be infeasible, due to its size and complex structure. For that reason, we introduce the notion of approximated summarized provenance, where we seek a compact representation of the provenance at the possible cost of information loss. Based on this notion, we have developed PROX, a system for the management, presentation and use of data provenance for complex applications. We propose to demonstrate PROX in the context of a movies rating crowd-sourcing system, letting participants view provenance summarization and use it to gain insights on the application and its underlying data.

Information-Balance-Aware Approximated Summarization of Data Provenance

“…We first recall some background of provenance information management, semiring provenance model, and the summarization of provenance from [1][2][3] before discussing our motivation and proposal.…”

“…Given a set of user-specified valuation , a mapped valuation of will be built for the summarized provenance expressions (denoted as ). In [3] a function (combiner function) is provided to perform this mapping. In simple words provides descriptions about how summarized annotations will be discarded or retained in the mapped valuation according to valuation choices of the original provenance.…”

“…Approximated summarization of data provenance has therefore been proposed in [3] to reduce the provenance size by grouping multiple "similar" data provenance annotations as a single annotation through mapping. Intuitively, as annotations are being merged to form new feasible annotations, each annotation would have to symbolize more and more annotations from the original provenance.…”

“…However, these are relatively loose constraints (e.g., annotations sharing at least one attribute in common may be grouped together) that are meant to keep the grouped annotations make sense, rather than to retain information that is useful for the users. On the other hand, to achieve a balance between provisioning accuracy and the size of provenance summarization, it has been proposed in [3] that the provisioning results derived from the provenance summary should be retained as much as possible compared to the one derived from the original provenance. Based on this requirement, the current provenance summarization algorithm in [3] searches for an approximated optimal provenance summary, by grouping semantically feasible annotations that will lead to maximum size reduction and minimal distance increment (combined with some weights), one pair at a time in a stepwise manner.…”

“…On the other hand, to achieve a balance between provisioning accuracy and the size of provenance summarization, it has been proposed in [3] that the provisioning results derived from the provenance summary should be retained as much as possible compared to the one derived from the original provenance. Based on this requirement, the current provenance summarization algorithm in [3] searches for an approximated optimal provenance summary, by grouping semantically feasible annotations that will lead to maximum size reduction and minimal distance increment (combined with some weights), one pair at a time in a stepwise manner. However, in this constraint, only the deviation in provisioning results, which is but one of the consequences of the information loss, is considered.…”