\chapter{Related work}
\label{ch:related}

Depending on the intended use, owners may share their data either as a whole (Microdata -- Section~\ref{sec:microdata}), or as computed statistics thereof (Statistical Data -- Section~\ref{sec:statistical}).
This is the basic division line that we set in this section.
Table~\ref{tab:related} summarizes all the works reviewed in this thesis, and provides a guide for the interested reader on some of the listed variables.
We proceed by identifying some common variables in the works of both categories, which are listed in Table~\ref{tab:related}.
There are three main columns, concerning the:

\begin{itemize}
  \item \textbf{Data} --- The first part of Table~\ref{tab:related} considers the nature of the input/output data of the algorithms in its first column.
  We identify here the considered kind of data from the set of categories defined in Section~\ref{subsec:data-categories}, i.e.,~stream, sequential, or time series.
  Particularly, we outline the cases where spatial data are explicitly considered; nevertheless, all other algorithms could be equally applied on location data as well.
  The second column of the Data part is about the model (batch or stream) of the privacy preserving process and the aggregation process, where applicable.
  The third column displays the publishing scheme, as one of the defined schemes in Section~\ref{subsec:data-publishing} one shot, continuous, sequential, or incremental.
  \item \textbf{Protection} --- The second part of Table~\ref{tab:related} contains four columns, namely the attack scenario, the base protection method, the acquired protection level, and the distortion applied.
  The different attack scenarios are described in Section~\ref{subsec:privacy-attacks}, whereas the base protection methods along with the applied distortion method are in Section~\ref{subsec:privacy-seminal}.
  The possible protection levels are: event, user, and w-event (see Section~\ref{subsec:privacy-levels}).
  \item \textbf{Correlations} --- The final part of Table~\ref{tab:related} is dedicated to correlations, which is an attack model not taken into consideration by all the works.
  Still, the continuous publication of data inevitably creates correlations even when they are not evident in a standalone data set.
  The different kinds of correlations can be found in Section~\ref{subsec:privacy-attacks}.
\end{itemize}

\input{graphics/table-related}

\input{microdata}

\input{statistical}

\input{discussion}