the-last-thing/text/related/main.tex

\chapter{Related work}
\label{ch:rel}
\nnfootnote{This chapter appeared in the special feature on Geospatial Privacy and Security of the $19$th journal of Spatial Information Science~\cite{katsomallos2019privacy}.}
% \kat{Change the way you introduce the related work chapter; do not list a series of surveys. You should speak about the several directions for privacy-preserving methods (and then citing the surveys if you want). Then, you should focus on the particular configuration that you are interested in (continual observation). Summarize what we will see in the next sections by  giving also the general structure of the chapter.}
% \mk{Moved to summary}

In this chapter, we survey works that deal with privacy under continuous data publishing covering diverse use cases.
We present $48$ published articles spanning $16$ years of research from $2006$ to $2021$, with $2015$ being the median, based on two levels of categorization (Figure~\ref{fig:rel-yrs}). 
% \kat{The related work section of your thesis, should make a connection/comparison to your work. This means that you should position the works presented wrt  your problem and your solution if the problems are the same. Put a small (or big) paragraph in the end of each of the two sections (microdata and statistical data) and name the similarities/differences }
% \mk{OK}

\begin{figure}[htp]
  \centering
  \includegraphics[width=.75\linewidth]{related/rel-yrs}%
  \caption{Number of reviewed published articles on continuous data publishing of microdata and statistical data per year.}
  \label{fig:rel-yrs}
\end{figure}

First, we group works with respect to whether they deal with microdata or statistical data (see Section~\ref{subsec:data-categories} for the definitions) as input.
The works are equally divided between the two data categories, while $55$\% of them propose location-specific techniques.
Then, we further group them into two subcategories, whether they are designed for the finite or infinite (see Section.~\ref{subsec:data-publishing}) observation setting.
$59$\% of the reviewed literature deals with finite data observation, $57$\% implements the streaming publishing mode, while $77$\% applies the global publishing scheme.
Finally, we identify the privacy-related aspects of each work in terms of the method and protection level that they apply, as well as the privacy attacks that they are considering with emphasis on the underlying data correlation (see Figure~\ref{fig:rel-stats} for the detailed cumulative statistics).

\begin{figure}[htp]
  \centering
  \subcaptionbox{Privacy method\label{fig:rel-meth}}{%
    \includegraphics[width=.5\linewidth]{related/rel-meth}%
  }%
  \subcaptionbox{Protection level\label{fig:rel-prot}}{%
    \includegraphics[width=.5\linewidth]{related/rel-prot}%
  }%
  \hfill
  \\ \bigskip
  \subcaptionbox{Privacy attack\label{fig:rel-atk}}{%
    \includegraphics[width=.5\linewidth]{related/rel-atk}%
  }%
  \subcaptionbox{Data correlation\label{fig:rel-cor}}{%
    \includegraphics[width=.5\linewidth]{related/rel-cor}%
  }%
  \caption{The privacy-related aspects of the reviewed literature in terms of (a)~the privacy method utilized, (b)~the protection level provided, (c)~the privacy attack considered, and (d)~data correlation therein.}
  \label{fig:rel-stats}
\end{figure}

% \kat{continue.. say also in which category you place your work}
Our work, which we present subsequently in Section~\ref{ch:lmdk-prv}, focuses primarily on microdata for its use case.
However, it is possible to deal with statistical data in specific scenarios.
For simplicity, we limit the conversation in microdata and plan to investigate more diverse settings in our future work.


\input{related/micro}
\input{related/statistical}
\input{related/summary}