Clemens Klug
parent
b5a106109e
commit
dfc27ffc81
|
|
@ -1,4 +1,4 @@
|
|||
\section{methodology}
|
||||
\section{Methodology}
|
||||
BioDiv2Go's Geogame2 (FindeVielfalt Simulation) was base case during the development of the analysis stack.
|
||||
It was chosen due to its well defined REST API, including log retrieval and user authentication.
|
||||
This section shows how the framework copes with the integration of another game with completly different architecture and log style.
|
||||
|
|
@ -86,4 +86,6 @@ While the implementation of a new client to download log files was straightforwa
|
|||
However, it was not the integration into the framework but the syntactical errors in the log files that was hard.
|
||||
While the BioDiv2Go parser requires less than 20 lines of code, the newly written parser scratches the 60 line mark with all the error handling code (see \autoref{code:bd2l} and \ref{code:ncl}).
|
||||
Once this obstacle is passed, the integration is nearly seamless.
|
||||
%TODO: webclient
|
||||
%TODO: webclient
|
||||
|
||||
As further challenge prooved - like with BioDiv2Go - the understanding of the structure of the log, i.e. deriving the games' internal state machine.
|
||||
|
|
|
|||
|
|
@ -1,3 +1,20 @@
|
|||
\begin{verbatim}
|
||||
…
|
||||
\end{verbatim}
|
||||
\section{Review}
|
||||
As shown in \autoref{sec:eval}, the proposed framework (see \autoref{sec:solution}) and its implementation (see \autoref{sec:implementation}) deliver what \autoref{sec:scope} asked for regarding the portability aspect.
|
||||
With the web interface depicted in \autoref{app:webif}, it is possible for non-expert users to generate pre-defined reports, while researchers can dive into the API of the framework either as preprocessing step or integrated into a larger project.
|
||||
Given the lean framework core, the development of new analyzers and rendering target is encouraged.
|
||||
This is backed by the focus on a standalone application instead of extensions to log processing systems struggling with spatial data in the required resolution.
|
||||
|
||||
\subsection{Results}
|
||||
Th selection of rendered results in \autoref{img:oebkml}, \ref{img:oebge}, \ref{img:retries}, \ref{img:trackfi}, \ref{img:time} showcases the already possible descriptive analysis capabilities.
|
||||
\autoref{img:trackfi} features a map view accessible through a browser, which aligns the active screen content of the mobile device with the spatial track.
|
||||
Selecting the preview image in the timeline below the map, a marker shows the position where this content entered the screen.
|
||||
|
||||
\autoref{img:speed} is based on the same analysis result used for \autoref{img:trackfi} with additional postprocessing in between.
|
||||
This aggregates the speeds of all selected game sessions and calculates the average.
|
||||
With a baseline reference of 1 kph speed, this plot can hint at gamefield design issues, e.g. overly long walking passages at high speed.
|
||||
|
||||
\section{Outlook}
|
||||
Considering the future, there are many analysis and rendering targets to come.
|
||||
As with any kind of software, sooner or later the point may be reached where the proposed architecture fails.
|
||||
Given the positive result of the integration evaluation, this sure seems pessimistic.
|
||||
The framework shows some self-debugging feature: Once a log parser is working, analyzers can help to find patterns of log messages to outline the logs structure.
|
||||
|
|
|
|||
|
|
@ -5,6 +5,10 @@
|
|||
\image{\textwidth}{../../PresTeX/images/oeb-ge}{Result visualized}{img:oebge}
|
||||
\image{\textwidth}{../../PresTeX/images/simu-retries}{Experimentational rounds}{img:retries}
|
||||
|
||||
\image{.7\textwidth}{../../PresTeX/images/track-fi}{ActivityMapper: Combined screen activity and spatial progress}{img:trackfi}
|
||||
\image{\textwidth}{../../PresTeX/images/speed}{Aggregated speed distribution of four gamefields}{img:speed}
|
||||
\image{.9\textwidth}{../../PresTeX/images/time-rel}{Time distribution of game sessions overview of four gamefields}{img:time}
|
||||
|
||||
\section{Containers}
|
||||
\subsection{Kibana test setup} \label{app:kibana}
|
||||
\lstinputlisting[language=yaml,caption={Docker-compose file for Kibana test setup},label=code:kibana,numbers=left]{code/kibana-docker-compose.yml}
|
||||
|
|
@ -25,16 +29,7 @@
|
|||
\section{Loader implementations}
|
||||
\lstinputlisting[language=python,caption={Log loader for BioDiv2Go},label=code:bd2l,numbers=left]{code/biogames.py}
|
||||
\lstinputlisting[language=python,caption={Log loader for Neocartographer},label=code:ncl,numbers=left]{code/neocart.py}
|
||||
\section{TODO}
|
||||
\subsection{Examples} %TODO ?!?!
|
||||
Configuration \& results
|
||||
%\twofigures{0.5}{../../PresTeX/images/oeb-kml}{Analyzer configuration}{img:oebkml}{../../PresTeX/images/oeb-ge}{Result visualized}{img:oebge}{Example: Generate KML tracks (BioDiv2Go; Oberelsbach2016)}{fig:oeb2016}
|
||||
|
||||
\section{Web interface}\label{app:webif} %TODO
|
||||
|
||||
ActivityMapper
|
||||
\image{.7\textwidth}{../../PresTeX/images/track-fi}{Combined screen activity and spatial progress}{img:trackfi}
|
||||
|
||||
|
||||
Graphs
|
||||
\image{\textwidth}{../../PresTeX/images/speed}{Speed distribution}{img:speed}
|
||||
\image{.9\textwidth}{../../PresTeX/images/time-rel}{Time distribution}{img:time}
|
||||
TODO: screenshots
|
||||
|
|
@ -10,11 +10,11 @@
|
|||
\chapter{Design of the analysis framework}\label{sec:solution}
|
||||
\input{content/3-solution}
|
||||
|
||||
\chapter{Implementation of the analysis framework}
|
||||
\chapter{Implementation of the analysis framework}\label{sec:implementation}
|
||||
\input{content/4-implementation}
|
||||
|
||||
\chapter{Portability evaluation of the analysis framework}\label{sec:eval}%TODO
|
||||
\chapter{Portability evaluation of the analysis framework}\label{sec:eval}
|
||||
\input{content/5-evaluation}
|
||||
|
||||
\chapter{!diskussion}%TODO
|
||||
\chapter{Discussion and outlook}
|
||||
\input{content/6-discussion}
|
||||
|
|
|
|||
Loading…
Reference in New Issue