--- /dev/null
+Intro:
+ Hyperleap + infotaintment
+ Relieve programmer fixing crawlers
+ System to generate crawler specification
+ Frontend useable for non programmers
+
+Frontend:
+ Runs in browser
+ Runs from apache and python
+
+Backend:
+ Converts the user patterns from frontend to nodelists.
+ Nodelists are merged into DAWG minimization to generate patterns(graphs).
+ The crawler reads the patterns and crawls the site.
+ Crawler results are send via an XML/XSD stream to the original backend.
+
+Results:
+ Few RSS
+ Much RSS misuse
+
+Future:
+ Extend to HTML (program to convert HTML to RSS)
+ Reuse interface
+ Low level matching can increase
+
+Questions:
+- Why is user interface easy to use
+ Direct feedback
+ Familiar interface with buttons and textboxes
+
+- Why did you choose RSS
+ We had to limit scope
+ RSS is very consistent in underlying structure
+ But RSS doesn't have any structure in itself but underlying because
+ they are generated
+
\begin{frame}
\frametitle{Frontend}
+ \begin{itemize}[<+->]
+ \item Useable by non programmers
+ \end{itemize}
\begin{figure}[H]
\only<2>{\includegraphics[width=\linewidth]{frontendfront}}
\only<3>{\includegraphics[width=\linewidth]{crawlerpattern}}
\begin{itemize}[<+->]
\item Based on Daciuk et al.
\item User patterns become nodes/subgraphs
+ \item Incremental construction
\end{itemize}
\end{frame}