chapter03.tex

\chapter{Methods of data structuring}
\label{ch:methods}

% This chapter contains a comprehensive survey of methods that
% are used to store and structure data in the wild.

% ACHTUNG: Nur das jeweils charachteristische sollte beschrieben werden
% Beschränkung aufs Wesentliche, Extremfälle, Besonderheiten
% Deutlich machen: Warum ist diese Typologie/Classification sinnvoll

This chapter holds the main empirical part of my thesis. Based on intensive
review of literature and standards, I give a comprehensive analysis of methods
and systems for structuring and describing data. The summary focuses on
conceptual properties: details of implementation such as performance and
security, are only mentioned briefly, if they show why specific techniques have
evolved. The goal of this analysis is to later find patterns and paradigms
independent from particular methods. For this reason I followed
\textcite{Meek1995} whose trick to become language-independent was ``to develop
a healthy disrespect for all languages, and look for faults in them all the
time.'' The division of methods in sections partly anticipates a more detailed
typology that will be developed in detail in chapter~\ref{ch:findings}. The
survey starts with character encodings (section~\ref{sec:characters}) that are
needed to express any textual data.  Identifiers
(section~\ref{sec:identifiers}) are used as part of all other methods as as
well. The most basic method to store and manage digital data are files
(section~\ref{sec:filesystems}) followed by databases
(section~\ref{sec:databases}). The analysis does not consider concrete database
systems, but general database models which \tacro[Database Management
System]{Database Management Systems}{DBMS} can be classified into.  Data
structuring or serialization languages (section~\ref{sec:dsl}) organize data in
general forms for storage and exchange; popular examples include \acro{XML},
\acro{CSV}, and \acro{RDF}. There is some overlap with markup languages
(section~\ref{sec:markuplanguages}), which mainly apply to text and similar
sequential data. Schema languages (section~\ref{sec:schemas}) express logical
schemas as data formats.  Conceptual modeling languages (\ref{sec:modelangs})
are used to capture a part of reality in formal language. They are often
combined with a graphical notation which is a strict form of a conceptual
diagram (section~\ref{sec:diagrams}).  Query languages
(section~\ref{sec:queries}) can define or be part of an \tacro{Application
programming Interface}{API} to select or identify a specific piece of data.

\pagebreak
\input{sec-characters}

\input{sec-identifiers}
\pagebreak

\include{sec-filesystems}
\pagebreak

\input{sec-databasemodels}
\pagebreak

\input{sec-dsl}
\pagebreak

\input{sec-markuplanguages}
\pagebreak

\input{sec-schemas}
\pagebreak

\input{sec-modelangs}
\pagebreak

\input{sec-diagrams}
\pagebreak

\input{sec-apis}

% ``The entity-relationship model adopts the
% more natural view that the real world consists of entities
% and relationships.

% See http://en.wikipedia.org/wiki/AmigaOS#Datatypes
% Datatypes / formats on OS level! (also in BeOS)

% OODBs. World is a set of objects.
% On the Representation of Roles in Object-Oriented and Conceptual Modelling
% \cite{Steimann2000} pointed out that roles, as introduced by
% \person[Charles]{Bachman} \cite{Bachman1977} are a third, fundamental
% concept beside objects and relationships.


% MOTIVATION: Many formats and data without common model or with no model at all (model not available, only logical schema exists...)
% REMINDER: one reality -> many possible models
% REMINDER: one schema -> multiple possible models
% REMINDER: modeling is design , is an art!
%
% model integration needed for information integration!
% but literature deals most with schema integration!

%----
% Sehr geehrter Herr Naumann,
%
% Ich sitze derzeit an meiner informationswissenschaftlichen Dissertation über
% die Modellierung von bibliographischen Daten. Dabei bin ich wieder auf das
% Probleme der Informationsintegration gestoßen. Im SSWS 200? hatte ich dazu
% an der HU ein Hauptseminar besucht und meine Prüfungsleistung als Wikipedia-
% Artikel abgegeben, vielleicht können Sie sich noch erinnern - wir hatten uns
% dann nochmal auf der WWW200? in Edinburgh getroffen.
%
% Ich habe folgende Frage ...
%
% Nach meinem Eindruck beschränkt sich der Großteil der Informationsintegration
% mit der Integration von Schemata wie SQL-Schemas, XML Schemas etc. Nach der
% traditionellen Einteilung der Datenmodellierung in ...
%
% 1. Kennen Sie gute Arbeiten, die sich mit der Integration von Informationen
% auf der konzeptuellen Ebene beschäftigen - d.h. unabhängig davon, ob Daten 
% in XML, SQL, RDF, CSV oder wie auch immer abgelegt werden?

% Anbei schicke ich Ihnen einen Auszug aus dem Grundlagenkapitel meiner
% Arbeit, wo ich u.A. kurz Datenmodellierung erkläre. Im Hauptteil nehme
% ich eine Bestandsaufnahme von verschiedensten Methoden zur Strukturierung
% von Daten vor (angefangen von Dateisystemen und Datenbanken über Sprachen
% wie CSV und XML und Schemasprachen wie XSD und OWL bis zu graphischen
% Notationssystemen wie ER-Diagramme und Mindmaps)


% Conceptual Mapping / Topic Maps / Structured Maps
% But Evans et al. (2003) argue that 'a truly flexible model-driven development 
% process should not dictate the language that practitioners should use to
% construct models, even an extensible one. Instead they should be free to use
% whichever language is appropriate for their particular domain and application,
% particularly as many languages cannot be shoe-horned into the UML family'. We
% follow this argument and suggest extending the model mapping of MDA (Caplat &
% Sourrouille, 2003) to 'language mapping'. 
% In: Rittgen, 2006
%
% XML Language for Conceptual Graphs exist!