|
|
|
|
Hyperset Approach to Semi-structured Databases and the Experimental Implementation of the Query Language Delta
by Richard Molyneux.
PhD thesis, University of Liverpool.
December 2008. |
|
|
|
Thesis appendix |
 |
|
Introduction
The following page is an online appendix for the PhD thesis of Richard Molyneux - Hyperset Approach to Semi-structured Databases and the Experimental Implementation of the Query Language Delta. The following resources are taken from the examples demonstrated in the thesis, and may prove useful for those interested in this approach.
Bibliographic information (BibTeX entry)
@phdthesis{m08,
author = {Richard Molyneux},
title = {Hyperset Approach to Semi-structured Databases and the Experimental Implementation of the Query Language Delta},
school = {University of Liverpool},
year = {2008},
address = {Liverpool, England}
}
|
|
|
Worked examples |
|
|
Bibliographic semi-structured/web-like database
Most of the worked examples in the thesis are based on the bibliographic database, represented visually by the directed graph (left) and equivalently by the system of set equations (right):
 | | BibDB = {
'book':b1,
'book':b2,
'paper':p1,
'paper':p2,
'paper':p3
}
b1 = {
'refers-to':b2,
'refers-to':p1
}
b2 = {
'author':"Jones",
'title':"Databases"
}
p1 = {
'refers-to':p2
}
p2 = {
'author':"Smith",
'title':"Databases",
'refers-to':p3
}
p3 = {
'author':"Jones",
'title':"Databases"
}
|
Although the above particular graph is acyclic, we can deal in general with arbitrary labelled graphs and arbitrary systems of set equations (with labels like 'book', 'refers-to', etc.). This is the reason for using the term hyperset as hypersets can involve cycles in the membership relation, unlike the ordinary sets.
It is important that different set names can denote the same (hyper)sets like b2 and p3 above. This fact might seem somewhat intriguing for the above WDB as b2 is declared as a 'book' and p3 as a 'paper'. Equality of sets b2 = p3 is exploited in some examples of queries below. Note that it is the problem of database designer to create a database in a reasonable way. Formally speaking there is no contradiction in the above example. (Forget the English meaning of words book and paper and you will see no contradiction. After all, databases are not about English.)
Atomic data like "Smith" are treated as singleton sets: "Smith" = {'Smith':{}}. In this sense atomic data are formally unnecessary and can be modelled by labels (and additional set equations).
Note that, for the purpose of making XML compatible with our hyperset approach, systems of set equations are represented in so called XML-WDB format (so that arbitrary XML elements can participate in set equations). Moreover, in the above example the bibliographic database is distributed into two fragments (subsystems of set equations) as illustrated by the colouring of the graph above. Set equations for set names BibDB, b1 and b2 are appropriately encoded into XML-WDB file 1, whereas set equations for set names p1, p2 and p3 are encoded into XML-WDB file 2 (see below).
|
|
|
|
|
|
