Semi-structured data
and the Web

Uli Sattler and Bijan Parsia

{sattler•bparsia@cs.man.ac.uk}

Goals of the course

The Web is

a very large (by several metrics) information structure

Indexable Web estimated size (Jan 2005): 11.5 billion pages

a complex, distributed, often interactive computing platform

and very large

This course will acquaint you with the representation of data on (or inspired by) the Web

semi-structured data
XML, HTML, and RDF/OWL
data, documents, and (Web) applications

New Course!

The semi-structured data part was taught last year

In conjunction with a Data Mining part

The Web part is new

There is still great flexibility in what's covered
Input welcome

In both areas, change is rapid

Change comes from the top (from big corps and standards bodies)
Change comes from the bottom (from small communities or individuals)
Separate hype and flash from fundamentals

Who are we?

Uli Sattler

Professor in the IMG group
Developed key algorithms which made practical the logics underlying the Web Ontology Language

Bijan Parsia

Lecturer in the IMG group
Participated in several Web Services standardization efforts, e.g., WSDL, WS-Policy

Data on the Web

Databases are designed to store relations/tuples w.r.t. a schema

follows a client/server architecture

Often data on the Web

does not come in tuples, but often in “documents” with less structure: semi-structured

is shared:

needs to be self-describing

to cope with uncoordinated reuse, evolution, etc.

in a multi-tier and decentralized architecture

Semi-structured Data

A semi-structured data format:

partially self-describing, can be enhanced by a schema (or more than one!)

person:
    {name: {first: "Uli", last: "Sattler"},
   tel: 56176,
   office: “Rm 2.24”},

person:
    {name: {first: "Bijan", middle: "John"},
   tel: 56199,
   tel: 50690,
   office: “2.32”}

How do we

describe the commonality
cope with the variation
prescribe certain aspects

without overprescribing!

SSD and XML

XML is a W3C standard for data exchange on the Web

based on embedded markup <b>..</b> to indicate structure

a textual representation of data
e.g., <person> <name><first>Uli</first> <last>Sattler</last></name> <tel> 56176</tel> <email>sattler@cs.man.ac.uk</tel>
</person>

Schema Langauges for XML

XML schema languages support structural constraints e.g.,

a <tel> contains a number
a <person> may contain one or more <tel>s

Like a grammar for XML documents
We can validate an XML document against a schema

i.e., check the (loose) structure against a schema (or several schemas)
need to describe tree structures with "holes" and "extra stuff"

We discuss 2 XML schema languages and their validation problems (automata-based):

DTDs

XML Schema

Foundations

The Web has some ubiquitous components

URIs, HTML, and HTTP

Web architecture claims 3 components

Identification (URIs)
Interaction (roughly HTTP)
Data formats (e.g., HTML)

Classic Web Applications (e.g., web sites)

In one sense, client/server

With a strong resemblance to old mainframe applications

But "screens" can come from anywhere

And transitions go anywhere

Web Services

Basic HTTP provides stateless, simple interactions

No notion of session
Limited set of basic operations
Request-response protocol

Applications may have complex interfaces

With strong dependancies between operations

For example, check credit>>verify order>>charge

But still want the advantages of Web like interop
More elaborate messaging structure

Well described interfaces

Sufficient to support composition

Rich Internet Clients

Consider the user experience

Web Apps have benefits

Data security and access
Integration with the Web
Cross platform
App always up to date

And downsides

Loss of data control and privacy
UI often clunky or hidously painful

And dangerous! "Edit the text locally then cut and paste to the textarea"

Requires network access to use
Requires server access to deploy!

How to improve the user experience?

Without losing Web benefits?

Practicalities

Period 3

Jan. 28 - Feb 25, 2008
Mondays, 9:00 - 17:00

Lectures

Semi-structured data and core XML

DTDs

XPath, XML namespaces, and XSLT

APIs: DOM and SAX

XML Schema and XQuery

XML documents and RDBMSs

URI, HTTP, and HTML

Data formats and Semantic Markup

REST vs. SOA

Service description

Service composition

(Rich) Internet Clients

21-09-2007

COMP60370: Semi-structured data and the Web

Semi-structured data
and the Web

Uli Sattler and Bijan Parsia

{sattler•bparsia@cs.man.ac.uk}

Goals of the course

New Course!

Who are we?

Data on the Web

Semi-structured Data

SSD and XML

Schema Langauges for XML

Query Languages

SSD & XML: Prerequisites

Applications on the Web

Foundations

Web Services

Rich Internet Clients

Practicalities

Practicalities (cont.)

21-09-2007

COMP60370: Semi-structured data and the Web

Semi-structured dataand the Web

Uli Sattler and Bijan Parsia

{sattler•bparsia@cs.man.ac.uk}

Goals of the course

New Course!

Who are we?

Data on the Web

Semi-structured Data

SSD and XML

Schema Langauges for XML

Query Languages

SSD & XML: Prerequisites

Applications on the Web

Foundations

Web Services

Rich Internet Clients

Practicalities

Practicalities (cont.)

Semi-structured data
and the Web