Labelling — Are You Serious?

Phil Archer, i-sieve Technologies

Online-Jugendschutz — geht's noch?
Hans Bredow Institute
Hamburg 25 May 2011
Slides & more at http://kwz.me/N5

Who am I?

i-sieve logo

Sentiment Analysis

Market intelligence

Custom filtering solutions

W3C logo

Web standards

I run online training courses, especially around mobile.

I was:

Full ICRA logo with text saying Internet Content Rating Association

ICRA was designed to:

Full ICRA logo with text saying Internet Content Rating Association

ICRA had:

Full ICRA logo with text saying Internet Content Rating Association

ICRA failed

Full ICRA logo with text saying Internet Content Rating Association
image of the small wooden ornament that sat on Harry Truman's desk while president saying the Buck Stops Here

Although the buck doesn't entirely stop with me, I'm not passing it on.

Three possible filtering methods

Lists

Generated by automated classification of content.

List is highly optimised for high performance, avoids having to classify the same page over and over again.

On the Fly Classification

If a URL is not on the list, the classification is done in real time.

This slows down the performance for the first user, but the result is added to the list.

Labels

Politically attractive.

Created manually.

Can be very detailed but more detail entails greater the effort.

Could help increase accuracy of on the fly classification.

Two successful filtering methods

Lists

Generated by automated classification of content.

List is highly optimised for high performance, avoids having to classify the same page over and over again.

On the Fly Classification

If a URL is not on the list, the classification is done in real time.

This slows down the performance for the first user, but the result is added to the list.

Begging the question about labels:

Why bother?

Time to say something positive…

two alpaca on top of a mountain above a lake in New Zealand, the mood is refreshing and positive!
Photo credit Bangkrood

Trusted, well-known data: film ratings

main image advertising the film Borat
Irish Film Classiciation Office info on Borat, includes line:'Politically incorrect' humour and satire on a seismic scale

Trusted, well-known data: film ratings

FSK classification for Barfuß auf Nacktschnecken

Trusted, well-known data: film ratings

Wouldn't it be good if Movie Pilot could include FSK data directly on this page? (a bit like Facebook does)

Screenshot of Movie Pilot page about Barfuß auf Nacktschnecken. No rating info shown

Trusted, well-known data: film ratings

Wouldn't it be good if Movie Pilot could include FSK data directly on this page? (a bit like Facebook does)

Screenshot of Movie Pilot page about Barfuß auf Nacktschnecken. Facebook code overlaid

Welcome to Ravensburg “Web 3.0 Region”

Wide angle shot of Ravensburg rooftops
Photo credit: Andreas Praefcke via Wikipedia
headshot of Dr Martin Hepp
Martin Hepp
On April 14, 2011, Ravensburg became the first city in the world to publish an almost complete set of high- quality information about shops, tourist attractions, medical services, and many other points of interest… www.lieber-ravensburg.de/developer

A Human View of eCommerce…

Screenshot of German version of Shopforia website, shows Amazon Kindle

A Machine's View of eCommerce…

Screenshot of RDF triples extrated from RDFa in Shopforia page
Good Relations, the Web Ontology for e-commerce logo, shows text and shopping cart

An active, successful metadata project

screenshot of one part of Good Relations wiki, shows list of tools available for creating the data
Screenshot of part of the Good Relations wiki

Good Relations & Search

screenshot from Martin Hepp's website showing effect of adding good relations data
From Hepp Research
LOD Cloud, September 2010
Linking Open Data cloud diagram, by Richard Cyganiak and Anja Jentzsch. http://lod-cloud.net/

Summary

A successful metadata system must:

  1. be relevant to consumers; visible in a way that fits in with their experiences elsewhere;
  2. benefit the content providers directly, usually by increasing visibility of, and trust in, their content;
  3. be available through a variety of trusted APIs that developers can very quickly start using as they see fit and that they already have the tools to handle;
  4. be published under an unrestrictive licence.

Online-Jugendschutz — geht's noch?

Yes, if:

A possible model: MyWOT

Trustworthiness

Vendor reliability

Privacy

Child safety

My Web Of Trust screengrab
www.mywot.com

A possible model: MetaCert

screenshot of metacert.com homepage, May 2011

A possible model: MetaCert

screenshot of metacert.com homepage, May 2011

Providing labels & other services for .xxx

Also working with Creative Commons & trustmarks

All labels delivered from MetaCert servers:

— more trustworthy;

— nothing for webmasters to add to site.

W3C standards compliant

Summary

A successful metadata system is:

  1. relevant to consumers;
  2. of benefit to content providers;
  3. available through trusted APIs;
  4. published under an unrestrictive licence.