Evolutionary and Swarm Computing for scaling up the Semantic Web

Evolutionary and Swarm Computing
for scaling up the Semantic Web

Christophe Guret (@cgueret), Stefan Schlobach, Kathrin Dentler,Martijn Schut, and Gusz Eiben

24th Benelux Conference on Artificial IntelligenceMaastricht University, October 25-26, 2012

What are we going to talk about?

Linked Data

Changing our point of view on soundness and completeness

Consider optimisation as an alternative to logical deduction

Two concrete examples of re-formulated problems

Short paper based on this publication

When solutions do not (quite) fit the problem ...

Copyright: sfllaw (Flickr, image 222795669)

Linked Data

Graph/facts based knowledge representation tool

Connect resources to properties / other resources

Web-based: resources have a URITry http://dbpedia.org/resource/Amsterdam

Interacting with Linked Data

Common goalsCompleteness: all the answers

Soundness: only exact answers

Motivation

In the context of Web data ?Issues with scale

Issues with lack of consistency

Issues with contextualised views over the World

Revise the goalsAs many answers as possible (or needed)

Answers as accurate as possible (or needed)

From logic to optimisation

Optimise towards the revised goals

Need methods that cope with uncertainty, context, noise, scale, ...

Answering queries over the data

Copyright: jepoirrier (Flickr, image 829293711)

The problem

Match a graph pattern to the data

Most common approachJoin partial results for each edge of the query

Solving approaches

Logic-basedFind all the answers matching all of the query pattern

OptimisationFind answers matching as much of the query as possible

Important implications of the optimisationOnly some of the answers will be found

Some of the answers found will be partially true

An optimisation approach: eRDF

Guess the answers to the query

Evolutionary algorithmEvaluate validity of candidate solution

Optimise with a recombination + local search

Some results

Tested on queries with varied complexity

Works best with more complex queries

Find exact answers when there are some

Finding implicit facts in the data

Copyright: givingnot@rocketmail.com (Flickr, image 6990161491)

The problem

Deduce new facts from others

Most common approachCentralise all the facts, batch process deductions

Solving approaches

Logic-basedFind all the facts that can be derived from the data

OptimisationFind as many facts as possible while preserving consistency

Important implications of the optimisationOnly some of the facts will be found

Unstable content

An optimisation approach: Swarms

Swarm of micro-reasonersBrowse the graph, applying rules when possible

Deduced facts disappear after some time

Every author of a paper is a personEvery person is also an agent

Some results

If they stay, most of the implicit facts are derived

Ants need to follow each other to deal with precedence of rules

Several ants per rule are needed

Take home message

Logic problems can be turned into optimisation problems

Trade offGained: scalability, speed, robustness

Lost: determinism, completeness, soundness

A lot of research still to be done!(and done quickly, Linked Data is growing fast...)

Click to edit the outline text formatSecond Outline LevelThird Outline LevelFourth Outline LevelFifth Outline LevelSixth Outline LevelSeventh Outline Level

/