IASI

IASI

Head: C. Reynaud

Artificial Intelligence and Inference Systems

Gemo: Joint team between LRI - Univ. Paris Sud & INRIA Saclay – Île-de-France

28 Septembre 2009 Franco-Japanese Workshop on Hypotheses Finding and its Applications 2

Objectives / Scientific Themes

Objective: Developing « Intelligent » Web Information Systems

- dealing with numerous, highly heterogeneous and distributed information

- providing fast access to meaningful data

Automate search, control and optimization tasks Achieve a new quality of service

- Semantic Web - Peer-to-Peer (P2P) - Data and Web Services

+ SAT Solvers, Diagnosis and diagnosability

Scientific Themes:

28 Septembre 2009 Franco-Japanese Workshop on Hypotheses Finding and its Applications 3

Semantic Web“An ontology, a support to Web Information Retrieval and

to the Integration of Heterogeneous Sources”

1. Web Information Retrieval• Filtering of documents (D. Mezaour PhD)

• Exploiting adaptive ontologies (C. Pruski PhD)

2. Semantic Annotation of Web Documents• Annotation of tables (F. Saïs PhD)

• Annotation of more or less structured documents (ongoing PhD - M. Thiam)

3. Semantic Integration of Heterogeneous Sources • Ontology Alignment (H. Kefi PhD - ongoing PhD – F. Hamdi)

• Reference Reconciliation (F. Saïs PhD)

4. Hidden Web• Automatic discovery, analysis,

querying (P. Senellart PhD)

Collaboration with Avin Mittal (Bombay), the MOSTRARE project, G. Gottlob (Univ. Oxford)

C. Reynaud

C. Reynaud

N. Pernelle

F. Saïs

P. Senellart

F. Saïs

WebCrawler

WebQueL

TARGET

XTab2SML

SHIRI-Annot

TaxoMap

LN2R

28 Septembre 2009 Franco-Japanese Workshop on Hypotheses Finding and its Applications 4

Web Information Retrieval

Surface Web

Hidden Web

microoganisme

TOP

aliment facteur

Ontology

Crawling

Query

Querying interface

Answer

SemanticEnrichment

Expansion

Adaptation

Filtering

28 Septembre 2009 Franco-Japanese Workshop on Hypotheses Finding and its Applications 5

Exploiting Thematic Web Data Warehouses and Integrated Heterogeneous Sources

microoganisme

TOP

aliment facteur

Ontology

Crawling

Filtering

Query

Querying interface

XML data warehouse

Semantic Enrichment

Local datawarehouse

AnswerWeb

External Source

Local Schema

Ontology Alignment

Reference Reconciliation

28 Septembre 2009 Franco-Japanese Workshop on Hypotheses Finding and its Applications 6

Peer To Peer Inference Systems

“Integration of Distributed Knowledge”

1. Distributed Reasoning • Decentralized consequence finding algorithm (P. Adjiman PhD)

• Reasoning in possibly inconsistent systems (N. Gia Hien PhD)

• Trust modeling (N. Gia Hien PhD)

• Conservative extension of a peer (ongoing PhD – N. Abdallah)

• Diagnosis in P2P framework (ongoing PhD – V. Armant)

SomeWhere *

SomeWhere+

2. P2P Data management Systems (PDMS) for the Semantic Web (P. Adjiman PhD)

3. Discovering of mappings (ongoing PhD – F.-E. Clavier)

• PDMSs based on OWL and RDFS

• Query rewriting algorithms

• Scalability

SomeRDFS

SomeOWL

SpyWhere

* registred software

F. Goadoué – P. Chatalic – L. Simon – P. Dague – V. Armant

C. Reynaud

28 Septembre 2009 Franco-Japanese Workshop on Hypotheses Finding and its Applications 7

Collaboration withthe BD group

Y. Papakonstantinou (UCSD)N. Polyzotis (UCSC)

P2P Warehousing«  Integration of Distributed Data »

1. P2P Management of XML Data

(N. Preda et Radu Pop PhD (CIFRE with Mandriva)

Ongoing PhD - S. Zoupanos)KadoP

OptimAX

2. P2P Management of XML Data annotated with Semantic Web Data (WebContent ANR project)

SomeRDFS KadoP

OptimAX

Key idea: XML indexing and query optimization using distributed hash tables (DHTs) and ActiveXML technology

Scales to millions of XML documentsOpen-source system

• Scalable management of XML data in P2P networks based on distributed hash tables (DHTs)

• Query optimization

I. Manolescu

28 Septembre 2009 Franco-Japanese Workshop on Hypotheses Finding and its Applications 8

Data and Web Services

• Modelling complex Web services

• Decentralized diagnostic algorithm

1. Active XML Active XML

2. Self-healability Web services

AXML is a declarative language for distributed information management and an infrastructure to support the language in a P2P framework

Simple idea: XML documents with embedded service calls

• Support intentional and dynamic data (some data are given via service calls)• Algebra and calculus• Monitoring • Verification (Docflow)• Prototype in Open Source

Collaboration withYuhong Yan (NRC, Canada)

(P. Bourhis PhD - B. Marinoiu)

(Ongoing PhD - Y. Li – L. Ye)

Collaboration with T. Milo (Univ. of Tel Aviv),

O. Benjelloun (Google Research), V. Vianu (UCSD)

S. Abiteboul

P. Dague

28 Septembre 2009 Franco-Japanese Workshop on Hypotheses Finding and its Applications 9

An Active XML document

<?xml version="1.0 " encoding= " UTF-8 " ?><newspaper xmls= " http://lemonde.fr "

xmlns:rss= " http://purl.org/rss " xmlns:axml= " http://activexml.net " >

<title>Le Monde</title><date>1-jan-2009</date><edition>Paris</edition> 

<weather> <axml:call service= [email protected] > <city>Paris</city> <unit>Celsius</unit> </axml:call></weather>

</newspaper>

Web Service call

28 Septembre 2009 Franco-Japanese Workshop on Hypotheses Finding and its Applications 10

And …

• On-line diagnosis of embedded systems (ongoing PhD – M. Batteux, CIFRE)

• Diagnosability analysis:

- formalization

- probabilistic diagnosability analysis

- link between observability and diagnosability

- distributed diagnosability checking

1. SAT Solvers

GUNSAT

2. Diagnosis and diagnosability

• Organization of the international SAT contest since 2002

• Algorithm experimentations • Working on incomplete algorithms for unsatisfiability

Collaboration with the CRIL Lab. of Lens

L. Simon

P. Dague

28 Septembre 2009 Franco-Japanese Workshop on Hypotheses Finding and its Applications 11

Reports

1. 2005-2008 IASI report

2. 2005-2008 LRI report

http://www.lri.fr/Eval2008/rapports/iasi.pdf

http://www.lri.fr/Eval2008/rapports/ra2008.pdf