Phd defense - Linked data based exploratory search - Nicolas MARIE

Preview:

Citation preview

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Linked data based exploratory searchPhd defense – Nicolas MARIE – 12.12.2014

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

10 blue links paradigm, Simple, fast

Exploratory searchbottleneck

Exploratory search: open search tasks having a general objective like learning or investigating a poorly known topic

Search is only a partially solvedproblem

[White, 2009]

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

10 blue links paradigm, Simple, fast

Exploratory searchbottleneck

Exploratory search: open search tasks having a general objective like learning or investigating a poorly known topic

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

10 blue links paradigm, Simple, fast

Exploratory searchbottleneck

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

<1990 Web

<2001 + Semanticweb

<2007+ Linked data

<2012 +ProprietaryKG

Semantic search refers to the incorporation of structured semantics in search

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

HumanComputer InformationRetrieval

Linked data based exploratory searchResearch questions

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

HumanComputer InformationRetrieval

Linked data based exploratory searchResearch questions

• How can we discover and rank linked resources to be explored starting from the user topic(s) of interest?

• How to address remote linked data source for this selection?

• How to optimize a data based exploration at the interaction level?

• How to evaluate such exploratory search systems?

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Plan

• How can we discover and rank linked resources to be explored starting from the user topic(s) of interest?

• How to address remote linked data source for this selection?

• How to optimize such data based exploration approach at the interaction level?

• How to evaluate such exploratory search systems?

• State-of-the-art review

• Conclusion

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Plan

• How can we discover and rank linked resources to be explored starting from the user topic(s) of interest?

• How to address remote linked data source for this selection?

• How to optimize such data based exploration approach at the interaction level?

• How to evaluate such exploratory search systems?

• State-of-the-art review

• Conclusion

• Contribution 1: several exploratory search algorithms optimized for linked datasets

• Contribution 2: their efficient execution over distant linked datasets

• Contribution 3: a novel interaction model and its implementation in the Discovery Hub web application

• Contribution 4: multiple users’ evaluations of the Discovery Hub results

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Plan

• How can we discover and rank linked resources to be explored starting from the user topic(s) of interest?

• How to address remote linked data source for this selection?

• How to optimize such data based exploration approach at the interaction level?

• How to evaluate such exploratory search systems?

• State-of-the-art review

• Conclusion

• Contribution 1: several exploratory search algorithms optimized for linked datasets

• Contribution 2: their efficient execution over distant linked datasets

• Contribution 3: a novel interaction model and its implementation in the Discovery Hub web application

• Contribution 4: multiple users’ evaluations of the Discovery Hub results

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

State-of-the-art review

Linked data• Browsers• Recommenders• Exploratory search systems

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

State-of-the-art review

Linked data• Browsers• Recommenders• Exploratory search systems

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

State-of-the-art review

Linked data• Browsers• Recommenders• Exploratory search systems

Trade-off

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Systems matrix

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Timeline

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Opportunities identification

Semantic search aspects

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Opportunities identification

Exploratory search aspects

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Plan

• How can we discover and rank linked resources to be explored starting from the user topic(s) of interest?

• How to address remote linked data source for this selection?

• How to optimize such data based exploration approach at the interaction level?

• How to evaluate such exploratory search systems?

• State-of-the-art review

• Conclusion

• Contribution 1: several exploratory search algorithms optimized for linked datasets

• Contribution 2: their innovative implementation over linked datasets

• Contribution 3: a novel interaction model and its application in the Discovery Hub web application

• Contribution 4: multiple users’ evaluations of the Discovery Hub results

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Algorithm - challenges

• Richness/complexity of linked datasets• Supporting diverse exploration needs

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Spreading activation

• Well-tried algorithm• Connectionist• Semantic friendly• Fast

1.0

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Semantic spreading activation

Filtering

Similarity

Semanticweighting

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

a

w

g

q

j

k

s m

i

p

e

o

b

n

1

class propagation domain:

Artist, Museum

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

a

w

g

q

j

k

s m

i

p

e

o

b

n

1

class propagation domain:

Artist, Museum

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

a

w

g

qs m

p

e

o

n

1

, dcterms:category, ?x

, dcterms:category, ?x

Similarity computation

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

a

w

g

qs m

p

e

o

n

1

a(...)a(...) * 4

a(...)

a(...)

a(...) * 3

a(...)

a(...)a(...)

a(...)

a(...) * 2

a(...) * 3

, dcterms:category, ?x

, dcterms:category, ?x

Category:People from ParisCategory:French paintersCategory:Impressionist painters

= +3

Similarity computation

a(...)

a(...)

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Composites interest queries: knowing my interest for X and Y what can I discover/learn which is related to all theseresources?

The BeatlesKen Loach

Polycentric algorithm

?????

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Composites interest queries: knowing my interest for X and Y what can I discover/learn which is related to all theseresources?

Ken Loach

Polycentric algorithm

Margaret Thatcher

?????

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Polycentric semantic spreading activationPropagation domain ( ) Propagation domain ( , )

Bands, Movies, Music genreMovies, Office Holder,

Political party

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Multi-perspectives exploration

The models and algorithms we proposeunveil topic knowledge nuances by allowing the exploration of topics through several perspectives.

In the graph context of linked data these perspectivescorrespond to different non exclusive sets of objects and relations that are informative on a topic regarding specific aspects.

Flexible querying and data processing

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

2 perspective-operations to expose the topic knowledge nuances • Criteria of interest specification• Controlled randomness injection

Multi-perspectives exploration

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

, dcterms:category, ?x

, dcterms:category, ?x

Classic similarity measure

, dcterms:category, ?a | ?b | ?c |...

, dcterms:category, ?a | ?b | ?c |...

Criteria spec. similarity

Multi-perspectives

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

, dcterms:category, ?x

, dcterms:category, ?x

Classic similarity measure

, dcterms:category, ?a | ?b | ?c |...

, dcterms:category, ?a | ?b | ?c |...

Criteria spec. similarity

Multi-perspectives

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Chosen level of randomness

Multi-perspectives

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Chosen level of randomness

Multi-perspectives

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Plan

• How can we discover and rank linked resources to be explored starting from the user topic(s) of interest?

• How to address remote linked data source for this selection?

• How to optimize such data based exploration approach at the interaction level?

• How to evaluate such exploratory search systems?

• State-of-the-art review

• Conclusion

• Contribution 1: several exploratory search algorithms optimized for linked datasets

• Contribution 2: their efficient execution over distant linked datasets

• Contribution 3: a novel interaction model and its application in the Discovery Hub web application

• Contribution 4: multiple users’ evaluations of the Discovery Hub results

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

• Computing the results from distant datasets• Computing the results on run-time

Algorithm execution - challenge

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Remote LOD sourceComputation at query-timefrom distant linked datasets

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

• Dbpedia 3.7• 3.4 million resources• 270 million triples

• Spread in both direction• Categories based similarity

Settings

Mirizzi and al., 2010

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Import size

Result quality Iterations

Trade-off

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Algorithm studied on 100.000 representativeDBpedia queries obtained with a random walker

J. Leskovec and C. Faloutsos. Sampling from large graphs, 2006

Algorithm behavior analysis

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Analysis - Number of iterations

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Analysis - Number of iterations

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Analysis – Triples imported

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Analysis – Triples imported

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Remote LOD sourceComputation at query-time

?

? ?

??

Analysis – Convergence polycentric

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Classic – top 5 artists

« French / not impressonist » criteria specification – top 5 artists

« Not French / Impressonist » criteria specification – top 5 artists

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Discovery Hub

Local Kgram instance

Data source selection

fr.dbpedia.org/sparql

it.dbpedia.org/sparql

de.dbpedia.org/sparql

es.dbpedia.org/sparql

dbpedia.org/sparql

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Discovery Hub

Local Kgram instance

Data source selection

fr.dbpedia.org/sparql

it.dbpedia.org/sparql

de.dbpedia.org/sparql

es.dbpedia.org/sparql

dbpedia.org/sparql

Analysis: 700 queries from the Discovery Hub

query-log

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Discovery Hub

Local Kgram instance

Data source selection

fr.dbpedia.org/sparql

it.dbpedia.org/sparql

de.dbpedia.org/sparql

es.dbpedia.org/sparql

dbpedia.org/sparql

Analysis: 700 queries from the Discovery Hub

query-log

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Analysis on random graphs

Monocentric algorithm behavior analyzed on 3700+ random graphs having diverse characteristics

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Analysis on random graphs

Monocentric algorithm behavior analyzed on 3700+ random graphs having diverse characteristics

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Plan

• How can we discover and rank linked resources to be explored starting from the user topic(s) of interest?

• How to address remote linked data source for this selection?

• How to optimize such data based exploration approach at the interaction level?

• How to evaluate such exploratory search systems?

• State-of-the-art review

• Conclusion

• Contribution 1: several exploratory search algorithms optimized for linked datasets

• Contribution 2: their efficient execution over distant linked datasets

• Contribution 3: a novel interaction model and its application in the Discovery Hub web application

• Contribution 4: multiple users’ evaluations of the Discovery Hub results

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

• Build an intuitive and supportive interface• Favor the users engagement

Interactions - challenge

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Application design http://discoveryhub.co

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Application design http://discoveryhub.co

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

http://discoveryhub.co

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED. Examples of evolutions

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED. Examples of evolutions

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Favoris

Nouvelle recherche TEMPS

Debut test Free Jazz 24s

Free improvisation 33s

(fiche) Avant-garde 47s

John Coltrane (vidéo) 1min 28

Marc Ribot 2min11

(fiche) experimental music 2min18 2min23

Krautrock 2min31

(fiche) Progressive rock 2min37 2min39

Red (King Crimson album) 2m52 2min59

King Crimson 3min05

(fiche) Jazz fusion 3min18

(fiche) Free Jazz 3min32 3min54

Sun Ra 4min18

(fiche) Hard bop 4min41 4min47

Charles Mingus (vidéo) 5min29

(fiche) Third Stream (vidéo) 6min20

Bebop 7min19

Modal jazz 7min26

(fiche) Saxophone 7min51 7min55

Mel Collins

21st Century Schizoid Band

Crimson Jazz Trio

(fiche)King Crimson

(fiche)Robert Fripp

Miles Davis

Thelonious Monk

(fiche) Blue Note Record

McCoy Tyner

(fiche) Modal Jazz

(fiche) Jazz

Chick Corea

(fiche) Jazz Fusion

Return to Forever

Mahavishnu Orchestra

Shakti (band)

U.Srinivas

Bela Fleck

Flecktones

John McLaughlin (musician)

Dixie Dregs

FICHE Dixie Degs

T Lavitz

Jordan Rudess

Behold… The Arctopus

(fiche) Avant-garde metal

Unexpected

FICHE unexpected

Dream Theater

King Crimson

(fiche) Jazz fusion

King Crimson

Tony Levin

(fiche) Anderson Bruford Wakeman Howe

(fiche) Rike Wakeman (vidéo)

Fin test

Ongoing work – Palagi, Giboin and al.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Plan

• How can we discover and rank linked resources to be explored starting from the user topic(s) of interest?

• How to address remote linked data source for this selection?

• How to optimize such data based exploration approach at the interaction level?

• How to evaluate such exploratory search systems?

• State-of-the-art review

• Conclusion

• Contribution 1: several exploratory search algorithms optimized for linked datasets

• Contribution 2: their efficient execution over distant linked datasets

• Contribution 3: a novel interaction model and its application in the Discovery Hub web application

• Contribution 4: multiple users’ evaluations of the Discovery Hub results

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

• Designing exploratory search evaluation protocols• Evaluating all the Discovery Hub algorithms

Evaluation - challenge

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Using

Cases

Compared to sSVM [Mirizzi and al., 2010]

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Hypotheses

1. The MSSA algorithm gives results at least as relevant as the sVSM one, even if it is not domain-optimized

2. The MSSA algorithm has less degradation than the sVSM algorithm

3. There is a greater chance that the results are less relevant but newer to

users at the end of the lists.

15 participants, questions

1. With the film A, I think I will live a similar cinematographic experience as with the film B ? Strongly agree, agree, disagree, strongly disagree

2. You and the film A ? Seen, known but not seen, unknown

Monocentric evaluations

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Very relevant

Not relevant at all

Monocentric evaluations

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Very relevant

Not relevant at all

Hypothesis 1: The SSA algorithm gives results at least as relevant as the sVSM one

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Very relevant

Not relevant at all

Known

Not known

Hypothesis 2: The SSA algorithm has less degradation than the sVSM one.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Very relevant

Not relevant at all

Known

Not known

Hypothesis 3: There is a greater chance that the results are less relevant but newer to users at the end of the lists.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Very relevant

Not relevant at all

Hypothesis 4: The explanatory features increase the users’ overall judgments positivity.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Random combination

Evaluated using

dbpedia:A dbpedia:B dbpedia:C dbpedia:D

= = = =

dbpedia:A & dbpedia:D

Hypothesis 5: the composite-query results

are interesting to the users

Hypothesis 6: a consequent proportion of the results are unexpected; they favor discoveries

Question 1: The result interests me: [Strongly

agree, agree, disagree, strongly disagree].

Question 2: The result is unexpected: [strongly

agree, agree, disagree, strongly disagree]

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

•61.6% of the results were rated as strongly relevant or relevant by the participants.

•65% of the results were rated as strongly unexpected or unexpected.

•35.4% of the results were rated both as strongly relevant or relevant and strongly unexpected or unexpected.

Hypothesis 5: the composite-query results are interesting to the users

Hypothesis 6: a consequent proportion of the results are unexpected

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Helpful

Not helpful at all

Hypothesis 7: The explanatory features help the users to understand the results

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Using

Information visualization

Serge Gainsbourg

Algorithm variantscompared:

• Basis• Criteria• Randomized - 0.5• Randomized - 1

Advanced querying evaluation

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Hypotheses

• Hypothesis 8: The users who specify their

criteria of interest find the results of the search

more relevant

• Hypothesis 9: The users who specify their

criteria of interest do not find the results less

novel

• Hypothesis 10: The stronger is the level of

randomness the more surprising the results are

for the users.

• Hypothesis 11: Even if the level of surprise is

high, the majority of the top results are still

relevant to the users.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Questions

1. Is the result in itself is surprising?

2. Is the relation between the result and the

query is surprising?

3. Is the result interesting?

4. Is the result too distant from the topic

searched?

Very close 1-2-3-4 Too distant.

Not agree at all 1-2-3-4 Totally agree

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Plan

• How can we discover and rank linked resources to be explored starting from the user topic(s) of interest?

• How to address remote linked data source for this selection?

• How to optimize such data based exploration approach at the interaction level?

• How to evaluate such exploratory search systems?

• State-of-the-art review

• Conclusion

• Contribution 1: several exploratory search algorithms optimized for linked datasets

• Contribution 2: their efficient implementation over distant linked datasets

• Contribution 3: a novel interaction model and its application in the Discovery Hub web application

• Contribution 4: multiple users’ evaluations of the Discovery Hub results

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

• Contribution 1: four exploratory search algorithms leveraging the linked data semantics

Monocentric

Polycentric

Criteria variant

Randomized variant

Contribution summary

• How can we discover and rank linked resources to be explored starting from the user topic(s) of interest?

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

• Contribution 1: four exploratory search algorithms leveraging the linked data semantics

Monocentric

Polycentric

Criteria variant

Randomized variant

Contribution summary

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

• How to address remote linked data source for this selection?

Contribution summary

• Contribution 2: efficient algorithm implementation over distant linked datasets

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

• How to optimize such data based exploration approach at the interaction level?

Contribution summary

• Contribution 3: a novel interaction model and its application in the Discovery Hub web application

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

• How to evaluate such exploratory search systems?

• Contribution 4: multiple users’ evaluations of the Discovery Hub results

3 user evaluations

Contribution summary

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Short-term improvements

-Facet and collection-levels interactions

-Collaborative exploratory search

-Post-query re-ranking

-New query-means

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Long-term perspectives

- Negative search

- Adaptive queries recommendation

- Collection-based knowledge creation

- On-the-fly distributed data selection

- Massive use and commercial success

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Other achievements

7 other publications in the fields of social networks and web sciences, sciences, including Web Sciences 2012Winner of the Challenge Jeunes Pousses 2013, in teamSeveral interventions during technological events

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Other achievements

7 other publications in the fields of social networks and web sciences, sciences, including Web Sciences 2012Winner of the Challenge Jeunes Pousses 2013, in teamSeveral interventions during technological events

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Other achievements

7 other publications in the fields of social networks and web sciences, sciences, including Web Sciences 2012Winner of the Challenge Jeunes Pousses 2013, in teamSeveral interventions during technological events

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Other achievements

7 other publications in the fields of social networks and web sciences, sciences, including Web Sciences 2012Winner of the Challenge Jeunes Pousses 2013, in teamSeveral interventions during technological events

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

• Nicolas Marie, Fabien Gandon. Demonstration of multi-perspective exploratory searchwith the Discovery Hub web application, ISWC2014, Riva Del Garda, Italy (demo)

• Nicolas Marie, Fabien Gandon. Survey of linked data based exploration systems, IESD2014, Riva Del Garda, Italy (long paper)

• Nicolas Marie, Fabien Gandon, Alain Giboin, Emilie Palagi. Exploratory search on topics through different perspectives with DBpedia, Semantics 2014, Leipzig, Germany (long paper)

• Nicolas Marie, Fabien Gandon, Myriam Ribière, Florentin Rodio. Discovery Hub: on-the-flylinked data exploratory search, I-Semantics 2013, TU Graz, Austria (long paper)

• Nicolas Marie, Fabien Gandon, Damien Legrand, Myriam Ribière. Exploratory search on the top of DBpedia chapters with the Discovery Hub application (demo + poster)

• Nicolas Marie, Olivier Corby, Fabien Gandon, Myriam Ribière. Composite interests’ exploration thanks to on-the-fly linked data spreading activation, Hypertext 2013, Paris (long paper)

16% acceptance rate

ESWC Best demonstration award

Thank you ! Questions ?

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Annexes

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

Tomaso Di Noia and al. Linked Open Data to support

Content-based Recommender Systems. I-Semantics

2012

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

[Passant, 2010]

[Mirizzi, 2010]

Wateilonis, 2010

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

p.223

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

p.216

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

COPYRIGHT © 2011 ALCATEL-LUCENT. ALL RIGHTS RESERVED.