30
-1- Philipp Heim , Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1 , Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization and Interactive Systems Group (VIS), University of Stuttgart, Germany 2 Interactive Systems and Interaction Design, University of Duisburg-Essen, Germany ESWC 2010 7 th Extended Semantic Web Conference Heraklion, Greece – June 1-3, 2010

-1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

Embed Size (px)

Citation preview

Page 1: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-1- Philipp Heim, Thomas Ertl, Jürgen Ziegler

Facet Graphs: Complex Semantic Querying Made Easy

Philipp Heim1, Thomas Ertl1 and Jürgen Ziegler2

1 Visualization and Interactive Systems Group (VIS), University of Stuttgart, Germany2 Interactive Systems and Interaction Design, University of Duisburg-Essen, Germany

ESWC 20107th Extended Semantic Web ConferenceHeraklion, Greece – June 1-3, 2010

Page 2: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-2- Philipp Heim, Thomas Ertl, Jürgen Ziegler

Outline

1. How to access information in the Semantic Web?

2. Faceted Search: An Introduction

3. Faceted Search in the Semantic Web

4. Facet Graphs

5. Evaluation

6. Discussion

7. Conclusion and Future Work

Page 3: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-3- Philipp Heim, Thomas Ertl, Jürgen Ziegler

Outline

1. How to access information in the Semantic Web?

2. Faceted Search: An Introduction

3. Faceted Search in the Semantic Web

4. Facet Graphs

5. Evaluation

6. Discussion

7. Conclusion and Future Work

Page 4: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-4- Philipp Heim, Thomas Ertl, Jürgen Ziegler

1. How to access information in the Semantic Web? Common Web:

Entering words in an input field (e.g. Google or Bing) Problem ambiguity:

Natural language is ambiguous!Finding the right information, however, requires the semantic of what should be searched to be specified by the user.

Solution:Artificial query languages like SPARQL that are uniquely defined. Access via SPARQL endpoints (e.g. DBpedia or the LOD cloud).

SELECT DISTINCT ?object ?label WHERE {

?object rdf:type <URI of Football player> . ?object rdfs:label ?label .

}

Page 5: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-5- Philipp Heim, Thomas Ertl, Jürgen Ziegler

1. How to access information in the Semantic Web? Problem required knowledge:

SPARQL requires the language to be learned by the user (rather a task for experts).

Solution:Intuitive graphical interfaces to express search queries that are semantically unique but do not require any extra knowledge.

Page 6: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-6- Philipp Heim, Thomas Ertl, Jürgen Ziegler

Outline

1. How to access information in the Semantic Web?

2. Faceted Search: An Introduction

3. Faceted Search in the Semantic Web

4. Facet Graphs

5. Evaluation

6. Discussion

7. Conclusion and Future Work

Page 7: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-7- Philipp Heim, Thomas Ertl, Jürgen Ziegler

2. Faceted Search: An Introduction

Example:

facets

facet(1) select

attribute

result set

(2) filter

(3) update

faceted search:

Page 8: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-8- Philipp Heim, Thomas Ertl, Jürgen Ziegler

2. Faceted Search: An Introduction

(1) select

Audiobooks

result set

(3) update

(2) filter(1) select

Audiobooks

number of results

facetcategory

facet count

facet attribute

Page 9: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-9- Philipp Heim, Thomas Ertl, Jürgen Ziegler

2. Faceted Search: An Introduction

Advantages:- Facets and their attributes are given (reduced effort)

- Attributes are categoriezed (understanding)

- No facet attribute can lead to an empty result set

- Resulting number of resultsis shown in advance

- Rapid update of result set (dynamic queries)

- Selected attributes are shown and can be deselected

Page 10: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-10- Philipp Heim, Thomas Ertl, Jürgen Ziegler

Outline

1. How to access information in the Semantic Web?

2. Faceted Search: An Introduction

3. Faceted Search in the Semantic Web

4. Facet Graphs

5. Evaluation

6. Discussion

7. Conclusion and Future Work

Page 11: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-11- Philipp Heim, Thomas Ertl, Jürgen Ziegler

3. Faceted Search in the Semantic Web

mspace (Hearst et al. 2002: Finding the Flow in Web Site Search):

Disadvantages: 1.No facet count(number of results to expect)2.First order facets only (directly connected to the result set)

Page 12: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-12- Philipp Heim, Thomas Ertl, Jürgen Ziegler

3. Faceted Search in the Semantic Web

Parallax (Huynh and Karger 2009: Parallax and companion: Set-based

browsing for the Data Web):

Advantage:1.Hierarchical facets possible(second or higher order facets)

Disadvantages: 1.Hierarchy not visible2.Browsing required

Page 13: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-13- Philipp Heim, Thomas Ertl, Jürgen Ziegler

3. Faceted Search in the Semantic Web

Tabulator (Berners-Lee et al. 2008: Tabulator Redux: Browsing and

writing Linked Data) :

Advantages:1.Hierarchical facets2.Hierarchy on one page

Disadvantages: 1.Attributes get partitioned in different subtrees2.Redundant attributes

Page 14: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-14- Philipp Heim, Thomas Ertl, Jürgen Ziegler

Outline

1. How to access information in the Semantic Web?

2. Faceted Search: An Introduction

3. Faceted Search in the Semantic Web

4. Facet Graphs

5. Evaluation

6. Discussion

7. Conclusion and Future Work

Page 15: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-15- Philipp Heim, Thomas Ertl, Jürgen Ziegler

4. Facet Graphs Idea: Facets and result set are represented as nodes in a

graph visualization

Theme SubjectYearDecade Story Title

result set

facets

categories(labeled edges)

Page 16: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-16- Philipp Heim, Thomas Ertl, Jürgen Ziegler

4. Facet Graphs

How to extract facets from RDF data?

club2

club1club

type

type

result set

class objects

venue2

venue1

venue

type

type

ground

ground

venue3 type

facet

properties

objects

class

facetresult set

1. Defining the result set class(e.g. German football club)

2. Extracting the facets(e.g. ground:Venue)

Page 17: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-17- Philipp Heim, Thomas Ertl, Jürgen Ziegler

4. Facet Graphs Prototypical implementation: gFacet

Page 18: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-18- Philipp Heim, Thomas Ertl, Jürgen Ziegler

4. Facet Graphs

gFacet:

1. Flash application (animation, independence, no installation)

2. SPARQL queries (standard, data set independent)

3. Open source (http://code.google.com/p/gfacet/)

General benefits of Facet Graphs:

1. Attributes for each facet are grouped into one node

2. All nodes are shown in a coherent presentation on one page

3. Semantic relations between the nodes are represented by labeled edges

4. Facets can be added and removed by the user (individual search interfaces)

Page 19: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-19- Philipp Heim, Thomas Ertl, Jürgen Ziegler

4. Facet Graphs

Hierarchical facets:

color-coding

Page 20: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-20- Philipp Heim, Thomas Ertl, Jürgen Ziegler

4. Facet Graphs

Multiple selections:

Page 21: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-21- Philipp Heim, Thomas Ertl, Jürgen Ziegler

4. Facet Graphs

Pivot operation:While using gFacet, users may change their minds about what they want to search.

Page 22: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-22- Philipp Heim, Thomas Ertl, Jürgen Ziegler

Outline

1. How to access information in the Semantic Web?

2. Faceted Search: An Introduction

3. Faceted Search in the Semantic Web

4. Facet Graphs

5. Evaluation

6. Discussion

7. Conclusion and Future Work

Page 23: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-23- Philipp Heim, Thomas Ertl, Jürgen Ziegler

5. Evaluation

Comparative study of Parallax and gFacet:

Three different types of tasks:

1. Find two players who are playing for a certain club.

2. Find two cities where players who are playing for a certain club are born.

3. Find one player who is playing for a certain club and is born in a certain city.

VS.

Page 24: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-24- Philipp Heim, Thomas Ertl, Jürgen Ziegler

5. Evaluation

Results:

Three different types of tasks:

1. Find two players who are playing for a certain club.

2. Find two cities where players who are playing for a certain club are born.

3. Find one player who is playing for a certain club and is born in a certain city.

comments to the statements ‘It was difficult to understand the relations

between the information’

Page 25: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-25- Philipp Heim, Thomas Ertl, Jürgen Ziegler

Outline

1. How to access information in the Semantic Web?

2. Faceted Search: An Introduction

3. Faceted Search in the Semantic Web

4. Facet Graphs

5. Evaluation

6. Discussion

7. Conclusion and Future Work

Page 26: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-26- Philipp Heim, Thomas Ertl, Jürgen Ziegler

6. Discussion

Are existing faceted search tools capable of supporting the Information Seeking Process (ISP) (Kuhlthau 1988) in the Semantic Web?

R3.5: Zooming functionalities that are capable of showing information in different levels of detail.

Page 27: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-27- Philipp Heim, Thomas Ertl, Jürgen Ziegler

Outline

1. How to access information in the Semantic Web?

2. Faceted Search: An Introduction

3. Faceted Search in the Semantic Web

4. Facet Graphs

5. Evaluation

6. Discussion

7. Conclusion and Future Work

Page 28: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-28- Philipp Heim, Thomas Ertl, Jürgen Ziegler

7. Conclusion and Future Work

Facet Graphs:- Facets as nodes in a graph visualization: Direct representation of

relationships between facets

- Connected representation on one page

- Hierarchical facets

- Color-coding: Understanding and traceing filtering effects

- Personalized search interface (add/remove facets)

gFacet:- Proof of concept

- Can query arbitrary SPARQL endpoints (e.g. DBpedia)

- Comparative study: Especially applicable for complex queries (Semanitc Web)

- However: Control remains a challenging task!

Page 29: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-29- Philipp Heim, Thomas Ertl, Jürgen Ziegler

7. Conclusion and Future Work

Future work:- Zooming functionalities + focus and context technique:

to handle massive amounts of facets in one graph

- Saving search interfaces: to share especially helpful combinations of facets to lower the barrier for new users

German football clubs and their players?

Do you want to load a search interface to explore:

Publications aboutthe Semantic Web?

US cities?

Page 30: -1- Philipp Heim, Thomas Ertl, Jürgen Ziegler Facet Graphs: Complex Semantic Querying Made Easy Philipp Heim 1, Thomas Ertl 1 and Jürgen Ziegler 2 1 Visualization

-30- Philipp Heim, Thomas Ertl, Jürgen Ziegler

Thank you for your attention.

visit gFacet at http://gfacet.semanticweb.org