Browsing-Oriented Semantic Faceted Search

Browsing-oriented Semantic Faceted Search
Andreas Wagner, Günter Ladwig and Duc Thanh Tran

Institute of Applied Informatics and Formal Description Methods (AIFB)

KIT – University of the State of Baden-Wuerttemberg and
National Research Center of the Helmholtz Association www.kit.edu

Agenda

Introduction and Motivation
Information Needs
Faceted Search Concepts
Contributions

Browsing-oriented Faceted Search …
Browsing-oriented Facet and Facet Value Spaces
Browsing-oriented Facet Ranking

Evaluation Results

2 Andreas Wagner, Günter Ladwig, Duc Thanh Tran Institute of Applied Informatics and Formal
Description Methods (AIFB)

INTRODUCTION & MOTIVATION


User Information Need
User Need Information

... ... ...

Example 1
Susan is a novice computer science student.
She is wishes to find information about work
of prestigious computer scientists. Fuzzy Need

Example 2 Precise Need
Susan is a grad-student. She is wishes to find
information about Knuth’s first book Funda- See, e.g., [1,2].
mental Algorithms.

Faceted Search

Faceted
... ... ...
Search

Faceted Search is…
a paradigm allowing users to
explore a data source through
fluent interaction of refinement and
expansion.

See, e.g., [3].


Faceted Search in a Semantic Web Context

Query and Data Model
Data model is a graph
Query model based on basic graph-patterns

Facet Model
Facets with are edge labels of (one ore more) node(s) contained in
the current result set
Nodes of these edges are facet values


Faceted Search in a Semantic Web Context

Facet Operations
Focus Selection
Query Result
Refinement Modifaction Exploration
Expansion

Focus Selection
Initial Query
knows name
?x ?y “Knuth“

works at Expansion
Refinement

“Stanford University“


Our Contributions

Browsing-oriented Faceted Search
Fuzzy information needs require different kinds of facets, and a
different grouping of facets. Strong need for browsing.
State-of-the-art focuses mainly on precise needs (or target a
generic scenario). See, e.g., [4,5,6,7].

Example 1
Susan is a novice computer science student.
She is wishes to find information about work
of prestigious computer scientists. Fuzzy Need

Example 2 Precise Need
Susan is a grad-student. She is wishes to find
information about Knuth’s first book Funda-
mental Algorithms.

8 Institute of Applied Informatics and Formal

Our Contributions Fuzzy Need

Challenges? Precise Need

How to handle high-dimensional facet values for browsing?
How to handle large facet value sets for browsing?
Facet & facet value ranking well-suited for browsing?

Contributions
State-of-the-Art?
See, e.g., [5,8,9].

Restricted Facet (Facet Value) Grouping Browsing-
Grouping focuses on facets only, no (flexible) means for oriented Facet
grouping large facet value spaces. (Value) Spaces
Search-oriented Facet Ranking
Browsing-oriented
Existing ranking approaches assume a precise Facet Ranking
information need (or are generic).
See, e.g., [4,5,6,7].

BROWSING-ORIENTED
FACETED SEARCH


Challenges?



Browsing-oriented Facet and Facet Value
Spaces

Facet Tree (FT)
A facet tree (i.e., hierarchical grouping of facets) is derived from
nodes and edges of the data graph, which are reachable from the
result set.
See, e.g., [8,9,10].
Result Set
(Set of Computer mary
Science Professors)
ann
P2 name
paul
70
P1

works at 150
P4 U2
U1 age 250
P3 U4
Focus 300
U3
12
Selection Andreas Wagner, Günter Ladwig, Duc Thanh Tran Institute of Applied Informatics and Formal

Browsing-oriented Facet and Facet Value Other Facet Operations
Spaces Focus Selection
Refinement
Facet Operation: Browsing Expansion
Browsing consists of (multiple) facet selections. However, facets
selected during browsing are not evaluated, i.e., the underlying
query does not change and thus the result set is not modified.

Result Set
(Set of Computer
Science Professors)
[ann − paul] ...? Compact,
P2 name intensional
representation of
P1 the facet space.

P4 works at
?u age [70− 300] ...?
P3
Focus
13
Selection Andreas Wagner, Günter Ladwig, Duc Thanh Tran Institute of Applied Informatics and Formal

Spaces

Extended Facet Tree
Employ clustering to extend the facet tree. Leaf nodes in the facet
tree containing more data values than a given threshold are
clustered, resulting in a set of data value trees.

Result Set
(Set of Computer ann
Science Professors)
[ann − paul] mary
P2 name [mary - paul]
paul
P1

P4 works at
Compact,
?u age [70− 300] ...
intensional
P3 representation of
the facet and
14 Andreas Wagner, Günter Ladwig, Duc Thanh Tran facet value space.
Institute of Applied Informatics and Formal

Spaces

Extended Facet Tree
We currently employ a simply divisive, hierarchical clustering.
Depending on the application setting, other clustering algorithms may
be better suited
Highlight outliers
Highlight expected values
...

Benefits
Entire facet and facet value space is (compactly) represented
User may drill-down, depending on how precise (fuzzy) her need is

Drawbacks
More interaction is needed, as facet tree is more fine-grained
See evaluation


Challenges?




A browsing-oriented ranking function incorporates different
notions (via their metrics): Small steps, uniform steps,
comprehensible result segments.
Notions (metrics) influence each other.
Depending on the application scenario, only a subset of the
notions (metrics) may suffice.

Small Steps Uniform Steps

Metric Metric
Metric Metric

Comprehensible
Result
Segments
Metric
Metric


Idea
For ranking a facet f, consider the facet and facet value space that
can be reached via f and result set modifications, which can be
performed via facet paths originating from f .

Use the extended facet tree, associated with a facet f, for assessing
the browsing quality of f.

Facet Extended Facet Tree

ann
name [ann − paul] mary
[mary - paul]
paul


Browsing-oriented Facet Ranking – Intuition

Idea
Via small result modifications, users get to know the Small Steps
result set bit by bit.
Small changes can be comprehended more easily by
users.
Metrics
Maximum Height
The height of the extended FT, directly reflects the maximum number of
possible facet operations.

Minimum Branching Factor
Trees with small branching factor lead to smaller result
modifications, as such trees tend to be higher.
A small branching factor reflects a small number of possible user
decisions.



Idea Uniform
We consider query modifications to be non-uniform, Steps
when they have varying impacts on the result set size.
When browsing, it is hard for users to choose
between non-uniform query modifications.
Such query modifications can be confusing and may
lead to irrelevant results.
Metrics
Height Balance
The extended FT is perfectly height balanced, when all leaves
are of equal edge distance to the root.
Facet Value Set & Binding Segment Size Balance
Balance the size balance w.r.t. facet value sets and binding set
segments, which may be reached via the extended FT.



Example: Binding Segment Size Balance
Uniform
Steps
Facet: name Facet Path: works at, age
[ann-paul]

P1 P2 P3 P4 P1 P2 P3 P4

ann [mary-paul] [70-300]

P1 P2 P3 P4 P1 P2 P3 P4
P1
[250-300] [70-150]
mary
P1 P3 P2 P4
P2 P3 P1 P4
70 150
paul
Binding Segment Tree 250 P1 P3 P2 P4

21 Andreas Wagner, Günter Ladwig, Duc Thanh Tran 300 Institute of Applied Informatics and Formal


Idea Comprehensible
Result
For users who are unfamiliar with a result set, it is
Segments
important that a facet operation leads to obvious and
comprehensible result modifications.

Metrics
Binding Distinguishability
A facet has a high distinguishability, when it leads to facet values
that precisely identify variable bindings. See [4].
Minimal Binding Segment Overlap
Binding segments with minimal overlaps are preferred to ensure
that facet operations along a facet tree lead to different result
modifications.


EVALUATION


Evaluation – Setting

We conducted a task-based user evaluation.

Participants
24 participants
Mixed group: 18 participants had a computer science background, 6
had non-technical background

Tasks: 24 tasks were chosen by domain experts and comprised
both precise and fuzzy information needs.

Data: we used the (complete) DBpedia dataset [11]

System: based on Information Workbench [12]


Evaluation – Extended Facet Tree

Tasks
Four tasks (C1-C4) for investigating the effects of our data value trees
Eight complex browsing tasks (B1-B8), to assess the quality of
browsing based on the facet tree

Baseline
System with a flat list of facets and no data value trees
We designed clustering (C) and browsing (B) tasks in a way, that we
were able to compare the effects of data value clustering en- or
disabled and facets grouped in lists or trees

How effective and how efficient is the extended facet tree (com-
pared to the baseline)?


Evaluation – Extended Facet Tree

Results

Results suggest that the use of our extended facet tree improves the
efficiency and effectiveness of the task completion, concerning
complex, fuzzy tasks.
Search is more efficient and equally effective, with regard to precise
and simple needs only.

Evaluation – Browsing-oriented Ranking

Tasks
Find (F) tasks comprise of 8 tasks (F1-F8), which involve precise and
fuzzy information needs. Goal is to find a concrete item of interest.
Explore (E) tasks comprises of 4 tasks (E1-E4), where users had to
explore a result set (fuzzy need), i.e., find outliers, interesting or
strange results.

Baseline: a system employing search-oriented ranking.

How effective and how efficient is the browsing-oriented ranking
(compared to the baseline)?


Evaluation – Browsing-oriented Ranking

Results

While browsing-oriented ranking might not provide an efficient way to
an item of interest, it is suitable for scenarios with no precise need
and large result sets to be explored.


CONCLUSION


Conclusion & Future Work

Current faceted search approaches imply a precise information
need (or are generic) and thus, focus on the search paradigm.
We target the browsing paradigm, where users only vaguely know
the domain or item of interest.
Our solution outperformed the state-of-the-art w.r.t. fuzzy infor-
mation needs.

Future Work …
Efficiency aspects?
When to switch between search- and browsing-oriented ranking?


REFERENCES


References
1. G. Marchionini and B. Shneiderman. Finding facts vs. browsing knowledge in hy-
pertext systems. Computer, 21(1):70–80, 1988.
2. G. Marchionini. Exploratory search: from finding to understanding. Commun. ACM,
49(4):41–46, 2006.
3. M. Hearst, K. Swearingen, K. Li, and K.-P. Yee. Faceted metadata for image
search and browsing. In CHI, pages 401–408. ACM, 2003.
4. S. Basu Roy, H. Wang, G. Das, U. Nambiar, and M. Mohania. Minimum-effort
driven dynamic faceted search in structured databases. In CIKM, pages 13–22.
ACM, 2008.
5. W. Dakka, P. G. Ipeirotis, and K. R. Wood. Automatic construction of multifaceted
browsing interfaces. In CIKM, pages 768–775. ACM, 2005.
6. D. Dash, J. Rao, N. Megiddo, A. Ailamaki, and G. Lohman. Dynamic faceted
search for discovery-driven analysis. In CIKM, pages 3–12. ACM, 2008.
7. J. Koren, Y. Zhang, and X. Liu. Personalized interactive faceted search. In WWW,
pages 477–486. ACM, 2008.
8. P. Heim, T. Ertl, and J. Ziegler. Facet graphs: Complex semantic querying made
easy. In ESWC, pages 288–302. Springer, 2010.
9. D. F. Huynh and D. R. Karger. Parallax and companion: Set-based browsing for
the data web. In WWW, 2009.

References
10. T. Berners-Lee, Y. Chen, L. Chilton, D. Connolly, R. Dhanaraj, J. Hollenbach, A.
Lerer, and D. Sheets. Tabulator: Exploring and analyzing linked data on the se-
mantic web. In Proceedings of the 3rd International Semantic Web User Interaction
Workshop, 2006.
11. C. Bizer, J. Lehmann, G. Kobilarov, S. Auer, C. Becker, R. Cyganiak, and S.
Hellmann. Dbpedia - a crystallization point for the web of data. Journal of Web
Semantics, 7(3):154–165, 2009.
12. http://iwb.fluidops.net/


BACKUP SLIDES


Faceted Search – Terminology
What are facets?
Conceptual dimensions of the current result set.
What are facet values?
Values of conceptual dimensions.

Dimension

Search Result Facets Facet Values


Spaces

Example: Facet Tree and Browsing


Spaces
Example: Extended Facet Tree


Browsing-Oriented Semantic Faceted Search

Recommandé

Recommandé

Contenu connexe

Tendances

Tendances (20)

Similaire à Browsing-Oriented Semantic Faceted Search

Similaire à Browsing-Oriented Semantic Faceted Search (20)

Dernier

Dernier (20)

Browsing-Oriented Semantic Faceted Search