feza baskaya – anne kakkonen - university of tampere - 20071 second international seminar on...
TRANSCRIPT
![Page 1: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/1.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 1
Second International Seminar on Subject Access to Information, Helsinki
30th November 2007
QUCCOOQUCCOOQuQuery ery CConstruonstrucction with tion with OOntntoologylogyOntology-based Search InterfaceOntology-based Search Interface
Feza BASKAYAAnne KAKKONEN
University of TampereDepartment of Information Studies
![Page 2: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/2.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 2
OutlineOutline
1. Background 2. Ontologies 3. Quccoo: Searching Unannotated
Collections through Ontologies 4. ShOE: Creating ontologies 5. Discussion, Conclusion
![Page 3: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/3.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 3
1. Background1. Background
Vast online information environments billions of digital documents many different natural languages distributed document production and
publication: no generally agreed rules general lack of control in the process much spam and other unwanted material
![Page 4: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/4.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 4
Background, 2Background, 2
Vocabulary mismatch hard to guess the best search keys; leads to
loss of search effectiveness especially in foreign languages hard to know word forms, compound
treatment Other problems – depending on one’s
search environment collection dependency, metadata dependency engine and query language dependency
![Page 5: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/5.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 5
2. Ontologies2. Ontologies Ontologies model semantics
concepts rich relationships support inference application means resource annotation closely related to thesauri
Belief: ontologies can solve the vocabulary problem represents the semantics of resources (documents)
better than pure natural language retrieval becomes correct and accurate desired: a universal world model, and a controlled
language for description and reasoning about this model
![Page 6: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/6.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 6
Issues in Classification and Issues in Classification and IndexingIndexing
Index languages - modeling - coverage, viewpoint maintenance - ageing, cost
Indexing - specificity, exhaustivity, consistency cost - where paid, who pays? The over-specificity the devices created
often lead to poor recall and thus they were soon mostly abandoned
![Page 7: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/7.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 7
Any Room for Ontologies?Any Room for Ontologies? Should one thus discard ontologies?
or other vocabulary control tools? In practice, realism tells us that
there will never be a comprehensive & up-to-date ontology – cf. UDC, which had large development community support
no one will annotate for free, for ever & consistently no one can do that exhaustively and from many viewpoints
emerging, e.g., in future in fact, less than 0.3% of web pages had Dublin Core
metadata (Rasmussen 2003) There is no alternative to searching unannotated
collections automatic annotation does not solve the problem - if one
aims at the good semantics required in the Semantic Web
![Page 8: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/8.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 8
3. Searching Unannotated 3. Searching Unannotated Collections through OntologiesCollections through Ontologies
Searching ontologies can provide conceptual organization support direct access to textual content
translate between concepts and textual variationtranslate between natural languageshide search engines / query languagesmay support other media / structures / features
be light-weight, narrow, and no world modelspersonal, group or small community supportversions, mutually incoherent, easily modifiableeasily disposable, perhaps tradable
![Page 9: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/9.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 9
Searching through OntologiesSearching through Ontologies
Need to solve the vocabulary problem from concepts to textual expressions three layers:
Concepts - for user interactionExpressions - for system useStrings to match - for system use
Need to provide a handy concept browser and query constructor
![Page 10: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/10.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 10
Three levelsThree levels Forest industry
forest industry paper industry saw mill ...
pl(saw, mill) al(industry) pl(paperi, tehdas)
Conceptual level
Linguistic level
String level
Concepts
Search keys
Character strings
Search termsCodes
&abbreviations
Search words
String constantsString patterns
![Page 11: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/11.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 11
QUCCOO: PrinciplesQUCCOO: Principles QUCCOO: QUery ConstruCtion with OntOlogies
for direct content access Based on the three levels … Aims to provide independence of …
expression variability (nutraceutical?) natural language (French?) collection (intranet, Web ,…) indexing (lemmatization, compounds?) availability of metadata & world model engine & query language (Lemur, Trip, Google, …)
You just select your concepts, targets and go! Point, click and go
![Page 12: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/12.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 12
QUCCOO: statusQUCCOO: status Web application, uses state-of-the-art
Servlet technology Supports diverse full-text database
engines (Trip, InQuery, etc.) as well web search engines (e.g., Google)
Supports diverse collections Intuitive; simple interface to access
information Supports multilingual search and various
index types
![Page 13: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/13.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 13
QUCCOOQUCCOO: Architecture: Architecture
search
Conceptual model request
Ontology server
PostgressRDBMS
PostgressRDBMS
Javaservlet
Javaservlet
Document servers
Request Concepts
Concepts
Query
Expanded QueryResultsResults
Client- Applet Server-Servlet
KBKB
DDBInQuery
DDBInQuery
DDBTRIP
DDBTRIP
DDBLemur
DDBLemur
WebGoogle
Results1. snippet2. snippet
...
n. snippet
c
ccccc
ccc
own keys
query
![Page 14: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/14.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 14
Quccoo Quccoo - - interfaceinterface
Ontology Tree
Concepts given by user
Search box
Options button
Search button
![Page 15: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/15.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 15
Quccoo Quccoo - - interfaceinterface
![Page 16: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/16.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 16
Quccoo Quccoo - - interfaceinterface
Search Engine Selection
![Page 17: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/17.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 17
Quccoo Quccoo - - interfaceinterface
LiberalitySelection
![Page 18: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/18.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 18
Quccoo Quccoo - - interfaceinterface
Expansion Level Selection
![Page 19: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/19.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 19
Quccoo Quccoo - - interfaceinterface
Ontology Selection
![Page 20: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/20.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 20
Quccoo Quccoo - - interfaceinterface
Database Selection
![Page 21: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/21.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 21
Quccoo Quccoo - - interfaceinterface
Query Lang. Selection
![Page 22: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/22.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 22
Quccoo Quccoo - - interfaceinterface
User ID
![Page 23: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/23.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 23
Quccoo Quccoo - - interfaceinterface
Ontology Lang. Selection
![Page 24: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/24.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 24
Quccoo Quccoo - - interfaceinterface
Query result page
![Page 25: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/25.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 25
Quccoo Quccoo - - interfaceinterface
Trip Database Engine results
Ontology in Finnish
![Page 26: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/26.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 26
Quccoo Quccoo - - interfaceinterface
Inquery Database Engine Results
Ontology in Finnish
![Page 27: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/27.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 27
Quccoo Quccoo - - interfaceinterface
Google Results in Finnish
Ontology in Finnish
![Page 28: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/28.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 28
Quccoo Quccoo - - interfaceinterface
Google Results in English
Ontology in Finnish
![Page 29: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/29.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 29
Quccoo Quccoo - - interfaceinterface
Google Results in Swedish
Ontology in Finnish
![Page 30: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/30.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 31
Quccoo Quccoo - - interfaceinterface
Ontology in English
Google Results in English
![Page 31: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/31.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 32
Quccoo Quccoo - - interfaceinterface
Ontology in English
Google Results in Finnish
![Page 32: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/32.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 33
Quccoo Quccoo - - interfaceinterface
Extra keyword(s) added by user
![Page 33: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/33.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 34
4. ShOE: Creating ontologies4. ShOE: Creating ontologies Search ontology editor - for creating
ontologies supports the 3 layer architecture of QUCCOO intuitive; easy to learn and use automatic support for the human editor
Multilingual in many aspects GUI, User Interface language can be changed Concepts names can be edited/displayed on-
the-fly in different languages Expressions can be edited/displayed on-the-fly
in different languages.
![Page 34: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/34.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 36
ShOE - ShOE - Main windowMain window
Concept properties
Concept hierarchy tree
Concept description
Tabs
Expression window
Search field
![Page 35: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/35.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 37
6. Conclusion6. Conclusion ShOE and QUCCOO are one answer to
problems in semantic information access light-weight disposable search ontologies for
full content access independencies of:
collections (partially), indexing ways,availability of metadata / annotationschanges of needs, variability of ”world models” search engines, query languagesvocabulary variation and natural languages
a compromise, different from semantic annotation or indexing, with control at the user end
![Page 36: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/36.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 38
User testingUser testing
![Page 37: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/37.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 39
Cross-language Web searchCross-language Web search Test persons
40 students from the University of Tampere and Pirkanmaa polytechnic
Ontology Combination of two ontologies: Food concepts and
geographical concepts 2 interfaces
QUCCOO + interface without ontology (basic Google search)
4 simulated search tasks Two tasks with one interface and two with the other
![Page 38: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/38.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 40
AnalysisAnalysis
Log files Queries Relevance assessments (scale 0-4)
Questionnaires Opinions about ontology and Quccoo-
interface
![Page 39: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/39.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 41
Results: search successResults: search success
No significant difference between systems QUCCOO performed better when
strong query structure was needed (“alcoholic beverage”)
In most self-formulated queries no phrases were used
→ QUCCOO helps persons who are not used to formulate structured queries
![Page 40: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/40.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 42
Results: opinionsResults: opinions
”Structure of the ontology was logical”
”Finding search concepts needed in the tasks in ontology was easy”
”Using the ontology was effortless” 92 % agreed in all
![Page 41: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/41.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 43
Results: opinionsResults: opinions
32/40 thought that QUCCOO-interface was easier to use
32/40 liked QUCCOO better Why?
Helped users to clarify task topic and to find related search keys
Made cross-language search easy (in 80% of direct searches some dictionary was used to help query formulation)
![Page 42: Feza Baskaya – Anne Kakkonen - University of Tampere - 20071 Second International Seminar on Subject Access to Information, Helsinki 30th November 2007](https://reader037.vdocument.in/reader037/viewer/2022110204/56649c6e5503460f94920c3b/html5/thumbnails/42.jpg)
Feza Baskaya – Anne Kakkonen - University of Tampere - 2007 44
DiscussionDiscussion
Thank you!
Over to you ... questions?