The objective of this framework is to sustain and encourage the collaboration between different kind of experts for modeling domains and for providing a semantic representation of the it. Examples of experts are the Domain Experts (i.e. those that knows the domain but usually lacks the modelling skills), and the Knowledge Engineers (those that have the skills but have not a clear understanding of the domain). During this talk, I will present the last version of MoKi, the wiki-based tool designed for supporting such a framework and I will show how this tool has been customized and extended in several projects in order to face the different challenges raised by the usage of semantic representations in different domains.
Swan(sea) Song – personal research during my six years at Swansea ... and bey...
Collaborative Modeling of Processes and Ontologies with MoKi
1. Collaborative modeling of
processes and ontologies with MoKi
Mauro Dragoni
Fondazione Bruno Kessler (FBK)
Shape and Evolve Living Knowledge Unit (SHELL)
https://shell.fbk.eu/index.php/Mauro_Dragoni - dragoni@fbk.eu
Semantic MediaWiki Conference Fall, Berlin, Germany – October 30th, 2013
1
2. Outline of the presentation
Introduction
Architecture for collaborative conceptual modeling
in wikis
MoKi and some of its real usages
2
2
3. What is all about?
Develop a theoretical and practical framework that:
Supports the integrated modeling of Processes and Ontologies;
Fosters the collaboration between domain experts and knowledge
engineers.
WHY?
need of a comprehensive model which requires the description
of both the dynamic component (processes) and the static
component (ontology);
need for an agile collaboration between domain experts and
knowledge engineers. Need to actively involve the domain
experts in the modeling process.
3
5. The research vision - architecture
MoKi
Knowledge Engineer
the Modelling W iKi ---
Modelling tool
Domain expert
Knowledge Engineer
Domain expert
Model
Formal representation of
integrated processes and
ontologies
Theoretical framework
Architecture for collaborative
conceptual modeling
5
7. An architecture for collaborative
conceptual modeling in wikis
1.
One element
One page
each element of the model is represented by a page in the wiki;
Mountain
Concept “Mountain”
A mountain is a large landform that
stretches above the surrounding land in
a limited area usually in the form of a
peak. A mountain is generally steeper
than a hill.
The highest
mountain on earth is
the Mount Everest
7
7
8. An architecture for collaborative
conceptual modeling in wikis
2.
Unstructured and structured descriptions
each page contains both structured and unstructured content;
Mountain
A mountain is a large landform that
stretches above the surrounding land in
a limited area usually in the form of a
peak. A mountain is generally steeper
than a hill.
The highest
mountain on earth is
the Mount Everest
(unstructured content)
L andf or m
¬H i ll
¬P lai n
∀madeOf (E ar th
Rock)
∃hei ght. ≥ 2500
M ountai n(M t.E ver est)
M ountai n(M t.K i li manj ar o)
(structured content)
8
8
9. An architecture for collaborative
conceptual modeling in wikis
3.
Different views to access the model:
different views to support different modeling actors;
Mountain
A mountain is a large landform that
stretches above the surrounding land in
a limited area usually in the form of a
peak. A mountain is generally steeper
than a hill.
The highest
mountain on earth is
the Mount Everest
(unstructured view)
9
9
12.
Wiki-based modeling tool;
Supports the integrated modeling of Processes and
Ontologies;
Provides modeling support both for domain experts
and knowledge engineers, fostering the
collaboration between them;
Based on the architecture presented so far.
12
12
13. Different views for different roles
Unstructured
view
Semi-structured
Fully-structured
view
view
13
17. Further features: key concepts extraction
Automatic extraction of key-concepts from a text corpus.
Key-concepts are terms characterizing the domain represented
by the corpus;
Extraction performed with the KX (Keyphrase eXtraction) system;
Extracted list enriched with additional resources (e.g. WordNet)
and aligned with the ontology under construction/consideration.
Useful for several ontology modeling-related tasks:
Boosting ontology creation/extension;
Ontology terminological evaluation/ranking (based on ontology
metrics);
Ranking of ontology concepts.
17
17
18. Further features: support for multilingual
ontology management
Language Expert
Language Expert
MoKi
Knowledge Engineer
Modelling tool
the Modelling W iKi ---
Domain expert
Knowledge Engineer
Domain expert
Domain-adapted
Dictionary-based
Translation Services
Domain-adapted
Statistical Machine
Translation Services
Public Machine
Translation Services
18
19. Further features: connections with
external components
Automatic suggestions of mappings between ontologies
Managing ontology evolution
Exposure service
Triple stores
SPARQL Endpoints
…
19
19
20. Usages of
IP FP6 EU Project [03/2006 – 02/2010]
Purpose: modeling of tasks/processes in an enterprise and of
the topics related to that task (competencies)
Used by:
4 SMEs
3 Universities
several related summer schools and university courses
20
20
21. Usages of
STREP FP7 EU project [01/2010 – 12/2012]
Purpose: build/revise an environmental ontology
Developed the new key concepts extraction functionalities
Used to automatically create part of the ontology (pollen)
21
21
22. Usages of
eContentplus EU Project [09/2007 – 08/2010]
Purpose: build/revise an ontology of organic agriculture and
agroecology
Used to foster collaboration between domain experts (FAO) and
knowledge engineers
Follow-up: Organic.Lingua
(FP7 Pilot Tipe B EU project [36 months, 03-2011 – 02/2014])
Extend MoKi to multilingua models and interface
22
22
23. Usages of
Italian national project [01/2010 – 12/2011]
Purpose: model processes for analysis/revision and
dematerialization
Used by 7 Italian regions:
Piemonte, Emilia Romagna 1 & 2, Puglia, Liguria, Marche, Trentino
Medium size models produced in around 2 weeks.
23
23
24. Usages of
OncoCure
Funded by Fondazione Caritro, Trento [2007 – 2008]
Purpose: modeling breast cancer clinical protocols encoded in
Asbru.
Customized version of the tool
Actively used mainly by KE
Positive feedback by the doctor who produced the clinical
guidelines in “reviewing” the model created.
24
24
25. Usages of
Pro.Mo.
Trentino Legge 6 Project [03/2012 – 12/2013]
Purpose: link the business process modeling level with the
system level
Each task represents one or more hardware services.
An intermediate level permits to link the hardware level with the
modeled processes.
Data coming from the hardware level are mapped with the models
and analysis operations are performed.
25
25
26. Evaluation
Performed within ProDe project (to be presented @ ISWC2011);
Users: 14 Public Administration employees distributed across 6
teams creating different integrated models;
Research questions considered:
RQ1: Is MoKi easy to use for domain experts?
RQ2: Is MoKi useful for collaboratively modeling domain knowledge?
RQ3: Are all the provided views useful or is there a “best” view among
the different interface views provided by MoKi for: (a) getting the model
overview? (b) navigating the model? (c) creating new entities?
Analyses performed:
Quantitative analysis of the data on the usage of MoKi (editing logs,
web-server logs, …);
On-line questionnaire filled by the real users.
26
26
27. Evaluation Results
RQ1 (ease of use):
The users perceive the tool as more than easy to use:
• 72% of employees spent only less than two days to learn how to use tool;
• the same percentage learned it autonomously.
RQ2 (usefulness for collaborative modeling):
The users positively perceive the overall usefulness of the tool for
the collaborative modeling of documents and processes:
• Correlation between the size of the subject’s team and his/her feedback
about tool usefulness for collaborative purposes (esp. in team with 3+ or
more users).
• Result further validated by the intensive usage of collaborative
functionalities by people in large team.
RQ3 (usefulness of provided views):
All the views provided by the tool have their own usefulness.
27
27
28. Current & Future Works
Implementation of the workspace manager for managing multiple
ontologies
Develop ad-hoc templates to guide DE in modeling activities
Support usage of ontology patterns
to speed up modeling activities, and limit modeling errors
Extend key-concepts extraction functionalities
describing an artifact is different than describing a role
Support extraction / identification of semantic relation in text (e.g. “isA”) between
concepts
Architectural improvements for speeding up the customization of the graphical
interface of the tool
28
28
29. Resources
Publications and demos:
ESWC2013, ESWC2012, ISWC2011, EKAW2010, ISWC2010,
SemWiki2009, ESWC2009, …
Released Open Source in July 2010 (version 1.2 – GPL2)
New release will come in January 2014 (version 3.0 – GPL3)
MoKi WebSite:
URL: http://moki.fbk.eu
On-line demos, code download, documentation, news, support…
29
29