Semantic Web in Group Formation Asma Ounnas Learning Societies Lab School of Electronics and Computer Science The University of Southampton, UK www. ecs . soton .ac.uk/~ao05r ao05r@ecs. soton .ac. uk
Mar 30, 2015
Semantic Web in Group Formation
Asma OunnasLearning Societies LabSchool of Electronics and Computer ScienceThe University of Southampton, UKwww.ecs.soton.ac.uk/[email protected]
Group Formation in e-learning
g1
g2
gN
Formation in terms of the constraints
+ Collaboration Goals
as a set of constraints
StudentsInstructor
Groups
Reasoning Web Summer School, Dresden, 07/09/2007
International
female
leader
I want David to be supervise my group
We want Web-based group formation in e-learning
Personalize the allocation of students
Give a degree of freedom for choosing constraints for the formation
Enable the formation of different types of groups
Reasoning Web Summer School, Dresden, 07/09/2007
Potential of the Semantic Web
Used in Personalization of Learning Objects
Used in Social networks
Used in CoPs and Expert Finders
Reasoning Web Summer School, Dresden, 07/09/2007
Constraint-based Group Formation The allocation of participants to groups based on
some constraints
Collaboration task has a set of goals, each is a set of constraints
Each constraint has a value
Maximize the utility of all constraints within all goals and minimize the deviation of the groups satisfaction => Optimal formation
Reasoning Web Summer School, Dresden, 07/09/2007
Assumptions
non-overlapping group formation
all groups have a similar size
All formed groups are stable while the formation is not announced by the instructor
Reasoning Web Summer School, Dresden, 07/09/2007
Hypothesis
Semantic Web technologies
can effectively automate the process of Web-based constrained group formation in the
e-learning domain.
Reasoning Web Summer School, Dresden, 07/09/2007
Group Formation Process
1. Initiating the formation: the instructor sets the constraints and the students’ list;
2. Identifying the members: analysing the students profiles to identify who should be in which group, the descriptions of the students
3. Negotiating the formation: the negotiation of the students’ allocations to groups involves running a specific algorithm that can satisfy the constraints.
=> Modeling + Constraint Satisfaction
Modeling Variables Task-related: experience, education level, knowledge,
skills, abilities (cognitive and physical), grades, interests, preferences of topics and experts.
Relation-oriented: gender, age, culture (race, ethnicity, national origin), social status, personality and behavioral style, social ties, trust.
Context-related: geographical location, availability schedules, and communication tools.
Reasoning Web Summer School, Dresden, 07/09/2007
Modeling For Group Formation
Group Type Task-related Relation-oriented
Teams All variables All variables
Communities Interests, topic preferences, experience, expert preferences
Expertise relationship, trust
Intensional Networks
Skills, abilities, experience None
Social Networks Interests, topic preferences Social ties, trust
Reasoning Web Summer School, Dresden, 07/09/2007
Implementation
Reasoning Web Summer School, Dresden, 07/09/2007
Groups listGroups list
Jena persistent storage (ontology instances)
Students Interface
(Extended foaf-a-matic)
Students Profiles
(SLP ontology)
Group Generator
(inference rules)Group Formation
Algorithms
Instructor InterfaceInstructor
Constraints
Instructor
Students
Semantic Learner Profile Ontology
Based on FOAF Added 14 classes and 34 properties Uses the University of Southampton CS
modules ontology Uses the Trust ontology Domain ontologies can be added
Reasoning Web Summer School, Dresden, 07/09/2007
Semantic Learner Profile Ontology<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#" xmlns:foaf="http://xmlns.com/foaf/0.1/" xmlns:slp="http://www.ecs.soton.ac.uk/~ao05r/LearnerProfile/slp.owl">
<foaf:Person rdf:nodeID="asma"> <foaf:name>Asma Ounnas</foaf:name>
<foaf:gender>Female</foaf:gender> <slp:teamRole>leader</slp:teamRole>
<slp:firstLanguage>Arabic</slp:FirstLanguage> <slp:takingModule><slp:moduleID>Web Technologies</slp:moduleID></slp:takingModule> <slp:preferredModule>Knowledge Technologies</slp:preferredModule> <slp:preferredTopic>e-learning</slp:preferredTopic> <slp:interest>Semantic Web</slp:interest> <slp:studentOf><foaf:Person><foaf:name>Hugh Davis</foaf:name></foaf:Person></slp:studentOf>
<slp:classmateOf><foaf:Person><foaf:name>Ilaria Liccardi</foaf:name></foaf:Person></slp:classmateOf> </foaf:Person></rdf:RDF>
Reasoning Web Summer School, Dresden, 07/09/2007
Student User Interface
Reasoning Web Summer School, Dresden, 07/09/2007
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
Instructor User Interface
Group Generator
CSP solvers to get the optimal formation data mining module for incomplete data Deduction rules eg.
IF John is a captain of the football team, THEN John is a leader
IF Sarah has a high grade in discrete mathematics AND Sarah has a high grade in Logic for CS, THEN Sarah will perform well in formal methods
Reasoning Web Summer School, Dresden, 07/09/2007
Evaluation Metrics framework for evaluating Group formation
Quality Using 3 algorithms for group formation with
incomplete data: without rules vs with rules. No claim for:
Proving that any particular set of constraints leads to better results in terms of the groups’ performance.
Claiming that any particular algorithm leads to best grouping.
Feedback
Questions
Reasoning Web Summer School, Dresden, 07/09/2007
Suggestions
Coments