Ontology-Based Techniques for Context-Aware Personalization of Educational Programs Amir Bahmani1, Dr. Sahra Sedigh2, and Dr. Ali Hurson1 1Department of Computer Science 2Department of Electrical and Computer Engineering Sixth Annual ISC Graduate Research Symposium April 13, 2012 1 Outline • PERCEPOLIS – Shortcomings of STEM Education – Modularity • • • • • Context-Aware Systems The Proposed Context-Aware System Personalization Processes Prototype Conclusions 2 Current Shortcomings of STEM Education • Static and linear curricula – Inability to keep up with advances in technology – Redundancy AND lack of reinforcement of topics among courses • Static and linear teaching practices Consequences – Prevalent pedagogy is not well-suited to learning style of millennial students. Low enrollment, retention, andused graduation rates in STEM programs. – Learning technologies are not effectively. • LackStudents of resources: faculty, facilities, who doskilled graduate are not preparedequipment for “professional practice.” 3 Solutions Proposed • By National Academy of Engineering: – Personalized learning – identified as one of 14 Grand Challenges in Engineering for the next century • By President Obama’s Strategy for American Innovation: – Use of learning technologies in higher education – listed as one of six educational objectives • Common sense (and overwhelming evidence) – Resource sharing – Teaching collaboration – Active and peer learning 4 Our Proposal 5 Modularity • The modular approach increases the resolution of the curriculum and allows for finer-grained personalization of learning objects and associated data collection. CS 388- High Performance Computer Architecture Performance Metrics RISC vs. CISC Arithmetic Logic Unit ... Beyond RISC 6 Personalization Hierarchy CS - Curriculum CpE 111 Performance Metrics Performance Metrics RISC CISC vs. RISC CISC Parallel and Serial ALU ALU Memory Beyond RISC Content Functional Address Accessible Accessible Superscalar ALU Memories Memories Superscalar for Beginner Study Prerequisite Relation … CS 388 Student Profile Superscalar for Intermediate Study Access Environment CS XXX Concurrency Programming VLIW Parallelism Pipelining VHDL Programming Superscalar for Advance Study Modules 7 Context-Aware Systems • Context-awareness: – The use of context in software applications that adapt their behavior based on the discovered context. • Any context-aware system contains two main parts: – 1) Context management subsystem concerned with context acquisition and dissemination – 2) Context modeling concerned with recognizing, representing, and manipulating context and situations. 8 Context-Aware Systems (cont’d) • An ontology is a representation of the universe; it shows how different entities are related. Cat Cat is-a is-a Lion Tiger Taxonomy has-a Tail is-a lives in Carnivore Jungle Ontology • Ontology-based modeling allows: 1. knowledge sharing 2. logic inference 3. knowledge reuse 9 Proposed Context-Aware System • The strengths of our system are: – Leveraging both individual and peer group information to offer better recommendations – Being flexible and user-friendly – Exceeding the functionality of competing alternatives – Updating the content of recommendations based on student’s environment 10 Related Literature • The C-CAST context management architecture supports mobile context-based services by decoupling provisioning, and consumption. – The system is built based on three basic functional entities: the context consumer (CxC), context broker (CxB), and context provider (CxP) • Hybrid Context Management (HCoM) uses semantic ontology and relational schema to represent graphical context data. Related Literature (cont’d) • A context aware framework (CAF) enables the context-aware applications and services, while being domain-agnostic and adaptable. – The CAF contains two core components: the data acquisition component and the context manager. Proposed Context-Aware System(cont’d) Inference Engine (IE) Context Management Layer Domain Ontology Inferred Context Context Interpreter Layer Context Provider Layer Recommender System (RS) Context State Recommendation Algorithms Recommendation Context Adaptive Presentation Operation Generic Ontology Context Attributes Context Manager (CM) Store / Retrieve Context Context Database PERCEPOLIS System Terms Context Verifier Recommendation Requests & Feedbacks Summary Schema Model Input Data Context Delivery 13 Software Agent Personalization Processes Personalization Processes Curriculum Course PERCEPOLIS Student Retrieve departmental rules Find potential courses based on student’s profile and department rules Prioritize the list based on Student’s interests and the result of collaborative filtering Select desired courses Overall check on the selected courses For each selected course Topic Subtopic Retrieve tentative list of topics Remove Topics have been taken Check whether the list satisfies the course constructor’s expectations. If “No”. Revise the list and add advanced topics For each selected topic Retrieve tentative list of topics Remove subtopics have been taken or are being taken Prioritize the list based on Student’s interests and the result of collaborative filtering Select desired subtopics For all selected subtopic Module Find the most appropriate modules for The selected subtopics based on student’s profile (Student's infrastructure and background) 14 Prototyping • The first version of the cyberinfrastructure prototype, based on the proposed contextaware system, is partially operational. • The prototype and profile databases have been implemented in Java SE 6 and MySQL 5.5.8, respectively. 15 Prototyping (cont’d) 16 Conclusion • In this work within the scope of PERCEPOLIS: – A new layered context-aware system is presented – The functionalities and strengths of the proposed system are verified by the help of the first prototype of the system • Future work includes enhancing and performing predictive modeling of the recommendation algorithms for performance and accuracy. 17 Questions? 18