Intelligence, mobility and learning Russell Beale School of Computer Science University of Birmingham UK
e-learning User freedom Strong informal learning –Task-based learning –Just-in-time learning –Multiplicity of learning episodes Focus on supporting the user everywhere, everytime
e-teaching and e-learning e-teaching – provision of materials and/or delivery mechanisms electronically –Intelligent tutoring systems –WebCT –Focus on teacher push e-learning – providing e-support for learning activities –Focus on user pull –May involve e-teaching Big difference!
We have the power… Technologies –802.11a-g, broadband, bluetooth, infra-red, GPRS, 3G –Pda, tablet, phone, laptop, desktop, wearable Software –J2ME, J2EE,.net, WSDL, IDEs Theories –OO programming, extreme programming, agent- based systems, component architectures, user- centred design
Evolution: e-m-u First we had E-lectronic –Digital media, delivery and control Then M-obile –Anywhere, anytime Not tied to desks, access when you like Now towards U-biquitous –Everywhere, everytime Environment enhanced, not just the user Access wherever, whenever
Key characteristics for supportive systems Timely –Access to things just in time Simple –Invisible, unobtrusive technology Relevant –Knowing about context and task for appropriate support Information –Not data – structured and organised
LIFE support systems Learning –Formal and informal, structured and lifelong Information –As and when appropriate Facilitation –Communication with people, systems, environment; recording experiences Entertainment –Social and individual, t.v. to movies, chat to internet
So we need to have effective… Learning support –For a variety of learning styles and users Mobility –Because people go everywhere, need to support tasks and activities everywhere Light, fast, well-connected, appropriate, tough… Intelligence –User models, task models, learning models –Evaluation of environmental context – bandwidth, devices, others –Context
Context Explicit context –User given. A dynamic profile. Implicit context –Inferred by system from available evidence Environment, user actions, active models etc. Context used to guide actions –More than filtering data –Looks into the future to predict and guide appropriate actions aside
Supportive architectures Want a software architecture to support this –Variable connectivity, multiple channels, ad-hoc conversations –Structured and informal interactions –Uncertain environments, embedded intelligence
Ecology of interaction Interrelated systems –Symbiotic interactions Low level reactive systems –Plants, bacteria c.f. phones, web pages Mid-level emergent behaviours –Ant colonies, bee swarms c.f. SMS Higher level intelligence –Animals c.f. basic AI systems Design ethos
Achieving intelligence Modular –Creates low level systems Highly connectable –Allows emergent behaviours in mid-level systems Use of appropriate intelligence –Moves towards higher level systems Design ethos
activeSpace Component-based architecture Independence of data flow and initiative –Data can flow in, out or bidirectional –Connections can be push (data driven) or pull (demand driven) –Enables mixed initiative interfaces, copes with diverse environment External linkage –Components don’t have to know about each other –Easily distributed –Allows emergent behaviour
activeSpace architecture
Specialised components Sources –Monitor environment or provide data E.g. GPS latitude, longitude Recognisers –Take data sources of one type and attempt to translate them into higher level concepts E.g. GPS to logical location (“in art gallery”) Services –Take data and offer something appropriate E.g. if in art gallery, provide web pages about the artists
Summary Intelligence an important part of mobile e-learning systems Component-based software architectures provide good way to achieve this
LIFE support systems should… Offer great advantages to user Be invisible Be clever Predict the future Be always appropriate Enrich your life They should be magical!