Context aware systems

Thesis Description:Context aware systems in Public Spaces (used for the Ifi administration)

= Scenario =

Scenario used from AEons proposal. Fabrice to describe the scenario

Sensor events: Mobile act as sensor, and mobile information is becoming part of business decisions. Three main areas in public


 * information - example: (calendar based), "the next travel possibilities to your meeting at UNIK is bus 352 from stop place Q"
 * entertainment - example from amusement park: you have visited attraction x, y, z. I recommend you to visit "v". These games might also be of interest for you
 * advertisement - example: closest shops, "discount", "you have bought x, product y fits perfectly to it".
 * public places or as app on your phone
 * time spent infront of an object gives you a recommender for similar objects

Profile:


 * user profile containing preferences
 * trust-based communities ("friends")

Context profile:


 * describing the situation

Decision making


 * recommender system for
 * smart home (light, heating, music) -> talk to Helge.Godø and Akershus Energi (alternative to Danfoss) - home control Z-Wave
 * eHealth (anormalities, behaviour, alarm) -> Dave (Dafferianto) eHealth at UiA
 * information on competing products, product review, prices -> Arvid.Fjesme (NOOM)
 * discounts for products and opportunities

,

Hanging around, shopping scenario
Context filtering as offer to the cloud



The big challenge in getting just the products you want is to leave enough information, but not leave that much that you will be spammed by emails afterwards.

Example 1 - dedicating shopping:


 * Alex has a puncture on his bike. He wants to find a bicycle shop nearby in order to get help with the repair.
 * He might open his mobile and search, opens an app and writes down his wishes, or he just sends an sms with keyword ..
 * Result is that he will get an offer from a shop nearby

Example 2 - "hanging around":


 * Catherine is hanging around with her friends in a certain area of the city
 * everyone has a "basic profile" out in the cloud, telling about her interests
 * based on this information and the "service offers" they will get an invitation to cinema, shopping, exhibition
 * how to not spam with offers - specific app on the phone offer collection with personalized alarm (flash of last offer, vibrate) - eksempel: GEEK, pulse

= Challenges =


 * privacy, access to user profile
 * mobile environment ("always online")
 * context description - what is really my content (automatic, learning profile)
 * non-intrusive messages

= Topics =


 * "background" recommender system
 * "context and preference learning" system (Philips: assume a 200 GByte hardisk on a STB, and 15 dedicated recordings will provide you with 95% of your viewing time)
 * Protege, creates ontology. Ontology (.owl) describes through semantics the preferences, context, ....
 * JavaAPI to access ontologies directly from programs
 * Reasoner for decision making

= Objectives =

Main focus will be on the context-sensor, contributing to a distributed user profile. Depending on the context, different informaiton is provided to the public space, to friends/colleagues and sensitive information is kept in the mobile.


 * Create a state-of-the-art review on potential middleware
 * include MUSIC, ...
 * semantic technologies for knowledge management (profiles, context) - ontologies


 * interface to LinkedOpenData, OData or "Tripple Store" (RDF)


 * Adopt an architecture

= TOC =

Title page, abstract, ...


 * 1. Introduction, containing: short intro into the area, what is happening
 * 1.1 Motivation, containing: what triggered me to write about what I'm writing about
 * 1.2 Methods, containing: which methods are you using, how do you apply them


 * 2. Scenario, optional chapter for explaining some use cases
 * one main scenario, and applicability of others; what is the outcome of the scenario ("switching action", display information, ...)
 * 2.1 user scenario, (bad name, needs something bedre)
 * 2.2 Requirements/Technological challenges, here: Semantic technologies, User Profiles/preferences, context-aware services, ... WLAN provisioning (FON.com), reasoning towards a group profile (advertisement, entertainment, information)


 * 3. State-of-the art/Analysis of technology, structure your content after hardware/SW (or other domains). Describe which technologies might be used to answer the challenges, and how they can answer the challenges. Here: Middleware for bindling the moduls together. Focus on engine, Web or mobile?
 * for all subchapters: provide an overview, then evaluate ("what is important for me?"), recommendation on which solution to go for
 * 3.1 technology A: Semantic technologies, Protege, Ontology
 * 3.2 technology B: User profiles (survey: SPICE, FOAF, tourist ontology,... -> none of them fit to my needs. Extend FOAF in areas of .....)
 * 3.3 Context awareness:
 * 3.4 Reasoning: overview, "why SWRL?"


 * 4. Implementation
 * 4.1 Architecture, functionality
 * 4.2


 * 5. Evaluation
 * 5.1 Security (man in the middle attacks, replay attacks...)
 * 5.2 Group Profile based service Personalization
 * 5.3 Future work
 * 6. Conclusions
 * References