The DYNAMOS approach to support context-aware service provisioning in mobile environments
Introduction
With their small size format and ubiquitous connectivity mobile devices such as smart phones and PDAs can offer interesting opportunities for novel services and applications in ubiquitous computing environments. On the other hand, given the huge amount of information and services potentially available in such environments, and the limitations anyway present in such devices (e.g., small-sized screens, limited processing power), the ways of searching content and discovering services are also evolving. Mobile users need to locate anytime, anywhere “relevant” content which is available in their daily environments, where relevance has a user-specific definition (e.g., cost, location, accessibility, etc.). According to the I-centric model proposed by the WWRF (Wireless World Research Forum), personalization, ambient awareness, and adaptability are the three core properties that set functional requirements on future service platforms (Arbanowski et al., 2004). For instance, a smart tourist guide should dynamically retrieve maps and information about shops, restaurants, and monuments located in the user’s vicinity and depending on the current task, preferences, and social context.
Various research proposals seek to accomplish this user-centric paradigm in mobile environments. Context and context-awareness (Dey et al., 2001) allow services to dynamically adapt to their computing environment. Sensed context information permits inferring user’s requirements and needs in order to make applications more personalized, friendly, selective, and responsive (Chen and Kotz, 2000). For example, mobility-aware recommenders such as the PILGRIM system (Brunato and Battiti, 2003) use the user’s location to filter web links of interest. Context-based applications such as tourism information services (Cheverst et al., 2000) or remembrance agents (Rhodes, 1997) proactively retrieve content of interest based on the user’s current task. Although these systems target different needs, they share a common design: the system collects different types of (contextual) information characterizing the user, and uses it to filter and rank relevant content items. However, this approach does not always reflect our daily life practice. Let us consider this example1
I don’t pay much attention to movie reviewers or restaurant critics. I’ve been disappointed by them too many times. Instead, I ask a friend I can trust, “Have you seen any good movies lately?” Frequently, people call or e-mail to tell me that they’ve eaten at a restaurant, or read a magazine article that they know I’ll love. Businesses thrive or whither based on this informal method of marketing. But is there a way to harness and direct the power of word-of-mouth advertising?
This extract highlights two motivations behind our work. Many information retrieval, information filtering, and recommendation systems are based on the construction of preference models; preferences can be built by directly asking the user to fill out a form, by collecting feedbacks upon system usage, or by observing her behavior over time. We argue that in many cases, systems supporting context-based content provisioning cannot assume to have complete access to user’s personal and contextual information. Reasons for this could be user’s reluctance due to privacy, security, and trust concerns, or mere laziness. Conversely, we observe that information about the user that the system cannot acquire might be known by other users such as friends, colleagues or persons in the proximity.
Our second motivation originates from the observation that the success of systems like Amazon have demonstrated the importance for customers to browse reviews and ratings expressed by previous clients. Typically, service content provided by professional content providers (meaning commercial service providers or public administration bodies) is officially expressed, impersonal, and utility-oriented. In addition, the information space easily becomes static and out of date as updating information has a cost and is time consuming. Rather than being a monolithic and static concept, we consider the service space as a dynamic and malleable region of interaction and experience, in which occupants or visitors are able to capture their live experience and create a record to other users for later access and review (Espinoza et al., 2001, Abowd, 1999).
Stemming from these considerations, we propose a hybrid model of context-aware service provisioning. It is our intention to complement the common user-centric model of content provisioning with peer-to-peer social functionalities,2 which are largely successful in Internet communities and on which we commonly rely in our daily life. Moreover, it is our objective to concretely evaluate such an approach in real-world daily scenarios involving mobile users equipped with off-the-shelf mobile devices, such as smart phones.
This paper3 describes the system platform together with an application prototype running on smart phones that we designed and implemented in the DYNAMOS project4 to support our hybrid model of service provisioning. The DYNAMOS system allows mobile users (i) to be proactively provided with a subset of relevant services available in the territory; (ii) to generate several types of contextual notes, attach them to geographical locations, and eventually share them with other users; and (iii) to annotate official service descriptions with personal observations, comments, or ratings to be shared with others. To evaluate the feasibility and usefulness of the hybrid model, we conducted field trials where a community of recreational boaters used our application during a sailing regatta in the Helsinki region.
The rest of the paper is organized as follows. Section 2 describes the hybrid model of context-aware service provisioning. Section 3 gives insights on how such an approach has been accomplished by the DYNAMOS platform and Section 4 details its implementation. Section 5 describes the investigated use case and discusses results from the field trials we conducted. Section 6 reviews related work. The paper concludes in Section 7.
Section snippets
Hybrid model of context-aware service provisioning
As Fig. 1 depicts, our hybrid model combines the service provider to consumer (B2C) and the consumer to consumer (C2C) models. The resulting model allows users to receive official content describing available services (service-generated content) and to generate unofficial content such as observations and personal opinions (user-generated content), attach it to service descriptions (service-generated, but user-annotated, content), and eventually share it in a small/medium-sized community of
DYNAMOS platform
In the context of the DYNAMOS project, we built a system platform and application prototype designed for smart phones to support the functionalities of the hybrid model. This section gives insights on the design of such a system.
Implementation
The DYNAMOS system and the prototype application on mobile phones have been implemented in Java. Event-based communication is realized through the Fuego Core Event Server (Tarkoma et al., 2006). The Fuego middleware is implemented in Java and provides a scalable distributed event framework and XML-based messaging services. This middleware also runs on mobile phones supporting Java MIDP 1.0. The application running on mobile phones has been implemented using Java 2 Micro-Edition (J2ME) with the
Case study: recreational boaters in the Turku Archipelago
To demonstrate the feasibility and usefulness of our hybrid model, we selected a community of recreational boaters as scenario of study. Recreational boaters are people who have either sailboating or motor boating as a hobby. We chose this scenario since it represents a typical case in which, given the intensiveness of the task and the dynamism of the situation, proactive and context-aware service provisioning can provide significant benefits.
Boaters constantly need up-to-date information about
Related work
Our research shares some concepts and techniques typically used in conventional recommender systems. With the emergence of e-services, recommenders have been often applied in various forms and fields, and especially in information retrieval. Recommenders are capable of ranking a list of similar items with respect to a certain criteria in order to provide recommendations, predictions, opinions that can assist users in evaluating items (Resnick and Varian, 1997, Schafer et al., 2002).
The two most
Conclusions
The hybrid model of context-aware service provisioning represents a novel approach to access and share information about services and content available in our daily environments. We designed and implemented a platform along with a prototype application running on smart phones, which support this model. We evaluated the feasibility of our approach through field trials in which the research subject was a community of recreational boaters. Our model has proved so far to be an interesting and
Acknowledgements
The authors would like to thank Tekes, ICT-Turku, Suunto, TeliaSonera, and VTT for funding the DYNAMOS Project. Thanks also to the Research and Training Foundation of TeliaSonera.
Oriana Riva is a researcher at the Helsinki Institute for Information Technology. She is completing her doctoral dissertation in computer science at the University of Helsinki on how to support mobile sensing applications in heterogenous pervasive environments. Her research interests include ubiquitous computing, ad hoc networking, and context-aware services for mobile users. She received her M.Sc. in Telecommunication Engineering from Politecnico di Milano, Italy in 2003.
References (48)
Classroom 2000: An experiment with the instrumentation of a living educational environment
IBM System Journal
(1999)- et al.
Incorporating contextual information in recommender systems using a multidimensional approach
ACM Transactions on Information Systems
(2005) - et al.
I-centric communications: personalization, ambient awareness, and adaptability for future mobile services
IEEE Communications Magazine
(2004) - et al.
FAB: content-based, collaborative recommendation
Communication of ACM
(1997) - et al.
Context-aware middleware for resource management in the wireless Internet
IEEE Transactions on Software Engineering
(2003) - Breese, J., Heckerman, D., Kadie, C., 1998. Empirical analysis of predictive algorithms for collaborative filtering....
- Brown, B., Chalmers, M., 2003. Tourism and mobile technology. In: K. Kuutti, E. H. K. e. a. (Ed.), Proceedings of the...
- et al.
PILGRIM: a location broker and mobility-aware recommendation system
Hybrid recommender systems: survey and experiments
User Modeling and User-Adapted Interaction
(2002)- et al.
E-Graffiti: evaluating real-world use of a context-aware system
Interacting With Computers: Special Issue on Universal Usability
(2001)
CARISMA: context-aware reflective middleware system for mobile applications
IEEE Transactions on Software Engineering
Developing a context-aware electronic tourist guide: some issues and experiences
Searching the semantic Web: approximate query processing based on ontologies
IEEE Intelligent Systems
A conceptual framework and a toolkit for supporting the rapid prototyping of context-aware applications
Human–Computer Interaction
GeoNotes: social and navigational aspects of location-based information systems
Design Patterns: Elements of Reusable Object-Oriented Software
Combining collaborative filtering with personal agents for better recommendations
A survey of trust in Internet applications
IEEE Communications Surveys and Tutorials
An algorithmic framework for performing collaborative filtering
Cited by (19)
A systematic review on the engineering of software for ubiquitous systems
2016, Journal of Systems and SoftwareDevelopment of context-aware workflow systems based on Petri Net Markup Language
2014, Computer Standards and InterfacesCitation Excerpt :Their method allows one to specify a context-aware pervasive system by means of a set of models, which describes both the system functionality and the context information, and generate code based on the models. Riva and Toivonen have investigated a hybrid approach that enhances context-aware service provisioning with peer-to-peer social functionalities in the DYNAMOS project [31]. Chaari, Ejigu, Laforest and Scuturici propose a generic adaptation framework that involves content adaptation and presentation adaptation to guarantee adaptation of applications to the context in a pervasive computing environment [7].
Context-aware systems: A literature review and classification
2009, Expert Systems with ApplicationsSymbian OS for developing mobile multimedia software platform for gymnastics
2017, International Journal of Emerging Technologies in LearningMobile multimedia recommendation in smart communities: A survey
2013, IEEE AccessA proactive personalised mobile recommendation system using analytic hierarchy process and Bayesian network
2012, Journal of Internet Services and Applications
Oriana Riva is a researcher at the Helsinki Institute for Information Technology. She is completing her doctoral dissertation in computer science at the University of Helsinki on how to support mobile sensing applications in heterogenous pervasive environments. Her research interests include ubiquitous computing, ad hoc networking, and context-aware services for mobile users. She received her M.Sc. in Telecommunication Engineering from Politecnico di Milano, Italy in 2003.
Santtu Toivonen is a senior researcher at VTT Technical Research Centre of Finland. Prior to joining VTT in 2001, Santtu worked at Sonera, a Finnish telecom operator. He holds a master’s degree from the University of Helsinki, with the major in cognitive science. He is currently pursuing his graduate studies in the same university, with the research focus on supporting users’ everyday cognitive tasks with context-aware systems. His professional interests include context-awareness, social media, usability, and the Semantic Web. He has published and acted as PC member and reviewer in several conferences and workshops in these research areas.