Exploiting the Knowledge Economy: Issues, Applications, Case Studies Paul Cunningham and Miriam Cunningham (Eds) IOS Press, 2006 Amsterdam ISBN: 1-58603-682-3
A Context-Aware e-Tourism Application Enabling Collaboration and Knowledge Exchange among Tourists Federica PAGANELLI , Gabriele BIANCHI , Valerio MELANI CNIT, Electronics and Telecommunications Department, University of Florence, v. S. Marta 3, Florence, Italy Tel: +39 055 4796382, Fax: +39 055 488883 Email:
[email protected];
[email protected];
[email protected] Abstract: Tourism is a key scenario for context-aware applications. Well known examples of context-aware e-tourism services are location-based content delivery and interface adaptation to current user activity. Our approach differs from previous ones as we consider a tourist not only as a target for content delivery, but also a source of valuable information, useful for other tourists and service providers as well. This work, in the framework of the Kamer project, describes an eTourism context-aware application, providing tourists on the move with proper context-aware services to support communication and knowledge exchange. The application integrates already available location-based content delivery services with a contextaware instant messaging service, and a tourist service provider reputation service, which supports tourists during decision-making processes. Here we describe main design issues and prototype implementation.
1. Introduction Kamer (Knowledge Management in Ambient Intelligence: technologies enabling the innovative development of Emilia Romagna) is an Italian regional project that involves two universities and one industry. The main objective of the Kamer project is to define novel models and technologies of Ambient Intelligence (AmI) and context-awareness supporting knowledge exchange and collaboration among individuals in mobility. The scenario that has been identified for testing out knowledge processes focuses on tourists visiting a cultural city (Piacenza) and its points of interest. Tourism is a key scenario for context-aware applications. During a journey, a tourist demands a set of services in a manner often depending on the current situation. Well-known examples of context-aware services for tourists equipped with mobile devices are locationbased content delivery (e.g., artwork descriptions or commercial offers), or humancomputer interface adaptation to the current activity (e.g., visit or relax). Our approach differs from the previous ones as we consider a tourist not only as a target for commercial promotion and cultural content delivery, but also as a source of valuable information, which can be useful for other tourists and for tourist service providers as well. We believe that seeing tourists not just as individuals, but as members of communities is a key concept for the design of innovative eTourism applications. We intend tourist communities as groups of tourists with common interests, where communication should be enabled and knowledge should be formalised and exchanged for the benefit of all members. A main objective of the Kamer project is to develop a prototypal eTourism contextaware application, providing tourists on the move with proper services to support the above-mentioned aspects of community building based on communication and knowledge Copyright © 2006 The Authors
exchange, together with already available location-based content delivery services. This paper aims at describing the first set of these services developed in their prototypal version: • An instant messaging service, supporting the creation of dynamic tourist communities, exchanging comments about common interests and accessed services. • A service provider reputation manager, supporting tourists during decision-making processes. Here, context awareness means not only awareness of user location, but also awareness of user preferences, current activity, closeness to points of interest and physical proximity to other tourists, especially those with similar preferences and itineraries. In order to cope with these requirements, the Kamer prototype includes a context management middleware that supports context data acquisition, storage and delivery to applications. This work describes main design issues and prototype implementation of the Kamer eTourism application and context management middleware. It is intended for academic and industrial researchers working in the field of innovative eTourism services and contextaware application design and development.
2. State of the Art Prior efforts to develop context aware applications in the tourism domain are numerous. Early work includes the GUIDE system [1], which provides tourists with location-based information by accessing a PDA. COMPASS [2] is an application providing tourists with context-aware recommendations and services. Gulliver Genie [3] is a prototype of a context-aware application, delivering proactive information services to tourists based on their location and preferences. Another example of a city guide is Sightseeing4U [4], which delivers location-based information via a multimodal user interface on a mobile device. These applications represent relevant contributions to the field of tourist context-aware service development. However, some issues are still not addressed. First of all, most of these context-aware tourist applications ([1], [2], [3]) do not offer mechanisms to extend pre-defined context data, store data in proprietary formats and are designed as strictly vertical applications [5]. As a consequence, context data management is not extensible in the sense of both context data acquisition and delivery to applications. Only Sightseeing4U offers abstract interfaces for context acquisition, thus allowing the integration of not previously foreseen context data providers. In any case, it is not clear if standard interfaces are provided in order to integrate external applications (e.g. via Web Services). Regarding assistance provided to tourists, these applications mainly provide locationbased content delivery and the social dimension of tourism is not sufficiently supported. A first attempt to investigate a new approach about mobile technologies for tourist service delivery, considering tourists as knowledge units, not only as content consumers, has been done in History Unwired [6]. At present, project results focus mainly on designing mobile solutions for improving knowledge exchange between tourists and locals. Based on this state-of-the-art analysis, our contribution proposes a novel approach for tourist community building, aiming to improve communication & knowledge exchange and leveraging on context-aware services. Moreover, our design approach aims at coping with requirements of extensibility and interoperability, by means of open formats for context data representation and standard interfaces based on Web Services, as detailed hereafter.
3. Scenario Here we describe a scenario illustrating Kamer eTourism Application functional requirements. Tom, the tourist, equipped with a PDA, starts his visit in the Piacenza city-centre. When Tom is near a tourist point of interest, the eTourism application proactively shows the Copyright © 2006 The Authors
description and multimedia contents related to the monument. The application also provides an IM service. Tom’s contact list is dynamically updated by the system to show the list of other Kamer tourists currently located close to him. For each contact in Tom’s list, the IM client displays some context data: the current physical location, profile data (e.g., age, nationality, preferences, tourist path) and current activity (e.g., relax, indoor visit). Tom can choose which tourists to contact according to these data. He decides to send a message to Julia, who is relaxing in a café nearby. A conversation about Piacenza is established. It’s lunchtime, the eTourism application alerts Tom that he is near a typical restaurant which fits his meal preferences. When Tom is prompted with this notification (e.g. “you are close to the restaurant...”), he can ask for the reputation of the restaurant and read textual comments left by other tourists. Optionally, he can contact feedback authors if still present on site (through instant messaging). He contacts Julia again to suggest a meeting in front of the restaurant. After lunch, he can insert his comment about the service.
4. Kamer eTourism Application Technology Description 4.1 Kamer System Architecture The context-aware Kamer eTourism application is based on the Kamer system, which provides basic services for the development of context-aware applications. The Kamer system architecture includes three types of logical components: • Context Providers: distributed sensors that gather context data (e.g., temperature, light). • A Context Management Middleware: distributed components, which take care of context data acquisition from heterogeneous context providers, ontology-based data representation and reasoning and delivery to context-aware applications. • Context Consumers: software entities that adapt their behavior to context data. More precisely, the Kamer eTourism application is configured as a context consumer.
Fig. 1 Kamer System Architecture
The implementation of this architecture is based on the service infrastructure provided by the Mobile Bristol Application Framework (MB) [7]. MB offers basic services and development tools intended to facilitate the rapid development and deployment of pervasive mobile applications. It offers modules for data acquisition from sensors (such as GPS and RFID for user’s location) and services for location-based content delivery (context consumers), thus providing the main skeleton of the Kamer eTourism application. Figure 1 represents the main blocks of the architecture: the above-mentioned MB, the Context Management Middleware (described below), and the eTourism context-aware application. The eTourism context-aware application provides tourists with an integrated Copyright © 2006 The Authors
access to the following services: location-based content delivery, instant messaging, service provider reputation management. The first one is provided by the MB and will not be described here, as it is out of the scope of this paper. The other two services are described hereafter. A Graphical User Interface (GUI) has been designed in order to provide tourists with an integrated access to these context-aware services via a Web browser on a PDA. 4.2 Context Management Middleware The Kamer System Architecture explicitly separates context management components from application logic. This choice aims at facilitating the design and development of contextaware applications and promotes reuse and extensibility of context management modules. The Context Management Middleware acquires data coming from heterogeneous sources (e.g. GPS, RFID, Web services, etc.) and processes them for data aggregation, inference of higher-level context data and delivery to applications. This middleware is based on the deployment and co-ordination of Context Management nodes (in short Context Manager, CM) in distributed physical locations, each monitoring the surrounding environment. Each CM is a J2EE module dealing with context data acquisition from context providers, data representation, reasoning and delivery to context-aware applications, managing queries from applications or notifying them when context has changed (Fig. 2). Context management is strictly dependent on the context data modeling approach. Our context data representation (Kamer Context Distribution and Reasoning Model), follows a hybrid approach, as it includes two sub-models: an object-oriented data model (Context Acquisition and Delivery Data Model), encoded in XML Schema, suited for distributed context data handling and exchange, and an ontology-based model (Context Reasoning Data Model) encoded in OWL syntax [8], used for ontology and rule-based reasoning. The transformation between OWL context data format to CADD is performed via XSLT. The OWL ontology model represents data relevant for describing the context of entities populating the tourist scenario. Entities of interest for application purposes include tourists, but also places, more especially Points Of Interest (POI). For instance, a POI can be an important square or church. These POIs are useful for location-based content delivery but also for grouping tourists based on their reciprocal physical proximity, as described below. Entity context is described in terms of five categories of context items: location, activity, environmental data (e.g. weather), social context and instrumental context. For the Kamer eTourism application, we divided the centre of Piacenza into a set of areas, each corresponding to a POI. User location is provided by MB, via a GPS receiver and an RFID reader mounted on the user PDA. Geographical coordinates and RFID readings are mapped into POIs’ symbolic positions. Weather information is provided by an external web service. Description of device characteristics (instrumental context) is provided manually by the user at application start-up. Other items (tourist activity and social context) are inferred via deductive reasoning. With the term user “social context” we mean the set of tourists who are located close to the user himself, or more precisely in the same POI area. Context reasoning and management features implementation is based on Jena [9], a Java framework for building Semantic Web applications. Rules can be written in the Semantic Web Rule Language [10] and then transformed, via XSLT, in the Jena textual rule format. The Context model representation is stored in a relational database (MySQL). Communication between CM and context-aware applications for context data delivery is based on Web Services standards and on a Publish & Subscribe (P&S) service notifying changes of context data items to subscribed applications.
Copyright © 2006 The Authors
Fig. 2 Context Management Middleware Architecture
As mentioned above, this architecture allows the deployment of a distributed set of CMs. When a user moves from a place monitored by a CM to another place monitored by a second CM, user context data management “moves” consequently. In order to co-ordinate distributed CMs’ activities each CM exposes a set of Web Services to exchange the current context of a tourist or the entire representation of the CM context model with another CM. The P&S service is managed transparently to applications: applications subscribe once to user context items and subscriptions can be transmitted from one CM to a second one. The configuration already implemented for the Kamer eTourism application scenario includes two CMs: a central CM acting as default context manager and managing the contexts of all tourists in the system; a local CM deployed in a museum (Palazzo Farnese in Piacenza) responsible for tourist context there. During the tour, tourist context data are managed by the central CM, except when he/she visits the museum. While in most context-aware tourism applications user context data are entirely managed on user PDA, our approach aims to distribute context data acquisition and management features, assuming a permanent connection. Main advantages include: • the possibility of exploiting context data consistency check and reasoning capabilities of ontology-based technologies, which at present cannot be deployed on mobile devices; • the extensibility of the system: each local CM can be configured with ad-hoc reasoning rules, related to the kind of place in which it is deployed (e.g. a museum or a theatre). 4.3 Kamer Reputation Manager The tourist service provider reputation management system, in short Reputation Manager (RM), provides functionality for creation, updating and visualization of tourist service provider reputation. The system is based on a knowledge base fed by tourists’ feedbacks after having used services on site. Feedbacks are structured in terms of quality labels (e.g. service quality, kindness, etc.) and related ratings (positive, negative, neutral, or absent). Users can leave also a short comment. Feedbacks are submitted by tourists through web forms. The application logic of the Kamer Reputation Manager processes the collected feedbacks and applies statistical computation in order to calculate the reputation of the service provider, in terms of global reputation as well as what is obtained by grouping feedbacks together from tourists with similar profiles. A User Profile is made up of the following manually entered data at application start-up: age, nationality, preferences on tourist itineraries and tourist services (hotel, restaurants, etc.).
Copyright © 2006 The Authors
The RM has been implemented as a J2EE module. The Reputation knowledge base is stored in a relational database (MySQL). 4.4 Kamer Instant Messenger The Kamer Instant Messenger aims at promoting knowledge exchange and collaboration between tourists, by enhancing traditional communication and presence features of instant messaging with user context awareness to promote the creation of tourist communities. This system provides users with additional information (i.e. current activity and location) about personal contact list members, in addition to all instant messaging basic features. Personal contacts also include tourists on site which are currently located close to the user. These contacts are organized in a dynamic group called “TouristsNearby”. This group changes according to the composition of the user social context, which is continuously updated by the Context Management Middleware. The composition of tourists groups can be related to the user location, as in our scenario, but also to other kinds of context-based data items (e.g. common interests, nationality, current activity, etc.). This approach, based on the Context Management Middleware, is thus more flexible than other strictly proximity-based approaches (e.g. those using Bluetooth). The system architecture is composed by an IM client, an IM server enhanced with two plugins for context data management and a module (CClient) managing the communication with the CM (Fig. 3). Among existing IM protocols, we chose the Jabber Protocol [11], an open XML-based standard. A wide community is working on upgrading the protocol and adding new features, including context-awareness [12, 13]. The IM server implementation is based on Wildfire [14], a Java server supporting most Jabber features and easily extended with custom plugins. The IM client is based on webJabber [15], a J2EE client running on a Tomcat server and using a web browser GUI. In order to maintain a light configuration on the PDA, we decided to deploy the IM client at the server side. Dynamically generated HTML pages are sent to the PDA and displayed by the web browser. The Client manages the interaction between the IM server and the CM. It is subscribed to the CM in order to be notified when IM users’ context changes. Notification content is translated into the Jabber xml vocabulary and sent to the IM server inside a Jabber XML stream (called XML Stanza in the Jabber standard) with a custom namespace. At present we are developing two server plugins for context data management: one for user activity and location data, another for user social context. The first one is a Jabber XML stanzas handler which processes Stanzas with the above-mentioned custom namespace. When a change of a user’s location or activity occurs, the IM client of that user and those of his/her contact list members are notified of the change, according to the standard Jabber protocol [16]. The second plugin is responsible for the social context updating. This plugin manages the composition of the “TouristsNearby” group for each IM user, according to the user social context composition. Regarding the IM client implementation, management of the “TouristsNearby” group has not required additional features. On the contrary, activity/location data management will require the development of ad-hoc application logic for data management and presentation. Fig. 3 Kamer Instant Messaging Architecture
Copyright © 2006 The Authors
5. Implementation and First Experiences At present we are implementing the prototype system in order to test the Kamer eTourism application with tourists visiting Piacenza. The Context Management Middleware, the Reputation Manager module and the content-delivery application have been already implemented, while the IM system is under development. During the design and development of Kamer eTourism application we are facing several issues related to the user interaction with a mobile tourist application. As a matter of fact, tourists are generally not accustomed to using PDAs, and they do not want to be overwhelmed by the amount of effort required to interact with technological devices during a tour. In order to make the interaction as pleasant as possible we chose the HP iPAQ hx4700 PDA, which is characterised by higher resolution display (640x480) with respect to similar devices. Moreover, the application allows users to configure content delivery service behavior: contents are proactively delivered without any interaction with the user, or the option “view content” or “skip” is shown before content delivery. The first choice requires a minimal user interaction, while the second one allows the user to control the information flow. Introducing context-awareness features in the IM system requires the display of additional information in a clear and non invasive way. In common IM clients for desktop environments, contact list and contact presence information can be displayed on the main page and events are notified with alerts. This approach is not satisfactory for a contextaware IM system for a PDA, where additional context information has to be displayed on a small size screen and alerts to notify context changes could be too invasive. We decided to show context information in a separate ad-hoc contact details page, which can be reached from the main page, and let the user choose what context data type he/she should be alerted. We are planning two stages of testing in Piacenza, aiming at evaluating user acceptance of implemented features and interface design solutions. The city centre has WiFi coverage and tourists will be provided with PDAs, equipped with GPS receiver and RFID reader. For these tests some RFID tags will be disseminated in indoor environments (mainly in the Palazzo Farnese museum) in order to detect user physical proximity to points of interest. The testing environment includes a main server and a secondary one. The MB framework, the main instance of CM (central CM), the eTourism web application, the Kamer Instant Messaging system and the Reputation Manager are deployed on the main server. The other server, located in Palazzo Farnese, contains the secondary CM instance, which manages context data of tourists inside the museum. On the PDA, the client-side part of MB is deployed for context data acquisition (i.e. user’s current location), while the eTourism application is accessible via Web browser. The first stage (planned in July) will be conducted in group sessions, involving groups of young Italians, namely students. These students, equipped with a PDA, will visit the centre of Piacenza along predefined tourist itineraries. This stage will be useful to test the usability of our application with users who are usually quite accustomed to using technological devices. A second stage will involve older age groups from different foreign countries. These kinds of users will probably have more heterogeneous technology skills and consequently usability tests could be more interesting. Moreover, users with different nationalities will provide useful feedback for improving the RM profiling mechanisms, for example introducing filters related to user nationality in the service reputation management. Test results will be reported in the conference presentation.
6. Conclusions and Future Work In this paper we have outlined our approach for eTourism based on the concept of tourists as content producers and consumers and members of a community. In order to support Copyright © 2006 The Authors
these aspects, communication and knowledge exchange tools have been developed in the form of instant messaging and service provider reputation management systems. The originality of our approach resides in augmenting these services with context-awareness and offering them to users on the move together with location-based content delivery. Here we have described main design issues and implementation details of main components of the Kamer eTourism application, i.e. the Context Management Middleware, the Reputation Manager and context-aware Instant Messaging System. Context awareness requires an extensive use of data on individuals, thus implying several issues concerning privacy. In order to adequately address these problems, our future work will focus on solutions for privacy policy specification and enforcement. Industrial relevance of the project is proved by the strict collaboration with HP Italy, partner of the project. Knowledge exchange between industry and academia, through coordinated working in the chain from idea to innovation, is strongly suggested for the effective development of AmI technologies and markets at EU level [17]. The project focuses on the introduction of mobile and context-aware technologies in the business area of Tourism and related information service delivery. As a matter of fact, it leverages on the widespread success of mobile telephony. The increasing adoption of “converged devices”, such as PDAs, which integrates heterogeneous technologies, is a relevant factor promoting the commercial exploitation of Kamer results. The rapid spread of WiFi and other evolving wireless technologies will probably lead to rapidly falling communications costs, eliminating a major obstacle to the uptake of mobile tourist services. Actors which can potentially benefit of the Kamer eTourism scenario include: content providers, Application Service Providers, mobile device producers, Wireless Internet Service Provider and mobile telephony operators. Roles and revenue sharing for each actor and related business models will be object of study in a future project activity.
References [1]
[2]
[3] [4]
[5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17]
K. Cheverst, N. Davies, K. Mitchell, A. Friday, Experiences of developing and deploying a contextaware tourist guide: the GUIDE project. In: Proceedings of the 6th annual international conference on Mobile computing and networking, Boston, Massachusetts, United States, 2000, pp. 20 – 31. M. Van Setten, S. Pokraev, J. Koolwaaij, Context-Aware Recommendations in the Mobile Tourist Application COMPASS. In: Proceedings of Adaptive Hypermedia and Adaptive Web-Based Systems, Third International Conference, Eindhoven, The Netherlands, August 23-26, 2004, pp. 235-244. M. J. O’Grady, G. M. P. O’Hare, Gulliver's Genie: agency, mobility, adaptivity. In: Computers & Graphics, V. 28(5) , 2004, pp. 677-689. A. Scherp, S. Boll, mobileMM4U - framework support for dynamic personalized multimedia content on mobile systems. In: Proceedings of Techniques and Applications for Mobile Commerce Track of Multi Conference Business Informatics 2004, Essen, Germany, 2004. W. Schwinger, C. Grün, B. Pröll, W. Retschitzegger, A. Schauerhuber, Context-awareness in Mobile Tourism Guides - A Comprehensive Survey. In: Technical Report, 2005. History Unwired. Project Web Site. http://web.mit.edu/frontiers/. Mobile Bristol Application Framework. http://www.mobilebristol.com. Web Ontology Language. http://www.w3.org/TR/2004/REC-owlguide-20040210/ . JENA: A Semantic Web Framework for Java. http://jena.sourceforge.net/resources.html. SWRL: A Semantic Web Rule Language Combining OWL and RuleML. http://www.w3.org/Submission/2004/SUBM-SWRL-20040521/. Jabber Software Foundation. http://www.jabber.org. Jabber Enhancement Proposal 0108 – User Activity. http://www.jabber.org/jeps/jep-0108.html . Jabber Enhancement Proposal 0112 – User Physical Location. http://www.jabber.org/jeps/jep0112.html. Wildfire server, JiveSoftware. http://www.jivesoftware.org/wildfire/. Jabber Web Browser Client. http://webjabber.sourceforge.net/ . RFC 3921: Extensible Messaging and Presence Protocol (XMPP). http://www.ietf.org/rfc/rfc3921.txt. IST Advisory Group, Ambient Intelligence: from vision to reality. In: ISTAG 2003 consolidated report, 2003.
Copyright © 2006 The Authors
