WWW/Internet - Portal do Software Público Brasileiro

ISBN: 978-972-8939-25-0 © 2010 IADISstudy in which TrailM was integrated with a ubiquitous commerce system (UbiTrade (Franco et al, 2009))and used for contextualized discovery of deal opportunities.The article is organized into six sections. Sect. 2 describes the TrailM architecture. The third Sect.discusses the TrailM use in the ubiquitous commerce guided by trails. Sect. 4 presents the TrailM and theUbiTrade prototypes as well as the integration between them, and still describes and evaluates the resultsobtained from a case study involving trails in discovery of deal opportunities. Sect. 5 discusses related worksand the contribution of this work. And finally, Sect. 6 presents the conclusions and the future works.2. THE TRAILM SYSTEMThis section presents the trails management system. Section 2.1 describes the main terms used by the system,mainly those related to the trails composition, and finally, Section 2.2 presents the system architecture.2.1 Entities, Trails, Ptrail, and TrailpointTrailM was designed to manage entities’ trails. In this context, entities represent persons, accessingcomputational resources (for example, a smartphone), or mobile objects (for example, a vehicle). We chooseentity, instead of user, because then the model could manage trails of any mobile object. In this scenario,different applications can be developed for trails. For example, a vehicle could have a trail, recordinginformation considered relevant, such as the occurrences of maintenances or failures.Applications that use TrailM must temporally store data that composes a trail. For this reason, we proposethat the applications have internally a structure called ptrail (piece of trail), which is composed of thefollowing attributes: entity, resource, event, context, and location. The values of the entity and resourceattributes are stored statically in the ptrail after that an entity logs in the model. Nonetheless, the values of theevent, context, and location attributes are automatically updated as the actions or movements made by theentity and monitored by the application. The records related to the contexts visited by an entity are stored inonly one trail. The trail is composed of a sequence of records of the ptrail. A record is composed of sevenattributes organized into three categories: identification, content, and temporality. The first category containsonly one attribute to identify the trail. The second category is used to store the ptrail content. The lastcategory contains date/time attribute, representing the time of the record creation.We named trailpoint the process that makes the trail composition. This process sends the valuescontained in the ptrail to be recorded in the entity’s trail, which is stored in a server. The trailpoint occurswhen the application automatically identifies that an entity performed an event, for example, entry or exitfrom a location, interaction with another entity, access to a file, among others. Figure 1 shows the trailcreation of a mobile device user. Each star represents the occurrence of an event, which causes a trailpoint.2.2 The System Architecture2.2.1 General OrganizationFigure 1. Example of trail compositionFigure 2 presents a general view of the system, which is composed of a server (TrailM Server) and a client(TrailM Client). In addition, TrailM considers the existence of an external provider that provides locationinformation. Next sections describe all these components, justifying why they are necessary.20