OBDA and Intermodal Logistics: Active Projects and Applications

In this paper we present the current state of affairs about our funded research projects concerning the investigation of Ontology-Based Data Access in the context of intermodal logistics. This application domain is particularly challenging for two main motivation. On the one hand, it is characterized by very large amounts of data; on the other hand, the design of a conceptual layer must strike a balance between efficient but potentially simplistic models, and sophisticated, but potentially highly inefficient ones. 1 Context and Motivation As reported in [3], the keyword Ontology Based Data Access (OBDA) defines scenarios where an ontology mediates the access to data which is physically stored in a (relational) data base (DB). In the last two decades, ontologies are playing an increasingly important role in different application domains – e.g., Semantic Web [1], medical knowledgebases [2] and software engineering [9] – to support knowledge sharing, representation and reasoning. The main feature of OBDA applications is the ability to deal with large amounts of data organized according to conceptually simple data models, and subject to specific and repetitive queries. In this paper, we show our current experience about practical uses of OBDA by considering business-size applications to Intermodal Logistic Systems (ILSs) related to two research grants of ours. The first application is about an information system which monitors operations in a network of intermodal terminals. The key feature of the network is to support reception, storage and shipping of goods packaged in containers – also, Intermodal Transport Units (ITUs) – so that the movement of goods by successive modes of transport happens without stuffing or stripping operations. The goal of the information system is to support decision making, e.g., by timely detecting incidents or performance degradations. This is achieved by collecting various data and events about the network and then computing Key Performance Indicators (KPIs). The second project concerns the design of a Decision Support System (DSS) for Cagliari International Container Terminal (CICT, http://www.cict.it). The DSS supports the terminal operator in critical decisions like choosing the berth to allocate to an incoming vessel, or the place whereto a given ITU must be assigned in a storage yard. Here we focus on the knowledge-based component of the DSS which fulfills two main task, i.e., monitoring the terminal information – as in the previous project – and helping to discover new relevant data that could be used to optimize a set of operations. In Section 2 we discuss both applications in detail. In both the aforementioned applications, the main focus of our investigation is the opportunity to build a monitoring information system using OBDA instead of mainstream technologies, e.g., relational databases. Some practical advantages of ontologies vs. relational data models include the ability to cope naturally with special kinds of relations, including taxonomies and part-whole relationships, and, in our case, the ability to handle heterogeneous attributes and NULL values. Even if the usage of OBDA could be a key enabler for intelligent data manipulation and integration, actual performance represents an outstanding issue. Current state-of-the-art ontology-based reasoning tools show acceptable performance and scalability in many cases of practical interest, even if reasoning about ontologies can be computationally (much) harder than querying a relational database. However OBDA remains challenging in scenarios characterized by large amounts of data like the ones we consider, and so the question of whether they can effectively support business-size applications is still open and worth investigating. In Section 3 we show our current progresses in this direction. In Section 4 we conclude the paper with some final remarks.