There are a multitude of freight databases, both public and private, that do not conform to any uniform standard or ontology. This makes the creation of one single authoritative ontology challenging, as the individual data sources do not and will not share a common vocabulary. To overcome this challenge a hybrid ontology 1 was developed where disparate data sources could be well described by their own unique ontologies and then mapped to a single upper-level “global” ontology. This approach allows for integration at the global ontology level to let all databases respond to the same queries, while remaining flexible enough to easily allow for data sources to be added or modified. This ontological structure was developed using the Role Base Classification Schema (RBCS). This schema uses two levels of classification at the global level: a primary grouping based on data category and a secondary grouping based on whether the data is an identifier or a descriptive feature. There are nine primary categorical groups: Time, Place, Link, Mode, Commodity, Industry, Events, Human, and Unclassified. Each of these primary groups has at most two subcategories for identifiers and features, which all data fields from local ontologies are then mapped to. In this way, the global ontology’s organization remains clear even when complex local ontologies are associated with it.
Referencing Page:
- 1Seedah, D. (2014). Retrieving Information from Heterogeneous Freight Data Sources to Answer Natural Language Queries. https://repositories.lib.utexas.edu/handle/2152/28341
- Log in to post comments