Commonsense aboutness for information retrieval.
Bruza, Peter D.
MetadataShow full item record
Information retrieval (IR) is driven by a process which decides whether a document is about a query. Recent attempts have been made to formalize properties of “aboutness”, but no consensus has been reached. The properties being proposed are largely being determined by the underlying model. Once the framework has been fixed, certain aboutness properties could be implied from it. Moreover, some properties may be sound only within a certain IR model, and some may lead to negative effects to the performance of IR systems. In this paper, by adopting an information-based and abstract framework, model independent, commonsense properties of aboutness are investigated. A set of properties characterizing commonsense aboutness and its dual - non-aboutness is introduced. Issues such as the soundness and completeness of aboutness inference are discussed.