A query language for discovering semantic associations, Part II: sample queries and query evaluation

Abstract

In our query language introduced in Part I (Niemi & Jämsen, in press) the user can formulate queries to find out (possibly complex) semantic relationships among entities. In this article we demonstrate the usage of our query language and discuss the new applications that it supports. We categorize several query types and give sample queries. The query types are categorized based on whether the entities specified in a query are known or unknown to the user in advance, and whether text information in documents is utilized. Natural language is used to represent the results of queries in order to facilitate correct interpretation by the user. We discuss briefly the issues related to the prototype implementation of the query language and show that an independent operation like Rho (Sheth et al., 2005; Anyanwu & Sheth, 2002, 2003), which presupposes entities of interest to be known in advance, is exceedingly inefficient in emulating the behavior of our query language. The discussion also covers potential problems, and challenges for future work.

An approach for developing a schemaless XML dataspace profiling system

“…Contrary to conventional query formulation, which requires that the user knows the semantic relationships in advance and, based on this knowledge, specifies the navigation among the available data, the goal in the discovery of unknown semantic relationships is to provide the user with navigation paths which make up these relationships. These kinds of paths are generally called semantic associations [47–49]. Formally, a semantic association is a sequence of binary relationships in which each binary relationship expresses a basic semantic relationship between two entities.…”

“…The chains of semantic associations can be arbitrarily long. In [48, 49], we introduced a query language capable of constructing semantic associations based on the underlying basic relationships. In our future work, we aim to study how these basic associations can be extracted through XML dataspace profiling.…”