Concept-based Queries: Combining and Reusing Linguistic Corpus Formats and Query Languages

Abstract

This paper proposes a methodology for querying linguistic data represented in different corpus formats. Examples of the need for queries over such heterogeneous resources are the corpus-based analysis of multimodal phenomena like the interaction of gestures and prosodic features, or syntax-related phenomena like information structure which exceed the expressive power of a tree-centered corpus format. Query languages (QLs) currently under development are strongly connected to corpus formats, like the NITE Object Model (NOM) or the Meta-Annotation Infrastructure for ATLAS (MAIA). The parallel development of linguistic query languages and corpus formats is due to the fact that general purpose query languages like XQuery do not fulfill the changing needs of linguistically motivated queries, e.g. to give access to (non-)hierarchically organized, theory- and language-dependent annotations of multi modal signals and/or text. This leads to the problem that existing corpus formats and query languages are hard to reuse. They have to be re-developed and re-implemented time-consumingly and expensively for unforeseen tasks. This paper describes an approach overcoming these problems and a sample application.

Resources

Details