Previous approaches of analyzing spontaneously spoken language often have been based on encoding syntactic and semantic knowledge manually and symbolically. While there has been some progress using statistical or connectionist language models, many current spoken-language systems still use a relatively brittle, hand-coded symbolic grammar or symbolic semantic component.In contrast, we describe a so-called screening approach for learning robust processing of spontaneously spoken language. A screening approach is a at analysis which uses shallow sequences of category representations for analyzing an utterance at various syntactic, semantic and dialog levels. Rather than using a deeply structured symbolic analysis, we use a at connectionist analysis. This screening approach aims at supporting speech and language processing by using 1 data-driven learning and 2 robustness of connectionist networks. In order to test this approach, we h a ve developed the screen system which i s based on this new robust, learned and at analysis.In this paper, we focus on a detailed description of screen's architecture, the at syntactic and semantic analysis, the interaction with a speech recognizer, and a detailed evaluation analysis of the robustness under the in uence of noisy or incomplete input. The main result of this paper is that at representations allow more robust processing of spontaneous spoken language than deeply structured representations. In particular, we show h o w the fault-tolerance and learning capability of connectionist networks can support a at analysis for providing more robust spoken-language processing within an overall hybrid symbolic connectionist framework.
Word graphs are able to represent a large miraber of different utterance hypotheses in a very compact manner. However, usually they contain a huge amount of redundancy in terms of word hypotheses that cover almost identical intervals in time. We address this problem by introducing hypergraphs for speech processing. Hypergraphs can be classified as an extension to word graphs and charts, their edges possibly having several start and end vertices. By converting ordinary word graphs to hypergraphs one can reduce the number of edges considerably. We define hypergraphs formally, present an algorithm to convert word graphs into hypergraphs and state consistency properties for edges and their combination. Finally, we present some empirical results concerning graph size and parsing efficiency.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.