|Date:||Mon, February 15, 2016|
|Place:||Research II Lecture Hall|
Abstract: Human dialogues require quite a lot of different abilities, e.g. recognition of speech signals, language understanding, semantic analysis, automatic cognitive processes and deliberate cognitive behavior, language production, and utterance of speech signals. An appropriate mathematical model consists of a hierarchical structure with several layers, where each layer exchanges information with the layers above and below. Such a hierarchical model is realized in the UASR (Unified Approach for Speech Synthesis and Recognition), which is a long-term research project originally initiated by Rüdiger Hoffmann at TU Dresden. Within this structure, 'semantics' connects speech recognition to cognitive processes and cognitive behavior to language production. I present a mathematical framework for analyzing semantic uncertainty, which perfectly fits into the UASR. I demonstrate that 'rational' dealing with uncertainty of specific semantic units corresponds to calculations in an algebraic structure 'inert semiring.' In addition, it is shown that certain well-studied cognitive biases and different levels of rationality can be modeled mathematically by using different labeled hypergraphs whose vertices are semantic units.
The colloquium is preceded by tea from 16:45 in the Resnikoff Mathematics Common Room, Research I, 127.