Of Politics, Behavior and Commands: Processing Information Unspoken for Sentiment Analysis and Spoken Interaction Applications

The registration and processing of information not uttered contributes to the enrichment of data in HCI applications and other forms of spoken interaction, including subtle emotions in Sentiment Analysis-Opinion Mining and, also, additional implementation in spoken commands i.e. for technical texts. The present approach and interactive application target to account for the new insights and factors and their integration into knowledge graphs with subsequent use in training data and neural networks. The main focus is on the data preparation stage for subsequent extensive implementation and quantitative evaluation. Insights from crowd-sourced data enable a differentiation between perceived linguistic and paralinguistic information not uttered compatible to language-specific and socio-cultural norms and perceived linguistic and paralinguistic information not uttered that is either strictly circumstantial/individual-specific or strictly domain/context dependent. The latter also applies to spoken commands. This, subsequently, enables a differentiation between circumstantial factors/evidence and socio-culturally-biased factors/evidence in data analysis and training data and its integration in knowledge graphs, outlining the distinct types of implementation for the enrichment of models and refining NLP tasks.