TR-2013-036


Krištof Anetta, Martin Takáč:  Towards Advanced Text Quality Evaluation Modelling




Abstract:  This paper tackles the phenomenon of judging text quality from the point of view of AI and cognitive science. Building on relevant literature, it starts with a general view of current computer text quality evaluation and then, discarding the fields where sufficient amount of research is taking place, narrows its focus down to questions of advanced text quality evaluation (literary style in particular). It suggests solutions on several levels, starting with subsemantic neural network evaluation and going through possibly implementable rules from writers‘ and editors’ practice both on the subsemantic and semantic level, extracted by means of content analysis out of a considerable corpus of mostly informal Internet articles.