How to Analyze Natural Language with Transparent Intensional Logic?

Logical analysis of natural language text is generally an under-specified task. However, within the project aiming at automatic processing of natural language (NL) text by means of logical analysis followed with the inference process, we need to have a "de facto" standard way for analysis of each NL sentence. First steps for introducing such standard are described in the presented text. The paper describes a semi-automatic way of building a corpus of logic formulae (constructions) in the formalism of the Transparent intensional logic (TIL) for real-world sentences in the Czech language. Output of a syntactic parser is used to determine the logical structure of the sentence and a verb valency lexicon is exploited for assigning TIL types. Using this information, an exemplary bank of TIL constructions is created automatically. This corpus of TIL constructions is then checked by human logic experts who iteratively consult the results with the respective theory of TIL transcription and the processing of input supportive lexicons. A user-friendly interface for such checking is presented at the end of the paper.