Mining information from time series in the form of natural language expressions

摘要

In this paper, we discuss three tasks of mining information from time series, namely finding intervals with monotonous trend, evaluation of it in sentences of natural language, and summarizing information using intermediate quantifiers. The mined information is presented in simple sentences of natural language. For estimation of the trend of time series, we apply the theory of first-degree F-transform. For generating sentences of natural language we apply the formal theory of fuzzy natural logic. Then the former are obtained by interpretation of special formulas.