On the computation of semantically ordered truth values of linguistic protoform summaries

摘要

Linguistic summaries provide a promise for extracting useful information from large datasets and present them in the form of natural language. Their applications can be found in various disciplines such as eldercare, financial time series etc. In this work we focus on linguistic protoform summaries and point out at some problems associated with the truth value computation methods present in literature. We develop a technique which produces more intuitive truth values for three different kinds of linguistic protoform summaries and illustrate this with help of some examples. Moreover, we show through a mathematical proof that our method is very robust and the truth values always follow the semantic order of the language they are representing.