The USC research team involved in the development of Obsidian, has just published the article Using Fuzzy Sets in a Data-to-Text System for Business Service Intelligence presented at the 16th International Workshop on Intuitionistic Fuzzy Sets and Generalized Nets.
It describes the techniques of fuzzy logic used in the automatic interpretation into natural language of the indicators presented by Obsidian in dashboards and reports, giving flexibility to temporal pattern detection algorithms, for example, in the comparison of monthly average values with the time series of historical values.
As can be seen in the following example, the text on the right is generated automatically from the 4 time series shown in the graphs on the left, and uses the fuzzy logic algorithms presented in this article for the elaboration of the text.
You can also watch a presentation about the automatic interpretation of indicators implemented in Obsidian.
This new publication joins the one already made in 2016 on process mining in service management tools, Process Mining in IT Service Management: A Case Study, publicada en [email protected] Nets/ACSD 2016: 16-30