Relational Representation of Uncertain and Imprecise Time Assess-ments: An Application to Artworks Dating.

Abstract

Imprecision and uncertainty appear together in many situations of real life and thereforesoft computing techniques must be studied to tackle this problem. Imprecise and uncertainvalues are usually expressed by means of linguistic terms, specially when they have beenprovidedbya human being. This is also the case of temporal information where, inaddition to handling time constraints, we may also have both uncertainty and imprecisionin the description, like in the sentence”It is very possible that Giotto’s Crucifix was paintedby 1289”. To manage both uncertainty (very possible) and imprecision (by 1289) in aseparate way would lead to a quite complicated computation and a lack of comprehensionby the users of the system. Because of these reasons, it is very desirable that bothsources of imperfection of time values are combined into a single value which appropriatelydescribes the intended information. In this work, we extend our previous research on thistopic and we study how to adapt it to relational systems in order to be useful. The finalgoal is obtaining normalized fuzzy values that provide an equivalent information about thedescribed temporal fact than the original ones, for making it possible to store and managethem in a fuzzy relational database. On the other hand, there will be some situationswhere more than one expert opinion about a time period must be taken into account andwe need to find a representative value of them all in order to be stored and managed. Forthe sake of simplicity, comprehensibility and the efficiency in computation (when usingtrapezoidal representation), the fuzzy average is used to find such a representative value.