On Temporal-fuzziness in Temporal Fuzzy Databases

We propose a framework for a database model unifying both imprecision and time aspects of data. Particularly, we emphasize that the fuzzy meanings of linguistic data can change with time; we call this property temporal-fuzziness. The major problems arising from the lack of the ability to handle this property have been studied, i.e. users and databases misinterpret the meaning of data. Thus, it is essential to treat temporal-fuzziness within an appropriate framework like ours. In our work, we employ concepts based on fuzzy set theory — possibility theory and linguistic variables — for modeling and evaluating uncertain/imprecise data from three domain types, i.e. quantitative, qualitative, and multivalued logic data domains. To model past data, the tuple time stamping method and the discrete time conceptual model are utilized. We introduce three kinds of measures for each domain type to evaluate data. They can provide upper bounded, lower bounded, and approximate answers to queries. Then, the Temporal Fuzzy Data Model and a metadatabase are proposed. The concept of metadatabase allows the time-variant property of fuzzy meanings to be modeled. Since based on the relational model, our model has a uniform and simple structure.