Context-Aware Fuzzy Databases

Highlights

• We have proposed an architecture of Context-Aware Fuzzy Database.

• We intend the development of intelligent, flexible and customized systems.

• Linguistic labels can be conveniently adapted according to the context.

• Users interacts with the system by means of SQL:99 with a context-aware behavior.

• We have shown a concrete medical application for a gait laboratory.

Abstract

The development of mechanisms to ease human machine interaction is an issue about which there is increasing interest within both the software world in general, and database systems in particular. A way to tackle this problem is to try to approach the natural way of user expression. The Fuzzy Sets Theory and its application to build Fuzzy Databases constitute a consolidated advance in the literature. Another way is to adapt the interaction of the system to the context where it is running. In this sense, this paper presents an approach to build a model of Fuzzy Databases that dynamically adapts to user context. In order to do this, we have studied the management of the context in Fuzzy Database applications and we propose an architecture for the development of intelligent, flexible and customized context-aware database systems. We also present a proof of concept implementation to be used with SQL:99 standard in Oracle ORDBMS. Finally, through a real application in the medical area, we demonstrate the feasibility of the proposal.

Introduction

The decision making process based on user information is quite complex. One of the main reasons for this complexity is the huge amount of available data, known as flooding effect [1]. In recent years, an area of Information and Communication Technologies (ICT) research has focused on increasing efficiency in obtaining relevant information to the user through the simplification of the query formulation process and the improvement of the relevance of the given results, producing intelligent systems that become more flexible and adapted to user needs [2].

Natural languages are intrinsically imprecise because they usually represent human perceptions. In the human communication process the use of descriptive terms (linguistic labels) plays a critical role. In general, imperfection (imprecision, vagueness or uncertainty) is present in all real world data [3]. Significant results have been achieved in the representation and processing of information affected by human expressiveness and subjectivity. In this situation, the Fuzzy Set Theory [4] has produced remarkable results in information systems for data modeling and flexible query processing [5], [6], [7].

The paradigms to develop systems are changing due to the advances in ICT, specially in mobile computing; consequently, system designers must consider new important variables (as the context and user preferences) to develop more dynamic and personalized applications. For example, smartphones and tablets have multiple sensors (such as GPS, gyro, proximity, accelerometer, barometer, heart rate, among others) that make it possible to develop applications that take into account both the information directly provided by the user, and the contextual information provided by these sensors.

Thus, a valuable vehicle for the realization of intelligent systems is the study of the context. Due to the effects that this concept has on human behavior, its proper simulation in computer systems becomes crucial to strengthen communicating, processing and managing information capabilities. Context-aware database applications are needed not only in scenarios where different types of users interact with the same data, but also where the expectations of the same user in her/his interaction with the system can significantly vary depending on information regarding, for example, her/his location in time and space.

In this paper, we use Fuzzy Databases with the purpose of facilitating context-aware interaction with imperfect data. Our main objective is to bring the indexicality of human language [8] to applications of Fuzzy Databases, providing context-aware applications in order to emulate what happens in communication among people.

The first contribution of this research is the introduction of an architecture for Context-Aware Fuzzy Databases. As we will see, thanks to the use of additional intelligent modules, the system can adaptively respond considering the current context and multiple customized settings for each user. The context can be determined by many affecting factors. The architecture that we propose is based on this idea of varying factors.

As a second contribution, we have also designed a proposal to enhance an existing Object Relational DBMS (Oracle, in this case) with the presented architecture. In this proposal, each user defines her/his own semantics for the labels that will guide her/his communication with the system. The proposed architecture is used to build a fuzzy database system where the context influences user communication with the system. An intelligent agent acts as a mediator that makes a transformation between the user's linguistic expressions and the raw data stored into the fuzzy database. In this way, objects are stored into the fuzzy database taking into account the context where the data input occurs and, when a query is run on this data in a different context, the system presents the information conveniently adapted to it. All the interaction with the system has been user friendly implemented for those familiar with the SQL:99 standard.

The aforementioned development of context-aware database systems capable of managing imperfect information is a real necessity in many fields. The medical area is an example where professionals make decisions based on imperfect data and the context plays a vital role. Therefore, the systems that provide doctors with data should be context-aware and should be able to manage data imperfection. In this way, decisions made by practitioners will be more appropriate as they will consider dependable information.

As a third contribution of this paper and in order to demonstrate the feasibility of our proposals, a real example has been implemented in the medical field. In this example, each user will define the semantics of descriptive terms, allowing for customized input data and for performing flexible queries according to her/his needs. This proof of concept implementation uses our proposal to provide an advanced data management tool for the study of the human gait, a medical concern in which the muscle skeletal joint behavior is quantified, and the repercussion of the neuromuscular commitment of the patients with pathological gait is shown.

Our work intends to be a contribution to the computational areas of Human Computer Interaction and Soft Computing. We propose the development of intelligent systems with the goal of filtering the overwhelming amount of information due to the advances of the ICT and mobile computing. We utilize and manage the user's context to personalize data storage and information retrieval when a user interacts with fuzzy databases.

This research highlights the value of the human perspective in actual computer systems; this is one of the reasons why we think the study of the context, in specific computing areas like Fuzzy Databases, is of primary interest. Our proposal is part of the guiding principle of Soft Computing which, according to Zadeh [9], exploits the tolerance for imprecision, uncertainty, and partial truth to achieve tractability, robustness and low solution cost. We use tools such as Fuzzy Databases, Object Oriented Data Modeling and Object Relational Database Management Systems (ORDBMS) to meet this challenging goal.

We adopted a strategy of non-experimental design, with top-down analysis, to address the problem presented in our study. The research has been carried out without the deliberate manipulation of variables, that is, the phenomena have been observed and analyzed in its natural environment.

Our model is product of a refining process by applying a method of successive approximations. First, we propose an architecture for the development of context-aware fuzzy database systems. Then, according to the proposed architecture, we develop a theoretical framework for user customizable semantics based on the Computational Theory of Perceptions, and the Object Relational Database Management System. Finally, we develop a proof of concept in the medical area to prove the feasibility of the theoretical framework proposed.

In Section 2, we review the theoretical background and the previous work about the context in Information Systems and Database Querying. In Section 3, we show the proposed architecture for Context-Aware Fuzzy Database applications, providing an explanation of fuzzy domains, context modeling and the data dependency model. In Section 4, we present a theoretical framework to achieve dynamic adaptation of the user's language when interacting with an ORDBMS. In Section 5, we develop the implementation of a proof of concept in the Children's Orthopedic Hospital of Venezuela. Finally, in Section 6, we present the concluding remarks and guidelines for future work.