Improving Human-Computer Interaction by Developing Culture-Sensitive Applications Based on Common Sense Knowledge

Abstract

Computing is becoming ever more present in people's everyday life. Regarding this fact, researchers started to think of ways for improving Human-Computer Interaction (HCI) so that computer applications and devices provide users with a more natural interaction, considering the context which users are inserted into, as humans can do (Harper et al., 2008). Developing culture-sensitive interactive systems seems to be a possibility for reaching such goal. For them, culture is a system of shared meanings which forms a framework for solving problem and establishing behavior in everyday life that has to be considered in interactive system development. Since cultural knowledge is often implicit, designers often have trouble even realizing that their designs carry cultural dependencies implicitly. Moreover, it is not possible to design by hand for every combination of possible cultures, nor is it practical to exhaustively test for every possible user culture A support is necessary for making this possibility come true and information and communication technology can offer it, as it is explained further ahead. This chapter presents the solutions of the Advanced Interaction Laboratory 1 (LIA) from the Federal University of So Carlos (UFSCar), Brazil, for developing culture-sensitive interactive systems. LIA's approach relies on using common sense knowledge for developing such kind of systems. That is because individuals communicate with each other by assigning meaning to their messages based on their prior beliefs, attitudes, and values, i.e. based on their common sense. Previous researches developed at the Lab have shown that common sense expresses cultural knowledge. So, providing this kind of knowledge for computers is a way of allowing the development of culture-sensitive computer applications (Anacleto et al., 2006a). One idea for providing computers with common sense is to construct a machine that could learn as a child does, observing the real world. However, this approach was discarded after Minsky and Papert's experience of building an autonomous hand-eye robot, which should perform simple tasks like building copies of children's building-block structures. In this 1 LIA's homepage (in Portuguese): http://lia.dc.ufscar.br www.intechopen.com Human-Computer Interaction, New Developments 2 experience, they realized that numerous short programs would be necessary to give machines human abilities as cognition, perception and locomotion and that it would be very difficult to develop those programs Another idea is to build a huge common sense knowledge base, store it in computers and develop procedures that can work on it. This seems to be an easier approach; nevertheless there are at least three big challenges that must be won in order to achieve it. The first challenge is to build the necessary common sense knowledge base, since it is estimated that, in order to cover the human common sense, billions of pieces of knowledge such as knowledge about the world, myths, beliefs, and so on, are necessary (Liu & Singh, 2004). Furthermore it is known that common sense is cultural and time dependent, i.e. a statement that is common sense today may not be a common sense statement in the future (Anacleto et al., 2006b). For instance, consider the statement "The Sun revolves around the Earth". Nowadays this statement is considered wrong, however, hundreds of years ago people used to believe that it was right. One possible idea to transpose this difficulty is to build the knowledge base collaboratively by volunteers through the Web, since every ordinary people has the common sense that computers lack (Liu & Singh, 2004, Anacleto et al. 2006a). In order to make the collection process as simple as possible to the volunteers, it is kind to think of collecting the common sense statements in natural language. Then the second big challenge arises: to represent the knowledge collected in natural language in a way that computers can make inferences over it.