Staniszkis, Witold and Staniszkis, Eliza. 2003. Intelligent agent-based expert interactions in

AN INTRODUCTION TO KNOWLEDGE-GROWING SYSTEM: A REVIEW ON THE NOVEL FIELD ON ARTIFICIAL INTELLIGENCE Arwin Datumaya Wahyudi Sumari, A.S. Ahmad, A.I. Wuryandari, J. Sembiring Departemen Elektronika, Akademi Angkatan Udara Sekolah Teknik Elektro dan Informatika, Institut Teknologi Bandung arwin91aauyahoo.co.id , asaisrgyahoo.com , acieklskk.ee.itb.ac.id , jakadepkominfo.go.id ABSTRACT The field of Artificial Intelligence AI has been progressing very advanced since its introduction in 1950s. The essential matter of AI is how to build an entity that mimics human intelligence in the way of learning of something to gain knowledge of it and use the knowledge to solve problems related to it. In this paper we review a novel field on AI called Knowledge-Growing System KGS that is, a system that is capable of growing its knowledge along with the accretion of information as the time passes. KGS is developed from our observation of how human’s brain gain new knowledge by fusing information gathered and delivered by human’s sensory organs. The result of the observation is a new knowledge- growing algorithm called as Observation Multi-time A3S OMA3S information-inferencing fusion method. Some examples will be delivered to show the application of KGS to real-life problems. Keywords: AI, information-inferencing fusion, KGS, OMA3S, INTISARI Bidang Kecerdasan Tiruan telah mengalami kemajuan pesat sejak diperkenalkan pada tahun 1950-an. Hal esensial dari Kecerdasan Tiruan adalah bagaimana membangun suatu entitas yang menirukan kecerdasan manusia dalam cara pembelajaran terhadap sesuatu untuk meningkatkan pengetahuan tentang sesuatu tersebut dan menggunakan pengetahuan tersebut untuk menyelesaikan permasalahan-permasalahan yang terkait dengannya. Dalam makalah ilmiah ini kami mengkaji satu bidang baru dalam Kecerdasan Tiruan yang disebut Sistem Berpengetahuan-Tumbuh SBpT yakni sistem yang memiliki kemampuan untuk menumbuhkan pengetahuannya seiring dengan bertambahnya jumlah informasi seiring dengan berjalannya waktu. SBpT dibangun dari pengamatan terhadap cara otak manusia dalam memperolah pengetahuan baru dengan cara memfusikan informasi yang dikumpulkan dan dikirimkan oleh panca indera manusia. Hasil dari observasi tersebut adalah algoritma penumbuhan- pengetahuan baru yang disebut dengan metoda fusi penginferensian-informasi Observation Multi-time A3S OMA3S. Beberapa contoh akan disampaikan untuk memperlihatkan aplikasi SBpT pada permasalahan-permasalahan di dunia nyata. Katakunci: Fusi penginferensian-informasi, kecerdasan tiruan, KGS, OMA3S INTRODUCTION Building an intelligent entity that is capable of mimicking human intelligence in some aspects has been a challenging research all over the world. Moreover, studies and research in field has been involving many science and engineering disciplines such as Philosophy, Psychology, Cognitive Science, Computer Science, Mathematics and Engineering as depicted in Figure 1 [1]. Of course, the approaches or techniques delivered by diverse researchers to emulate human intelligence were difference because they modeled the human intelligence based on their understanding of it combined with the science-and- engineering bases they had. Refer to [2], McCulloch-Pitts viewed the human intelligence comes from the mechanism occurs in the human’s nervous system in a model of human brain’s neuron known as Artificial Neural Networks ANN. Lotfi Zadeh saw that human beings do not always think crisply, that is, “yes” or “no”, about a phenomenon but tend to think in between. Based on this reason he coined Fuzzy Set theory that since then it has been used widely in designing and implementing intelligent-based systems. Figure 1 : Disciplines of AI Human beings also have an ability to find the best compromise solution among several given alternatives and it is called as optimization problem. This perspective gave a birth to a new approach called as Evolutionary Computing where Genetic Algorithm GA is a part of it. Another approach that comes purely from mathematics field is Probabilistic Reasoning PR. The world is very unpredictable and full of uncertainties. In order that a system can apprehend this situation and solve the problems, it has to deal with these uncertainties. In solving problems, human being always tries to find the most likely solutions that he has ever experienced in his life. These experiences are stored in his brain in form of knowledge that is grown along with the accretion of information he sees, hears, etc from his environment as the time goes by. Suppose, he is uncertain with the best available solutions, he will gather or collect more information in order to obtain comprehensive information regarding the problems. This mechanism is performed repeatedly until he ascertains with the best results to be taken as the basis for making a decision or an action. The illustration previously described shows that human being gets more intelligence when his knowledge grows from nothing to some extent that makes him able to apprehending the phenomena in his environment. Knowledge can be grown if there is information delivered to and processed by the brain. The comprehensive information can be obtained by combining or fusing information from all sensory organs by a technique called information-inferencing fusion. The comprehensive information will become his new knowledge regarding the phenomena he observes. The whole mechanism in obtaining the new knowledge is called as Knowledge-Growing KG. Our endeavor in this field has come up with an emulation of the KG in human brain called Knowledge-Growing System KGS with Observation Multi-time A3S OMA3S information- inferencing fusion method as the KG mechanism [3]. Regarding to this matter, the structure of the rest of the paper is as follows. Section II covers the literatures review regarding to the state-of-the-art of AI research especially that is related to the emulation of how brain obtains new knowledge. In Section III, we will deliver the development of KGS and followed by Section IV where the examples of KGS application will be presented. The paper converges in Section V with some concluding remarks. LITERATURES REVIEW Some literatures reviews have been given in Section I. In this section we focus on two important things that form the KGS namely information fusion and AI. Up to now it is difficult to find literatures which review the relation between information fusion and AI, even though the essential concept of information fusion itself adopts the mechanism occurred in living things’ brain to obtain more complete information. Most of the applications of information fusion are for targets tracking and identification with the ultimate aim to deliver comprehensive information for decision-making purposes, see [4] for detail.

1. The Intelligent Behavior of the Human Brain’s Information-Inferencing Fusion

We did more deep investigation on the mechanism occurred during information fusion in the brain as depicted in Figure 2. We have concluded that there is an intelligent behavior performed by the brain when fusing the information gathered from the environment and delivered by the sensory organs which is in this case called as the information multi-source. The process of obtaining new knowledge from the delivered information follows three consecutive processes namely information fusion, information inferencing, and information-inferencing fusion [3]. Figure 2 : Human information-inferencing fusion system [5] On the first process, information delivered from the sensory organs is fused to gain comprehensive information regarding the observed phenomenon. The result of this process is then reasoned to obtain the inferencing of the fused information. The inferencing at this stage is the inferencing regarding the phenomenon at the first observation or at time t. The next observation at the next subsequent time will result in inferencing at time t+1 and so on. The comprehensive information after some observation times is obtained by fusing the collection of the information-inferencing over time to become fused information-inferencing. This comprehensive information is called as new knowledge of the observed phenomenon [6].

2. The Topology of AI

Some literatures on the state-of-the-art of AI approaches have been delivered in [2, 6], but up to now we still have not found any single literature that studies the relation between human intelligence and information fusion. Therefore, the challenge in emulation the KG mechanism in human brain is what is the most appropriate approach that will be selected for this purpose [6]. Ahmad 2006 in [2] proposed a topology of AI which consists of three categories, namely smart systems, knowledge-based systems, and computational-based systems. One of subcategories under knowledge-based systems is intelligent programming which is aimed to emulate the human’s process of thought. In this approach, the knowledge acquisition is done by means of thinking process which is formulated in form of algorithms and implemented on computer software. The Ahmad’s topology of AI is depicted in Figure 3.