DEVELOPMENT OF A TRUST MEASUREMENT MODEL IN SERVICE COMPUTING SYSTEMS

• Main Author: Biantoro, Yudhi
• Format: Dissertations
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/79129
• Institution: Institut Teknologi Bandung

The rapidly growing role of information technology in fulfilling various human needs is inevitable. Such rapid development relies on computing based services. The success of information technology is based on the interaction between computer dependent entities involved in business services. Service computing requires dynamic entity interactions to facilitate the achievement of optimal business goals. Therefore, the relationshi p between the entities involved depends on trust. When trust is high, service computing transactions can be greatly enhanced. However, when it is on the decline, transactions become few to non existent. Based on the interaction dynamics, the role of trust is very important in various fields, including the service computing field. For this reason, it is necessary to measure trust for the sustainability of business services. Based on several studies, there are many definitions of service computing according to each researcher's point of view, but there is a common view that service computing uses computing technology. It can be said that service computing is an encapsulation o f the application of computing technology. Thus, various computing technologies used to ensure the continuity of business services are part of the service computing system. Many researchers in the field of computing technology had undertaken study on trust models. Various trust measurement models have been used and applied in various technology fields such as IoT, big data, QoS, and so on. Based on the implementation of computing technologies that are developing today, it shows that various computing technologies ar e part of service computing. However, research on trust models in service computing systems is not explicitly available, only in the technology area, so there is an opportunity to develop trust models in service computing systems. For this reason, the opportunity to build a trust measurement model in service computing systems is open. To build a trust measurement model for service computing systems, it is necessary to first study the relationship between computing technology and service computing. This research proposes a trust measurement model for service computing systems based on the extraction of various existing models, techniques, and components of trust in computing technology from previous studies. The extraction process was conducted using meta analysis technique by extracting various models, techniques, and components from trust models of computing technology. The extraction resulted in superior groups called domains. The resulting domains are converted into mathematical formulas. The elements forming the domains need to be described based on their reference sources, so that the mathematical formulas can be calculated. This research produces a generic model of trust measurement in service computing systems that aims to be more flexibly applied to service computing with various fields of computing technology. This research also produces a trust threshold value of 0.67 wit h machine learning techniques. There are four learning machines that are used as candidates to determine the trust threshold value. Based on the analysis, the random forest technique is the best technique, and is then used to determine the threshold value. The threshold value of trust is needed to know the limit of untrust and trust. Values above 0.67 mean trust, and conversely values below 0.67 mean untrust, while the value of the trust parameter with a range between 0 and 1 is presented for each entity in volved in the interaction in the test conducted. Model testing using the available dataset consists of seventy six entities that interact with each other in a business purpose. The testing parameters include SIC, MC, SRT , and SW which are superior domains obtained from reference sources, while STV is a novelty parameter containing trust values. Furthermore, it is also presented how the workflow of implementing the model aims to facilitate the development of a trust measurement model system for programmers.