Katsunori Oyama

Situ: A Situation-Theoretic Approach to Context-Aware Service Evolution

Evolvability is essential for computer systems to adapt to the dynamic and changing requirements in response to instant or delayed feedback from a service environment that nowadays is becoming more and more context aware; however, current context-aware service-centric models largely lack the capability to continuously explore human intentions that often drive system evolution. To support service requirements analysis of real-world applications for services computing, this paper presents a situation-theoretic approach to human-intention-driven service evolution in context-aware service environments. In this study, we give situation a definition that is rich in semantics and useful for modeling and reasoning human intentions, whereas the definition of intention is based on the observations of situations. A novel computational framework is described that allows us to model and infer human intentions by detecting the desires of an individual as well as capturing the corresponding context values through observations. An inference process based on hidden Markov model makes instant definition of individualized services at runtime possible, and significantly, shortens service evolution cycle. We illustrate the possible applications of this framework through a smart home example aimed at supporting independent living of elderly people.

Human-Intention Driven Self Adaptive Software Evolvability in Distributed Service Environments

Evolvability is essential to adapting to the dynamic and changing requirements in response to the feedback from context awareness systems. However, most of current context models have limited capability in exploring human intentions that often drive system evolution. To support service requirements analysis of real-world applications in distributed service environments, this paper focuses on human-intention driven software evolvability. In our approach, requirements analysis via an evolution cycle provides the means of speculating requirement changes, predicting possible new generations of system behaviors, and assessing the corresponding quality impacts. Furthermore, we also discuss evolvability metrics by observing intentions from user contexts.

Reasoning about Human Intention Change for Individualized Runtime Software Service Evolution

While software evolution has been studied extensively in software engineering, few of these efforts have involved a systematic exploration of human epistemological attitudes, such as human desire and intention, as the driving force of software service evolution. Our work proposes a theoretical framework to monitor and reason about human intention and its changes, which in turn can be used to determine how software and services should evolve to be individualized and better serve each user. Extending the Situ framework, we explore the service satisfiability problem through sub-world coverage following Kripke semantics, which enjoys wide application in AI and other fields related to human epistemic reasoning.

Requirements Analysis Using Feedback from Context Awareness Systems

User intentions to obtain services evolve over time. Since the changes of intention may occur at any time, some system requirements may become temporarily obsolete or contrary to a user intention. However, user intentions often indicate potential, valuable goals for upgrading the system to a new version or engaging new development. Our research tackles requirements analysis issues via feedback from context awareness systems to identify user intentions and capture instant definitions of goal in a robust manner. This paper presents a context-aware goal elicitation process by exploring the aspects of data, information, knowledge and wisdom. Furthermore, captured user intentions and goals are shown in a case study of healthcare system, and issues for the goal elicitation are explored.

A Human-Machine Dimensional Inference Ontology that Weaves Human Intentions and Requirements of Context Awareness Systems

Changing system requirements, especially for context awareness (CA) systems, often cause modifications in the software systems in order to adapt to dynamic environments. Since the requirements may become temporarily obsolete or contrary to human intentions, the CA systems need to be tuned to resolve the conflict. On the other hand, most CA design methods rely on pre-defined requirements and reasoning engine, thus, fail to address all the possible situations. Consequently, services provided by such a CA system are limited to accommodate some situations and unable to react as expected. Therefore, it is critical for CA systems to capture exceptions at runtime, infer changed human intentions, and adapt to these changes. This study focuses on inference of ever-changing human intentions and monitoring human intentions to handle system evolution. In this paper, we present an inference mechanism of human intentions via the human-machine dimensional inference ontology (HDIO). This ontology gives inference rules based on the BDI logic to deduce human intentions from contexts. Furthermore, the inference exercises of a healthcare system example shows how user intentions relate to system requirements and how they help improve self-adaptability of CA systems.

HESA: A Human-Centric Evolvable Situation-Awareness Model in Smart Homes

Situation Awarenessis a paradigm that contains data gathering, logic reasoning and responsive interaction to capture situations, which describe system states, from dynamic environments. Focusing on evolvability issue, we proposes a generic Situation Awareness System (SAS)model, called Human-centric Evolvable Situation-Awareness (HESA). In this paper, we present the HESA architecture, accommodation/evolution processes, and how we use Context/Situation Generation approach to provide human-centric fault-tolerant evolvability.

On Distributed Run-Time Software Evolution Driven by Stakeholders of Smart Home Development (Invited Paper)

Stakeholders of smart home development often request personalized requirements based on existing services. Requirements under the environment may change at any time even when services are provided correctly, since the stakeholders eventually change their intentions. Evolvability is essential to adapting to the dynamic and ever-changing requirements in response to contextual feedback. To support software evolution for real-world applications in distributed service environments, this paper introduces architecture for distributed run-time software evolution and discusses issues of software evolution in Smart Home environments. Smart Home environments exist in feature-rich networks, and should be ideally maintained with updates at run-time.

Situation-Oriented Requirements Elicitation

In this paper we present a new human-centered requirements elicitation methodology that effectively considers end-users desire, behavioral and environmental contexts. We follow a methodology that uses a computationally rich definition of situation as a 3-tuple where d denotes human desire, A denotes the action vector, and E denotes the environment context vector. The proposed method of human-centered requirements elicitation is based on the situation -- transition structure which is a directed weighted graph that represents transition from one situation to another. We illustrate the proposed methodology through some case studies with open access data sets. Requirements thus elicited appear to be valid after manual inspection. Future directions along this line of research are then asserted.

Integration of 3D Trajectory Maps into a Local Distribution Map of Radiation Dose Using Unmanned Aerial Vehicle

Environmental situation awareness in this study focuses on monitoring techniques to discover public safety and potential risks at local spots. Changing amount of radiation dose recently has become a great concern by citizen near nuclear power plants. Observation approach to resolution of the concern in this study is the application of Unmanned Aerial Vehicle equipped with a situation monitor in order for covering observation points at the local spots which cannot be accessed by walking or wheeled vehicle. This paper presents an ontology based map integration process that merges 3D trajectory maps into a local distribution map in Google earth. 3D to 2D projection is introduced to visualize changes in spatial situation at ground level by calculation of radiation dose based on distance decay.

A Concept Lattice for Recognition of User Problems in Real User Monitoring

User problems encountered during the use of a software product are often hard to identify even after the software product is thoroughly-tested and then released. There are inevitably unexpected situations introduced or triggered by transient use patterns, however, these unexpected situations can be hardly eliminated due to various human factors on user interaction. This study of situation-oriented real user monitoring develops an application of concept lattice to represent use patterns from user interaction and perceive potential user problems. Data structure of user problem is modeled based on user problem ontology. Once a user problem is recognized, observation data of the use patterns are the subject for further analysis. In this paper, use patterns in a file upload service are used to demonstrate how the concept lattice is used and discuss about validity of the causal factors found in the concept lattice.