Hideki Hara

Justifications for bullying among Japanese schoolchildren

The present research investigated the justifications for bullying as a function of the various roles that children assume in bullying. Contrary to expectations, most children judged bullying to be right or at least permitted regardless of their role in the episode. However, children with different types of involvement in bullying showed different justification strategies. In particular, bullies were more likely to blame the victims than were children assuming other roles. Gender differences were found in the justifications provided. Boys were more likely to blame the victims, whereas girls were more likely to engage in denial of injury. These results support the notion that justifications for bullying function to sustain bullying.

Reusable software components based on an agent model

Software reuse is a key technology to improve the efficiency of design and maintenance of large software systems. One of the problems in reuse technology is in retrieving components which a programmer wants to modify based on the given specification. We have been studied a framework called ADIPS framework which provides agent based distributed application systems. The framework consists of virtual machines where multi-agent systems work and repositories which manage agent based components. The authors propose a reuse model of software components applying the framework to it and an architecture of the reusable components based on the agent model.

A concept of context-aware computing based on symbol grounding perspective

Conventional context-aware system refers to an intelligent mobile system that can sense the Real Space (RS). Its essential functions are to understand the situation in RS as well as to supply suitable information or services to users depending on the perceived situation. In order to realize that, pattern recognition technologies based on signal or image processing (non-symbolic processing) as well as natural language processing technologies based on knowledge processing (symbolic processing) should be implemented. Historically, symbolic processing and non-symbolic processing have independently evolved but about 20 years ago Steven Harnad had proposed ‘Symbol Grounding Problem’ and argued that it is necessary to integrate non-symbolic processing and symbolic processing. However, there are very few discussions dealing in both context-aware computing and symbol grounding mechanism. Therefore, this paper defines the desired functions for context-aware system in the view point of symbol grounding mechanism and proposes a novel context-aware system, symbiotic grounding system, as an advanced symbol grounding system.

B-dash: agent platform for mutual cognition between human and agents

In this paper, we describe an agent-based platform called “B-DASH” for the Symbiotic Computing. Mutual cognition between human and agents is a key technology of the Symbiotic Computing. In order to promote that mutual cognition, B-DASH has a blackboard mechanism to integrate and share information.

Behavior of agents based on mental states

In multiagent systems, agents communicate with each other to satisfy an entire goal of the systems. To achieve this goal, an agent has to determine its behavior based on its situation in the operational environment. In previous works, an agent using information of external environment has been proposed. However, it is required to utilize various information with respect to each agent to realize a flexible communication among various agents. In this paper, we propose a basic framework of agent communication based on the internal information of an agent, called mental states, and also demonstrate an experimental result of the behavior which evaluates agents with mental states.

Low-level cognitive process model for symbol grounding in context-aware system

This paper proposes a low-level cognitive process model as a part of context-aware systems that can recognize situations in real space as well as provide suitable information or services to users based on the recognized situations. Generally context-aware systems need to deal with sensory data (non-symbolic processing) as well as language resources such as Web (symbolic processing). In this paper, we employ symbol grounding perspective and propose a neural network-based models for converting sensory data into symbolic representation. In addition, this paper exhibits an evidence for a basic question “why do we take a symbol grounding perspective?” in the view point of accessibility of symbolic representation inside systems.

Design of a Symbiotic Sensor Network

In this paper, Design of a concept of a Symbiotic Sensor Network (SSN) and its implementation methodology are proposed. SSN can support people who install sensor network into ordinary home with 1)autonomous action and 2)interactive function.

Proposal of agent organization method on heterogeneous agent platform for symbiotic computing

To achieve cooperative problem solving for use in providing human activity surrogates and supporters, various agents have been designed and developed using many agent platforms. Generally, realizing cooperation or construction remains as a difficult problem confronting agent organizations using different agents running on a heterogeneous agent platform, in contrast to similar agents running on a homogeneous platform. To overcome that problem, a systematic method to construct an organization with different agents on a heterogeneous agent platform is proposed.

Conversational labeling method for sensor nodes in a daily living environment

In this paper, we will propose an easy installation method for a sensor network using a smartphone. The system can be installed and initialized through dialogue between our system and an end user. The method provides easy way to install sensor network in a daily living environment where an end user lives.

ADIPS framework and its application

The aim of this research is to develop a flexible reasoning support environment (cyberspace) that can perceive, reason and act to achieve multiple goals under dynamic, uncertain and complex conditions found in information systems. Current information system architectures are too rigid to cope with heterogeneity, inaccuracy and inflexibility of the real world requirements. In order to overcome such limitations we propose a distributed multiagent software architecture for flexible intelligent information systems with the following goals: (a) Construct a framework for the development of agent-oriented systems in distributed environment called ADIPS, as an infrastructure and two classes of systems are built on it, and (b) develop a reasoning support environment called cyber-office.