Diffusion of innovation is usually difficult, because many members in society had not adopted innovation initially and people are usually rational: they select risk dominant solution and keep initial state. Cascade phenomena, which are sequences of adoption by agents, are the important driven forces for the society to make a successful diffusion of innovation, emergence of social norm and opinion formation. There are certain kinds of networks (ex. social network) under these cascade phenomena and the topologies will affect the dynamics. The cascade phenomena can occur in specified range of conditions (or known as cascade windows) defined by both the average degree of the underlying network topologies and the threshold of each agents (or nodes). This paper shows what kind of network topology maximizes cascade of innovation in terms of cascade window using evolutionary optimization method and how the network topology drives cascade phenomena at wider conditions than other networks. From results of the optimization, the best networks have both cluster of hub nodes and cluster of nodes with a few links, which are called vulnerable nodes. By a detailed consideration of topologies of the best networks, the authors also propose a network model (P model) to maximize the cascade of innovation, in which a mechanism of probabilistic growing network is used. At each discrete time step, one new node and constant number of new links are introduced and added to hub nodes or vulnerable nodes in the network depending on a certain probability. The P model has better scalability compared with evolutionary optimizing method in terms of time complexity and can make networks that drive cascade phenomena at wider conditions than evolutionary optimized networks. The success of P model strengthens the importance of both cluster of hub nodes and cluster of vulnerable nodes for maximizing cascade of innovation on networks.
This paper discusses how the neighbours affect the decision of consumer behaviour over diffusion of innovation. An agent-based model of diffusion is proposed on an online social network which have both “scale-free” and “regular” properties. The findings of the studies of consumer activity in order to show the following points: 1) the informative effect can cause a take-off, but it is not sufficient to reach the completion of diffusion, 2) the combination of the informative and normative effects can easily bring a take-off, which is a point in time within the adoption curve that the existence of a sufficient amount of adopters of an innovation or a product. After the take-off, the diffusion is accelerated and reaches the completion in the end, 3) the informative effect makes information propagate fast, and so does the normative effect over a network that has characteristics of scale-free and high cluster, 4) in a selective advertisement, the most effective approach is non-selective advertisement for all consumers. This paper shows that it is inadequate to think that opinion leaders only adopt a product and transmit the information of usability impressions to other consumers in order to trigger diffusion on online human-relationship networks. Rather, diffusion is promoted entirely by active communication among non-opinion leaders which have received such information from opinion leaders.
Social media, which support communication on the Internet, continue to spread rapidly. SNSs and micro-blogs are typical services. This research proposes methods for efficient information sharing through social media, especially intra-firm systems for information sharing of research or expertise. In this research, a simple model of social media is provided, whose users share information by directly referring to other users. The authors focused on a particular intra-firm social media and analyzed the usage of its communities. By applying a clustering method, they are classified into six groups and analyzed their characteristics and occupation rate. Finally, the authors proposed a method that supports information sharing and evaluated it by agent-based simulation using our proposed model. The authors clarified that not only encouraging users to post articles but also giving them an opportunity to refer to the existing community' s history is effective for information sharing.
This paper focuses on hierarchical-game based multi-attribute negotiation between multiple MAs (Manufacture Agents) and multiple MSAs (Material Supplier Agents) of SCN (Supply Chain Network). The attributes of the wholesale price of the product, the quantity of the order and the lead time of the order are negotiated simultaneously. A modified hierarchical-game based negotiation protocol is proposed based on the previous work. It is a changeable hierarchical game. It is a two-layer game when all the orders are in abilities of all the MSAs. However, it is a three-layer game when there are some orders out of abilities of some MSAs. The second layer game is not necessary, it is triggered to find coalitions only if the order is out of ability of MSA. The third layer game is used to determine the final quotation of the negotiation. Then, the first layer game is aimed to find the optimal allocation scheme to maximize the total profit of SCN based on the results of the second and the third layer games. Numerical example is provided to illustrate the proposed protocol.
In this paper, the authors examine strategic communication including ambiguous utterances. Recently, game theoretical approaches are introduced in the study of pragmatics. Parikh's paper is considered as one of the representative studies. He shows that there exist equilibria including an ambiguous utterance as a part of the equilibrium strategy in his model. However, his model has some limitations. First of all, in his model, interests between the sender and the receiver perfectly coincide, and secondly he analyzes with a static solution concept. In this paper, Parikh's model of ambiguous utterances is extended by changing the degree of coincidence of interest between the sender and the receiver, and is analyzed by evolutionary dynamics. The authors found that ambiguous utterances can be a stable outcome with certain conditions. Agent-based simulation also confirmed this finding.
This paper presents an evolutionary model of a population where individuals play a repeated Leader Game where two agents each has two options to choose, D (active/aggressive action: Daring, Dive) and C (passive action: Concede, Careful). The simulation results show that the evolutionary process for the Leader Game brings about a linear dominance hierarchy among strategy (automaton) classes. The author also investigates the robustness of the result against the perturbation of the game structure, the change of memory size of automata and the introduction of behavioral error.
Recently, a need has arisen for facsimile model construction with close correspondence to the real world for use in evaluating the effectiveness of specific policies. When constructing facsimile models requiring a large amount of data, however, parameter estimation using data fitting becomes problematic. Furthermore, there are often problems where, owing to the limitations of field research, it becomes impossible for analysts to obtain the data necessary for model construction. The paper proposes a solution to this data collection problem by using “virtual grounding” as a method for creating valid agent models. The proposed method constructs an agent model by isolating qualitative features of the real world situation that are targets for modeling at a stage where the complete dataset is not yet available, and uses standard models for which utility has been previously demonstrated. Following this construction, a number of sample participants modeled as agents repeatedly make hypothetical decision-making actions within the model environment, and the model parameters are estimated based on the results of these decisions. This paper demonstrates the utility of the virtual grounding method by using an example of modeling visitor agents to Tokyo Disney Sea.
This paper focuses on a collective adaptive situation as an complex phenomena in social dynamics and aims at specifying its situation towards a validation of agent-based social simulations. For this purpose, the fuzzy c-means clustering (FCM) method is applied into the experimental data of both the human subject experiments and the agent-based social simulations in the cross-cultural game, Barnga, and compare these results from the viewpoint of a collective adaptive situation. The analysis of these data has revealed the following implications: (1) the collective adaptive situation can be specified by the FCM method; and (2) the classified results of both of the experimental data are close to each other, indicating that agent-based social simulations can be validated from the viewpoint of a collective adaptive situation.
Although a lot of researchers have interests to social simulations recently, many of their outcomes can show just abstract lessons over simple settings in their social simulations. It is understandable because complicate settings will make their social simulation tool difficult to understand and tune its parameters. There is a dilemma between simple settings with abstract lessons and difficult settings with difficult tool tuning. In this paper, it is shown that how social simulation tools can be used to make a policy proposal for polling place assignment. In the previous study of authors, a model was proposed for a decision making for each voter in an election. The change of the polling places was considered in a city, Takatsuki, Osaka Prefecture, Japan, to increase the voter turnout and to reduce the number of polling places. A method was also proposed to identify regional polling parameters in an election using real voter turnout records. With those regional polling parameters, selection and assignment of polling places to each region was done using an Evolutionary Multiobjective Optimization (EMO) algorith. Several sets of solutions were found that increase the voter turnouts and reduce the number of the polling places. In this paper, the authors show several practical results after consulting the office of the board of elections in the target city. In order to utilize outcomes of social simulations practically, they are advised that they should show the difference between the current voting assignment and the proposed one in order to announce those who should vote in a different station. They are also advised to show how to support those who should go to farther polling stations after the change of polling stations. To meet these requests from the office, they simulate a case with free shuttle bus service using our model.
The Great East Japan Earthquake on March 11th, 2011 struck eastern coastal area of the Japanese main island and tsunami following the earthquake posed a devastated condition. Most infrastructures and lifelines, such as water supply systems, petroleum gas supply systems and electric power systems including Fukushima nuclear plant were damaged. Logistics systems are no exception. The authors developed “Support System for Transportation under Disaster Circumstance” comprised of three tools. This paper focusses on one of the tools, transportation simulator for relief goods and demonstrated functions of the simulator and simulation examples.
This paper describes implementation of a pedestrian simulation environment to carry out an exhaustive analysis of huge-scale pedestrian flow. Our environment consists of the CrowdWalk pedestrian simulator and the PRACTIS simulation controller. CrowdWalk has three pedestrian movement models that are based on a one-dimensional space model. PRACTIS carries out an exhaustive and efficient analysis in a cluster environment to support effective decision-making, enabling pedestrians to be guided according to various criteria. The authors apply CrowdWalk and PRACTIS in a firework festival to verify their guidance planning, and examine some case studies.
This paper presents an agent-based simulation model to analyze performance of organization with heterogeneous members. A hierarchical landscapes model with organizational and personal landscapes is proposed and it puts difference of skills and values into difference of personal landscapes. The use of this model shows that an organization needs to have a certain amount of diverse members to improve the whole organizational utility under the changing environment. This is because while the uniform members stay at a state with higher individual utility even if there are diverse members in the organization, the diverse members discover a new state with higher organizational utility and then take others to that state.
As robot technology improves, autonomous robots expect to work in place of humans. It is cost efficient when such autonomous robots are supposed to perform multiple tasks. This paper deals with such multi-task robots and proposes a task selection technique for multi-task autonomous robots. When an autonomous robot is implemented for multiple purposes, it is important that the robot be able to appropriately select its own tasks. The proposed method, named collaborative task casting, takes degrees of task achievement into account and balances task occupation to ensure that adequate robot resources are directed towards each task. The experimental simulation results showed that the proposed method had advantage over existing methods in terms of minimizing unsatisfactory state and adaptively worked in many possible settings.