Logical Analysis And Algorithm Research On Search Strategy In Complex Network

The research of this paper has apparent interdisciplinary features, From the theoretical perspective, cross-over study on logics, artificial intelligence and complex network will formally describe the search process of the complex network as well as behaviors and status of participated searchers in combination with theories such as modal logic, BDI logic, LORA, etc. It has great significances to establish an intelligent theory system which can treat the search process of a complex network. From the application perspective, the search process which researches the complex network based on the logic theory will result in programs and systems of significant application values. It has important research prospects in solving issues on intelligent information search in social network, computer network, WWW and transportation network of the true life.This paper focused on the research of several common and typical algorithms in complex network search and put forward a new BDM search algorithm and a social network search algorithm based on special themes. Besides, through logical analysis of the complex network searching process and cut-over from agent to the complex network, it established a new BDI agent-based complex agent network(CAN) model and a BDI logic-based CAN search model. Considering the above search strategies, the paper explained the reasoning process and description of the BDI logic-based CAN search model, analyzed logical bases for rational selection of the search agent and service agent in the complex agent network and proposed a scheme for a new intelligent search algorithm BMA for the LORA-based complex agent network. Main contents of the paper are presented as follows:(1) To improve the search efficiency and reduce the search cost, it proposed a new BDM search algorithm and a social network search algorithm based on Specific Topics. In addition, it also carried out simulation experiments, performance analyses and logical analyses on the results of search strategies.(2) Based on first-order predicate logic and the world semantics, this paper showed the relation between the search process and logical reasoning process, depicted search status sets of the complex network and discussed the relation with predicate logic status sets of the first-order modal. All possible network search status sets in search strategies are depicted by the first-order predicate logic. Features of search status are also analyzed.(3) It mainly cited the natural calculation method and the BDI Agent, proposed a new complex agent network(CAN) model and presented realization thoughts of the CAN model and a formalized definition of the model. The method describes status and behaviors relevant to the agent node through individualagent models as well as the complicity and interaction between the complex network and agents to better describe the complicity of problems relevant to the complex network dynamic behavior between individuals and groups(e.g. network search, transmission, phase change, transfusion, etc.).(4)Based on BDI(Belief,Desire and Intention) models of individual agent and BDO(Belief, Desire and Obligation) models of group agent, it constructed a BDI logic-based complex agent network search process model, described the model of the dynamic search scale and analyzed state factors affecting the search process from two levels of the complex agent network search process: in the individual level, it mainly researched thinking property memories such as individual agent modeling, belief, wish, intension etc. and their treatment, how to transfer the BDI agent model of the evolutionary process of social behaviors and introduced a BDI agent based reasoning process; in the group level, it mainly researched the BDI agent model based on the macroscopic description of the complex coordination regulations among Agents. It mainly considered mass phenomena such as unity, collective commitment, and regulation, etc.(5) For the LORA(Logic of Rational Agents) logic expanded on the basis of the dynamic logic and the BDI logic, it depicted the search behaviors of the complex Agent network from the dynamic performance that the diversity of the network search status and changes with the time for the coordination process with agent based LORA. It also proposed a search model mechanism and tectonic model to solve the dynamic coordinated search of several Agents of the complex agent network. To solve the intelligent search problem of the complex network, on the basis of the CAN network model, it proposed a BMA search strategy in the CAN. The method transmits query requests by intelligence and random movements through allocating certain amount of agents in the CAN network to solve the search problem in the complex agent network. We call the search algorithm of completing the search task through search agents in the CAN as the BMA search strategy based on multi agents that the Mobile. It also proposed and solved key technologies on the BMA search mechanism, BMA search algorithm description, BDIbased agent search model etc.Finally, it had a general introduction to experiment software for algorithms in this paper, specified features and experiment environments of simulation software and finally proposed parameter setting situations relevant to the experiment. It tested the search algorithm performance indexes SSR, LC and RPL of the two classic algorithms and the BMA algorithm proposed in this paper, analyzed SSR, LC and RPL of MBA, etc. The basic performance, expansibility and adaptation of the BMA algorithm proposed in the paper were also assessed.