Application Of Ant Colony Foraging Model Used In Co-occurrence Network Clustering

There are so many networks exist in our real life, such as Personal networks、Internet network and so on. In all kinds of network structure, the number of nodes with similar characteristics will be clustered into one group and once the number of nodes in a network is large, this net can be called complex network. Because of it’s complexity in internal structure, it is more difficult to cluster the complex network. But with the gross of mathematics and computer science, more and more new and innovative clustering algorithms and tools for clustering appeared in the different disciplines and field of research and provide a theoretical basis for them. Those algorithms and tools have been rapidly developed and applied.The heuristic algorithm bionic creatures-Ant colony foraging model is used into clustering analysis of the elements of literature, not only because of it is an innovative and initiative method, but also a new way of thinking for scientometrics and develop a new vision For scientometrics’ subsequent development.The Ant colony will release "pheromone" on the traveled path for foraging activity, which help to guide the following ants find food more quickly, while other ants walked through the path wil] also release "pheromone", it can be called "pheromone" positive feedback phenomenon and "pheromone" on the path will fade away gradually with the time pass away. Sometimes individual ants random choose other paths and those paths will be another way for foraging. Because "pheromone" on the shortest foraging path volatile less and this way will gradually become the main road, at last the shortest route will be repeated by most ants. The similar phenomenon in human society we called information foraging also have the same manner and process, that is the process when scholars from different fields of scientific research seeking information. Scholars retrieve information based on previous results found by other scholars according to surviving pheromones, such as keywords, references, etc., which provide important scientific evidence and information foraging path for later scholars. When the scholars output their own research findings it can also produce "pheromone", which can help other scholars. At the same time, they can put forward innovative research questions and develop new fields. Throughout the process, information foraging and ant colony foraging are similar both in the mechanism and in the process, so we can use foraging ant colony algorithm for clustering analysis in the literature analysis filed.By constructing ant colony foraging model and applying the model in four typical co-occurrence network cluster analysis in the field of scientometrics, we found that ant colony foraging clustering model worked well in the Citation and journals co-occurrence networks, and also have certain clustering effect on Author Co-citation network, but for literature keyword co-occurrence network, the clustering result is relatively weak, which is due to the model focus on the movement of pheromone.The ant colony foraging pheromone in citations, author and journals co-citation network are obvious and can make a strong guiding role. But to keyword co-occurrence networks, although it can describe the whole documents, but the guiding role of pheromones is relatively weak, thus affecting its clustering effect. It can be concluded that ant foraging model program is more convenient to use in the specific operation because of its design and the positive feedback ant colony clustering algorithm principle, also it can be able to quickly reach the operations optimal solution and has strong robustness. At the same time when using foraging model it’s also very easy to find knowledge flows and cross-impact between these clusters and we can find the flow of knowledge and support between different research areas the situation through contact among clusters.