Research On Overlapping Community Detection Method Based On Ant Colony Algorithms In Text Networks

The main channel for the dissemination of resources online is the information carrier based on the application of text.Enormous amount of data allows users to select information from massive choices when inquiring,and it is difficult to distinguish between true and false.Thus,how to search the target information efficiently and accurately is an urgent problem to be solved.The burgeoning of complex networks has broadened the vision of the research in social structure,scholars have successively discovered community structures from various network data sets.Therefore,the discovery of the text network community structure has also become an important issue.However,overlapping community detection has a prominent effect in the researching areas of the hidden structures in real networks.This paper discovers a research of the text weighted complex networks based on the ant colony algorithm of the text network overlapping community.Firstly,a weighted network model was built in this paper on account of text similarity.Making use of the comprehensive characteristics of the nodes in the word co-occurrence network to extract keywords that can reflect the topic of the text,and by means of Wikipedia knowledge base,it conducts the word-concept matching.The synthesized measurements of relevance between words are taken by the distance of links and category.Then this paper proposes a text similarity method that considers semantic information and language structure synthetically.The calculation method is determined by the cosine similarity of the text keyword vector and included words,and then it constructs a complex network of text sets with text as nodes and similarity as edge weight.Secondly,a local extended ant colony algorithm to optimize overlapping community detection(LEAC-OCD)is proposed in the paper.The algorithm forms the triangle phantom with the core nodes and the common nodes from neighbors as the initial value of the ant colony algorithm.Constructing a modular weighted community clustering function is an adaptive function of the algorithm to estimate and quantify the stability of the community structure,and it applies the modulo ordered table coding method to achieve a rough division of community structure.Then,through the free movement of ants inspired from transfer mechanism,it alters the attribution of ant's position and utilizes the post-processing strategy to obtain the results of the overlapping community.The convergence speed of this algorithm is faster and the time complexity is lower.Experiment shows that the LEAC-OCD algorithm is superior to other classical algorithms on simulated data sets and artificial data.Finally,the LEAC-OCD algorithm is applied to the discovery of text network community.It initially uses the GN algorithm to divide the effect of different data sets to determine the effective threshold interval for text similarity,builds a text network with different thresholds,and then exert the LEAC-OCD algorithm and other overlapping community detection algorithms to divide the text network.The results indicates that the value of the algorithm modularity function in this paper is relatively high so that it can effectively divide high-quality text communities.