Semantic similarity based clustering and modeling using Latent Dirichlet Allocation (LDA)

Abstract

Privacy has become a substantial issue once the applications of big data are dramatically growing in cloud computing. In recent years, we have a tendency to focus on privacy and propose a unique a novel approach that is termed Dynamic data encryption Strategy (D2ES). Our planned approach aims to by selection encipher knowledge and use privacy classification ways under temporal order constraints. This approach is intended to maximize the privacy protection scope by employing a selective coding strategy within the specified execution time necessities. During this paper, is intended victimization semantic similarity based mostly clustering and topic modeling victimization Latent Dirichlet Allocation (LDA) for summarizing the big text collection over Map reduce framework. The account task is performed in four stages and provides a standard implementation of multiple documents account. The conferred technique is evaluated in terms of quantifiability and varied text account parameters particularly, compression ratio, retention ratio, ROUGE and Pyramid score are measured. The benefits of Map scale back framework are clearly visible from the experiments and it's additionally incontestable that Map reduce provides a quicker implementation of summarizing giant text collections and may be a powerful tool in big Text data analysis.