Rocchio Nearest Centroid and Normalized Neural Network based Lead Generation in Social Media Marketing

Abstract

The employment of the internet and social media has transposed consumer behavior and the methods in which business organizations carry over their business. Social and digital marketing recommends noteworthy changes to business establishments via cost curtailment, enhanced brand understanding and surged sales. Despite enormous amount of potentialities, noteworthy disputes subsist from gloomy digital oral message in addition to trespassing and troublesome online presence of brand. Deep learning (DL) has fascinated escalated awareness owing to its notable processing power in tasks, to name a few being, speech, image, or text processing. Due to its aggressive evolution and extensive accessibility of digital social media (SM), examining these data utilizing conventional materials and methods is substantial or even complex. Also with the large growth in the volume of data, the multifariousness in data heterogeneity, are the most distinguished reasons, why and how the SM data mounted. In this paper we study the impact of tweets on distance learning to understand people’s opinions and to discover facts. However, adding redundant features minimizes the generalization capability of the model and may also minimize the overall accuracy of a classifier. We introduce Rocchio Nearest Centroid Laplacian Feature Selection model that combines Rocchio Nearest Centroid and Laplace function for selecting relevant features or tweets. Next an Arbitrary Normalized Attention-based Recurrent Neural Network Lead Generation algorithm is designed aggregating characterizations from preceding and succeeding tweets while generating lead via digital marketing tweet funnel. We validate and evaluate our method using data from distance learning dataset. Experiments and comparisons on distance learning data show that, compared to existing SMM methods, considering generalization capability and digital marketing tweet funnel results in improvements in processing time, lead generation accuracy and precision to a significant extent.