Online Content based Approach of Network Extraction for Finding Weight, Size and Features

Abstract

In an online community, moderating the content created by users is a great challenge. Nevertheless, this paper is very costly to man-made workers. To end this, an automatic machine can arouse great interest to identify breaches of user content. Some ways to solve this problem, but word filtering or regular expression matches are most often seen in practice. The key drawbacks are their insecurity and their context-insensitive existence deliberately concealed by the users. In addition, they are reliant on languages and may need the appropriate training service. In this paper, we suggest an automated harassment detection system that totally ignores the content of the post. We derive and describe a network of conversations from raw chat logs by means of topological steps. We then use this to train our harassment detection task as a classifier. On a data, thoroughly test our program, with a massive French multiplayer online gaming user feedback. Set the most suitable parameters for network extraction and speak about the discriminatory potential of our applications, both topologically and temporally. Our system hits 83.89 F-measurements, which improves existing methods with the full feature set.

Keywords: Allocation of weight, classification algorithms, detection of misuse, network extraction, word filtering

Cite this Article: Nalli Vinaya Kumari, T Aravind, M Vineeth, M Akshita. Online Content based Approach of Network Extraction for Finding Weight, Size and Features. Journal of Mobile Computing, Communications & Mobile Networks. 2020; 7(1): 23–30p.