Moderation Techniques for Social Media Content

Abstract

Social media are perhaps the most popular services of cyberspace today. The main characteristic of social media is that they offer to every internet user the ability to add content and thus contribute to participatory journalism. The problem in that this content must be checked as far as quality is concerned and in order to avoid legal issues. This can be accomplished with the help of moderation. The problem is that moderation is a complex process that in many cases requires substantial human resources. This paper studies the moderation process and proposes a moderation model that can guarantee the quality of the content while retaining cost at an affordable level. The model includes various moderation stages which determine the applied moderation technique depending on the publication record of the user that submits the content.

The Ethics of AI Ethics: An Evaluation of Guidelines

“…However, it is omitted that producer and consumer practices in the context of AI technologies may in themselves contradict sustainability goals. Issues such as lithium mining, e-waste, the one-way use of rare earth minerals, energy consumption, low-wage "clickworkers" creating labels for data sets or doing content moderation are of relevance here (Crawford and Joler 2018;Irani 2016;Veglis 2014;Fang 2019;Casilli 2017). Although "clickwork" is a necessary prerequisite for the application of methods of supervised machine learning, it is associated with numerous social problems (Silberman et al 2018), such as low wages, work conditions and psychological work consequences, which tend to be ignored by the AI community.…”

Design and Application of an AI-Based Text Content Moderation System

“…e system first checks the format of the input text, and if error has been detected, the system will not process the text anymore, but will prompt that the input format is incorrect, and the user needs to re-enter the text; after format check is passed, the text content will be pushed to the AI-based TCM system, which will call the API to moderate the text content. If the content check is passed, an approval message will be sent to the system administrator, who can conduct random content check and show the moderated text comments on the webend and app-end [14]. If the text content contains illegal content such as prohibited words, spams, or advertisement and fails to pass the moderation, it will be classified as unapproved type, and then, the AI-based TCM system can clarify the violation type of the text content, prompt alert to the user that the text he/she entered contains illegal content, and point out the violation type, and the relevant information will be recorded in the system log; for the type of the violation that cannot be determined by the system, the text will be pushed to the system administrator for manual moderation.…”

The Risks, Benefits, and Consequences of Prepublication Moderation: Evidence from 17 Wikipedia Language Editions

“…In many situations, this time spent on the work of reviewing might be wasteful and inefficient if it means that moderators have less time to spend contributing to the community in other ways. Furthermore, waiting for an extended amount of time for contributions to be explicitly approved creates "a lack of instant gratification on the part of the participant" which may reduce individuals' desire to contribute again [77].…”