With the development of edge computing, a large number of tasks can be offloaded to the edge server for computing, among which the dispatching and scheduling of dependent tasks has attracted extensive attention.The offloading of dependent tasks mainly has the following problems: how to select an appropriate edge server for dispatching, how to arrange the scheduling order of edge servers to better schedule tasks, and how to solve the task dependency problem.In this paper, we proposal a dispatching and scheduling method DAMD, based on reinforcement learning and multi-agent reinforcement learning, to solve the above three problems.Specifically, as the first step of DAMD, a reinforcement learning approach is designed to estimate the network load and dynamically dispatch tasks to the appropriate edge servers.Each edge server is regarded as an agent by a multi-agent reinforcement learning method, the second step of DAMD, which comprehensively considers the dependency relationship between tasks and the scheduling relationship between servers to achieve the efficiency and fairness of task scheduling.Finally, the results show that our method can better complete the task within the deadline and greatly reduce the average response time according to the time sensitivity requirement.

With the development of big data technology, processed data has become an important source of value. Data has played a pivotal role in the development of enterprises, especially internet enterprises. However, Internet enterprise platform companies generally infringe on personal privacy in various stages of information collection, processing and application, and Internet enterprise platform data protection research is of great significance. The study found that the current problems of data protection on Internet enterprise platforms include: extremely weak user data protection measures, intellectual property risks throughout the whole process of big data processing, and infringements that have both new and high-tech characteristics. The high ambiguity in the definition and attribution of “data rights”, the low cost and high concealment of infringements, and the value difference between intellectual property protection and digital economy are the main causes of these problems. As far as the protection path is concerned, we should start from the three aspects of technology empowerment, governance empowerment and legal empowerment, and work together to promote the proper protection of Internet enterprise platform data.

The internet has reconstructed information boundaries in the modern world, and along with mobile internet has become the most important source of information for the public. Simultaneously, the internet has brought humanity into an era of information overload. In response to this information overload, recommendation systems backed by big data and smart algorithms have become highly popular on information platforms on the internet. There have already been many studies that attempted to improve and upgrade recommendation algorithms from a technical perspective, but the field lacks a comprehensive reflection on news recommendation algorithms. In our study, we summarize the principles and characteristics of current news recommendation algorithms and discuss “unexpected consequences” that might arise from these algorithms. In particular, technical bottlenecks include cold starts and data sparsity, and moral bottlenecks are presented in the form of information imbalance and manipulation. These problems may cause new recommendation systems to become a “warped mirror”.

Improvements in big data and machine learning algorithms have helped AI technologies reach a new breakthrough and have provided a new opportunity for quantitative research in the social sciences. Traditional quantitative models rely heavily on theoretical hypotheses and statistics but fail to acknowledge the problem of overfitting, causing the research results to be less generalizable, and further leading to societal predictions in the social sciences being ignored when they should have been meaningful. Machine learning models that use cross validation and regularization can effectively solve the problem of overfitting, providing support for the societal predictions based on these models. This paper first discusses the sources and internal mechanisms of overfitting, and then introduces machine learning modeling by discussing its high-level ideas, goals, and concrete methods. Finally, we discuss the shortcomings and limiting factors of machine learning models. We believe that using machine learning in social sciences research is an opportunity and not a threat. Researchers should adopt an objective attitude and make sure that they know how to combine traditional methods with new methods in their research based on their needs.