No abstract

The goal of a technology-assisted review is to achieve high recall with low human effort. Continuous active learning algorithms have demonstrated good performance in locating the majority of relevant documents in a collection, however their performance is reaching a plateau when 80%-90% of them has been found. Finding the last few relevant documents typically requires exhaustively reviewing the collection. In this paper, we propose a novel method to identify these last few, but significant, documents efficiently. Our method makes the hypothesis that entities carry vital information in documents, and that reviewers can answer questions about the presence or absence of an entity in the missing relevance documents. Based on this we devise a sequential Bayesian search method that selects the optimal sequence of questions to ask. The experimental results show that our proposed method can greatly improve performance requiring less reviewing effort.

Coal mine safety management involves many interested parties and there are complex relationships between them. According to game theory, a multiparty evolutionary game model is established to analyze the selection of strategies. Then, a simplified three-party model is taken as an example to carry out detailed analysis and solution. Based on stability theory of dynamics system and phase diagram analysis, this article studies replicator dynamics of the evolutionary model to make an optimization analysis of the behaviors of those interested parties and the adjustment mechanism of safety management policies and decisions. The results show how the charge of supervision of government department and inspection of coal mine enterprise impact the efficiency of safety management and the effect of constraint measures and incentive and other measures in safety management.

In this paper, we try to find the right control method for the game behavior instability in coal mine safety management. Through the analysis and comparison of the system stability with inflexible and flexible costs and penalties, it can be concluded that the dynamical game system with flexible costs (incentive rewards) and flexible penalty mechanism can significantly reduce the dynamics of unsafe behaviors in coal mine safety supervision. A combined mechanism of incentive rewards and flexible penalty is put forward to improve the stability of the dynamical system and control the instability of behaviors effectively. The results of model simulation show that the combined mechanism has very good property and can optimize and control the instability of behaviors and strategies of the interested parties. Based on the theoretical conclusions, some control strategies and policy advice are proposed for the improvement of the system and measures of safety management for government departments.