To detect malicious URLs more timely, machine learning based malicious URL detection methods have replaced traditional blacklist methods. These studies aim to improve the accuracy and speed of detection from various aspects such as URL segmentation, URL embedding methods, machine learning models, etc. However, the security issues inherent in these machine learning based malicious URL detection methods have been overlooked. Adversarial example attacks are one of the security issues faced by machine learning based malicious URL detectors. In this paper, we proposed a new adversarial example attack method against malicious URL detection based on machine learning and it has better performance than existing methods. Besides, we compared the robustness of different URL embedding methods and machine learning models with our attack methods and existing attack methods. At last, we analyzed the reasons why our proposed method performs better and the reasons why the context-considered embedding method has high resistance to adversarial example attacks.

In order to solve the shortcomings of using blacklist method to detect malicious URLs, such as slow update speed, the research of using machine learning to detect malignant URLs are increasing. These researches have proposed their own methods and obtained great accuracy, but the summary research on malicious URLs detection is insufficient. In this paper, we propose a three-step framework for malicious URLs detection and we overview 14 related works by our three-step framework and find that almost all researches of malicious URLs detection using machine learning can be classified by three-step framework. We evaluate some machine learning models and context-considering methods and their suitability by our three-step framework. According to the results, we verify the importance of considering context and find that context-considering embedding methods are more important and the malicious URLs detection accuracy improved with context-considering methods.

Nowadays, the ubiquity of sharing economy and the booming of ride-sharing services prompt Mobility-on-Demand (MoD) platforms to explore and develop new business modes. Different from forcing full-time drivers to serve the dispatched orders, these modes usually aim to attract part-time drivers to share their vehicles and employ a 'driver-choose-order' pattern by displaying a sequence of orders to drivers as a candidate set. A key issue here is to determine which orders should be displayed to each driver. In this work, we propose a novel framework to tackle this issue, known as the Information Disclosure problem in MoD systems. The problem is solved in two steps combining estimation with optimization: 1) in the estimation step, we investigate the drivers' choice behavior and estimate the probability of choosing an order or ignoring the displayed candidate set. 2) in the optimization step, we transform the problem into determining the optimal edge configuration in a bipartite graph, then we develop a Minimal-Loss Edge Cutting (MLEC) algorithm to solve it. Through extensive experiments on both the simulation and the real-world data from Huolala business, the proposed method remarkably improves users experience and platform efficiency. Based on these promising results, the proposed framework has been successfully deployed in the real-world MoD system in Huolala.