In blockchain networks adopting the proof-of-work schemes, the monetary incentive is introduced by the Nakamoto consensus protocol to guide the behaviors of the full nodes (i.e., block miners) in the process of maintaining the consensus about the blockchain state. The block miners have to devote their computation power measured in hash rate in a crypto-puzzle solving competition to win the reward of publishing (a.k.a., mining) new blocks. Due to the exponentially increasing difficulty of the crypto-puzzle, individual block miners tends to join mining pools, i.e., the coalitions of miners, in order to reduce the income variance and earn stable profits. In this paper, we study the dynamics of mining pool selection in a blockchain network, where mining pools may choose arbitrary block mining strategies. We identify the hash rate and the block propagation delay as two major factors determining the outcomes of mining competition, and then model the strategy evolution of the individual miners as an evolutionary game. We provide the theoretical analysis of the evolutionary stability for the pool selection dynamics in a case study of two mining pools. The numerical simulations provide the evidence to support our theoretical discoveries as well as demonstrating the stability in the evolution of miners' strategies in a general case.

Aspect-based sentiment analysis (ABSA) aims to identify views and sentiment polarities towards a given aspect in reviews. Compared with general sentiment analysis, ABSA can provide more detailed and complete information. Recently, ABSA has become an important task for natural language understanding and has attracted considerable attention from both academic and industry fields. The sentiment polarity of a sentence is not only decided by its content but also has a relatively significant correlation with the targeted aspect. For this reason, we propose a model for aspect-based sentiment analysis which is a combination of Convolutional Neural Network (CNN) and Gated Recurrent Unit (GRU), utilizing the local features generated by CNN and the long-term dependency learned by GRU. Extensive experiments have been conducted on datasets of hotels and cars, and results show that the proposed model achieves excellent performance in terms of aspect extraction and sentiment classification. Experiments also demonstrate the great domain expansion capability of the model. INDEX TERMSAspect-based sentiment analysis, online reviews, neural networks.

No abstract

No abstract

No abstract