With the rapid growth of social tagging systems, many efforts have been put on tag-aware personalized recommendation. However, due to uncontrolled vocabularies, social tags are usually redundant, sparse, and ambiguous. In this paper, we propose a deep neural network approach to solve this problem by mapping both the tag-based user and item profiles to an abstract deep feature space, where the deepsemantic similarities between users and their target items (resp., irrelevant items) are maximized (resp., minimized). Due to huge numbers of online items, the training of this model is usually computationally expensive in the real-world context. Therefore, we introduce negative sampling, which significantly increases the model's training efficiency (109.6 times quicker) and ensures the scalability in practice. Experimental results show that our model can significantly outperform the state-of-the-art baselines in tag-aware personalized recommendation: e.g., its mean reciprocal rank is between 5.7 and 16.5 times better than the baselines.

No abstract

No abstract

Abstract. Tropospheric ozone (O3) has replaced PM2.5 or PM10 as the primary pollutant in the North China Plain (NCP) during summer in recent years. A comprehensive understanding of O3 production in response to the reduction of precursor emissions over the NCP is urgently demanded for effective control policy design. In this study, the air quality modeling system RAMS-CMAQ (Regional Atmospheric Modeling System–Community Multiscale Air Quality), coupled with the ISAM (Integrated Source Apportionment Method) module is applied to investigate the O3 regional transport and source contribution features during a heavy O3 pollution episode in June 2015 over the NCP. The results show that emissions sources in Shandong and Hebei were the major contributors to O3 production in the NCP. Not only the highest local contribution of O3 mass burden but also more than 30 % contribution of O3 mass burdens in Beijing and Tianjin came from the emissions sources in these two provinces, respectively. Conversely, the urban areas and most O3-polluted regions of the NCP were mainly dominated by conditions sensitive to volatile organic compounds, while “both control” and NOx-sensitive conditions dominated the suburban and remote areas, respectively. Then, based on the sensitivity tests, the effects of several hypothetical scenarios of emissions control on reducing the O3 pollution were compared and discussed. The results indicated that the emissions control of industry and residential sectors was the most efficient method if the emissions reduction percentage was higher than 40 %. However, when the emissions reduction percentage dropped below 30 %, the power plant sector could make significant contributions to the decrease in O3. The control strategies should be promptly adjusted based on the emissions reduction, and the modeling system can provide valuable information for precisely choosing the emissions sector combination to achieve better efficiency.

No abstract