This paper proposed a bi-directional attention based text-keyword matching model for law recommendation (BiAKLaw). In this model, BERT is utilized as a basic matching model, bi-directional attention mechanism is implemented to extract token-level alignment features and keyword-level differential features, and these features are fused with keyword attentive semantic representations for a better matching effect. The experimental results on the traffic accident and intentional injury datasets demonstrate that, compared with BERT, the proposed model increases F1 evaluation metric by 3.74% and 3.43% respectively.

This research used three technical means of hollow bricks, customized bamboo basket I and customized bamboo basket II to carry out the mangrove restoration test under difficult site condition in Shenzhen Bay in 2016. The effect of plantation was evaluated in terms of colonization technology, economic cost, mangrove population characteristics and sediment environmental quality after 4 years. The results showed that 1) three technical methods in the mangrove plantation area could create an Aegiceras corniculatum restoration. The preservation rate of A. corniculatum gained over 87%. 2) The cost of the hollow brick technology was lower than those of the groups of customized bamboo basket I and customized bamboo basket II. 3) The total organic (TOC) and total phosphorus (TP) of the sediments in the mangrove planting area were significantly higher than those in the intertidal zone, which indicated that mangroves could promote the retention of organic matter and phosphorus and reduce the risk of their escape to the offshore marine environment. The conclusion is that the technical means of hollow bricks, customized bamboo basket I and customized bamboo basket II can successfully restore mangroves under the difficult site condition of Shenzhen Bay. Considering the restoration effect and economic cost, it is recommended to use hollow brick technology to restore mangrove under difficult site conditions.

In order to explore the impact of large-scale circulations on O₃ pollution in Chengdu, based on the 700 hPa geopotential height field in the Sichuan Basin (SCB) and using the T-PCA method, the atmospheric circulations in the summer of 2015–2019 (June–August) were classified into five circulation patterns, and the characteristics of O₃ pollution and O₃ sources in Chengdu under different circulation patterns were identified. Among the five types of circulation patterns, the northeast-high pressure (NEH) pattern and the under-high pressure field (UNF) pattern generally led to higher air temperature, and lower humidity and cloud cover in the SCB. Air temperature change compared to the previous days showed an obvious increasing trend in the SCB under the UNF pattern, kept stable under the NEH pattern, and indicated cooling effects with varying intensities under the other three circulation patterns (the southeast-high pressure pattern, the northwest-low pressure pattern and the northwest-high pressure pattern). The meteorological conditions under NEH and UNF were more conducive to the photochemical production of O₃ in the SCB. During the study period, O₃ concentrations and the percentages of over-standard days of O₃ in Chengdu under NEH and UNF were significantly higher than those under the other three circulation patterns, and thus NEH and UNF were the main circulation patterns that were related to O₃ pollution in summer. Months with more severe O₃ pollution in Chengdu tended to have higher proportion of days under the NEH and UNF patterns. The simulation studies of O₃ sources in Chengdu in summer suggested that O₃ was mainly derived from the SCB (55.0%), with local contribution of Chengdu (31.6%), and other important contributing cities including Deyang (5.4%), Chongqing (4.0%), Ziyang (3.9%) and Meishan (2.2%). However, O₃ sources in Chengdu under different circulation patterns showed different characteristics. Under the NEH pattern, O₃ transport within the Chengdu Plain had notable impact on O₃ pollution in Chengdu, with a significant increasing contribution from Deyang (14.9%). By contrast, under the UNF pattern, O₃ pollution in Chengdu became more localized, with nearly half (46.8%) of O₃ contributed by local emissions. 

According to the ground and vertical meteorological characteristics of atmospheric circulation types during the autumns and winters (Nov.?Feb.) of 2013?2018 in Shijiazhuang, 11 types of atmospheric circulation were classified into five categories. The correlation between each category and PM_2.5 pollution was discussed by combining the transport trajectories and the air quality data. Among those categories, Category I (type NW and type N, accounting for 16% of days) was the best condition for air pollution diffusion, characterized by the dominant west or northwest winds near surface with steady direction and high speed, as well as high planetary boundary layer (PBL) height. Category II (type NE, accounting for 9%) and Category III (type E and type SE, accounting for 12%) were the second-best conditions, characterized by high wind speed. Category II was dominated by north wind and medium PBL height, while Category III was dominated by northeast wind and low PBL height. Category IV (type A, 37%) was unfavorable for air pollution diffusion, which was characterized by low-speed, near-surface winds and low PBL height. Category V (type UM, type C, type S, type SW, type W, 26%) was the worst condition, characterized by near-surface winds with low speed and variable directions, as well as low PBL height and intensive temperature inversion in lower atmosphere. The potential areas of sources contributing to PM_2.5 pollution in Shijiazhuang under different circulation types varied with different transport trajectories. Concerning the correlation between the PM_2.5 pollution in Shijiazhuang during the autumns and winters and the diffusive conditions of different circulative types, the PM_2.5 pollution tended to occur (the frequency of pollution was from 78% to 96%, and the frequency of heavy pollution and above was more than 55%) when the region was controlled by Category V; the condition of the PM_2.5 pollution tended to change slowly under Category IV, but the PM_2.5 concentration was more likely to increase under the continuous Category IV days; the frequency of PM_2.5 pollution was lowest under Category I.