Analysis of the Main Factors Influencing Food Production in China Based on Time Series Trend Chart

Based on the annual sample data on food production in China since the reform and opening up,we select 8 main factors influencing the total food production( growing area,application rate of chemical fertilizer,effective irrigation area,the affected area,total machinery power,food production cost index,food production price index,financial funds for supporting agriculture,farmers and countryside),and put them into categories of material input,resources and environment,and policy factors. Using the factor analysis,we carry out the multi-angle analysis of these typical influencing factors one by one through the time series trend chart. It is found that application rate of chemical fertilizer,the growing area of food crops and drought-affected area become the key factors affecting food production. On this basis,we set forth the corresponding recommendations for improving the comprehensive food production capacity.

A New Time-Aware Collaborative Filtering Intelligent Recommendation System

Aiming at the problem that the traditional collaborative filtering recommendation algorithm does not fully consider the influence of correlation between projects on recommendation accuracy,this paper introduces project attribute fuzzy matrix,measures the project relevance through fuzzy clustering method,and classifies all project attributes.Then,the weight of the project relevance is introduced in the user similarity calculation,so that the nearest neighbor search is more accurate.In the prediction scoring section,considering the change of user interest with time,it is proposed to use the time weighting function to improve the influence of the time effect of the evaluation,so that the newer evaluation information in the system has a relatively large weight.The experimental results show that the improved algorithm improves the recommendation accuracy and improves the recommendation quality.

一种联合LTR和社交网络的Top-k推荐方法

个性化推荐可以有效地解决网络中数据爆炸问题,绝大多数的工作利用用户评分信息来降低评分预测误差值(MAE).相比于精确的评分预测,用户更关心推荐给他的top-k项目排序列表,而最小化MAE并不意味着良好的top-k项目列表.由于评分数据存在稀疏问题,社交网络中信任信息在top-k推荐系统中起着越来越重要的作用.本文提出了一种基于信任的面向top-k排序的推荐方法,BTRank.该算法基于LTR方法,结合用户评分以及用户信任信息来构建项目排序模型,有效地提高对所有用户的top-k排序列表质量.同时,考虑到用户兴趣会随着时间演变而变化,本文设计了时间效应模型函数用于处理用户历史评分数据.在真实数据集上的实验中表明本文提出的算法效果明显优于传统的推荐算法以及同类top-k排序推荐算法.

融合用户兴趣漂移的Top-N推荐算法

传统推荐算法多基于用户兴趣的静态属性获得用户偏好,忽略了用户兴趣漂移问题,为此,提出了解决该问题的融合用户兴趣漂移的Top-N推荐算法。利用长短期记忆网络(LSTM,long short term memory)处理时序数据的优势表示用户短期兴趣漂移规律,用矩阵分解得到的固定向量表示用户的长期兴趣,将注意力机制纳入LSTM隐藏状态的表示中来获取用户长短期兴趣关联。实验结果表明,所提算法与当前流行算法相比,在Top-N项目推荐中具有更优性能。

时间上下文的协同过滤Top-N推荐算法

在推荐系统中,通过收集和分析用户在系统中的所有行为信息,创建用户独有的偏好模型,从而根据该模型推断出用户可能感兴趣的物品。传统协同推荐算法一般都是利用收集的用户行为信息,根据偏好模型分析用户的行为特点,筛选出向用户推荐的物品列表,但推荐列表大同小异。为了提高推荐的准确性和精确性,让用户在不同的时间可以看到不同的推荐结果,提出了以传统协同过滤算法为基础的改进算法。在分析用户行为信息,建立用户行为特征时,考虑不同时间下用户历史信息也不同,时间越近越能反映当前用户行为特征。用户在较短时间间隔内感兴趣的物品,具有更高的相似度。故以时间作为权重因子引入到算法中,加重近期行为在算法中所占的比重,优先向用户推荐与他浏览过的物品类似的物品,从而提高推荐物品的多样性。在典型数据集上的实验表明,在保证推荐准确度的前提下,融合时间的推荐算法准确率和召回率明显提高,验证了该算法的有效性。

Carbon emission scenarios of China's power sector:Impact of controlling measures and carbon pricing mechanism

The study constructs a low-carbon path analysis model of China's power sector based on TIMES model and presents a comparative analysis of carbon emissions under Reference,Low-Carbon and Enhanced Low-Carbon scenarios,and the main difference of the three scenarios is manifested by policy selection and policy strength.The conclusions are drawn as follows:(1)The peak of carbon emission in China's power sector will range from 4.0 GtCO2 to 4.8 GtCO2,which implies an increment of 0.5e1.3 billion or 14%e35%from the 2015 levels.(2)Introducing carbon price is an effective way to inhibit coal power and promote non-fossil fuels and Carbon Capture,Utilization and Storage applications(CCUS).The carbon emission reduction effects will gradually increase with carbon price.When the carbon price attains to CN￥150 t
1CO2,the CO2 emission can decrease by 36%than that without carbon price.(3)CCUS is one of important contributing factor to reduce CO2 emission in power sector.Generally speaking,the development of non-fossil fuels and energy efficiency improvement are two main drivers for carbon mitigation,but once the carbon price reaches up to CN￥106 t
1CO2,the CCUS will be required to equip with thermal power units and its contribution on carbon emission reduction will remarkably increase.When carbon price increases to CN￥150 t
1CO2 in 2050,the application of CCUS will account for 44%of total emission reduction.(4)In the scenario with carbon price of CN￥150 t
1CO2,power sector would be decarbonized significantly,and the CO2 intensity will be 0.22 kgCO2(kW h)
1,but power sector is far from the goal that achieving net zero emission.In order to realize the long-term low greenhouse gas emission development goal that proposed by the Paris Agreement,more efforts are needed to be put to further reduce the carbon emission reduction of power sector.Based on the above scenario analysis,the study proposes four recommendations on the low-carbon development of China's power sector:(1)improve the energy efficiency proactively and optimize the energy structure of power sector gradually;(2)promote the low-carbon transition of power sector by using market-based mechanism like carbon emission trading scheme to internalize the external cost of carbon emission;(3)give more emphasis on and support to the CCUS application in power sector.

模糊TOPSIS时变权重二次量化云服务推荐

云服务数量的快速增长给云服务选择决策带来了挑战。为协助决策者在进行云服务选择时提供决策依据,更好地满足用户使用需求,提出一种并行模糊TOPSIS时变权重二次量化云服务推荐算法。首先,基于熵的权重赋值方法确定准则权重,解决了准则权重不好选取的问题,并采用时变权重方法作为补充设计一种新的模糊TOPSIS评估方案,时变权重的引入可以对不同时期QoS等信息的重要性进行二次量化区分,有助于提高融合决策的合理性。其次,基于时间段的模糊TOPSIS多准则融合决策信息,提出一种可并行云服务选择框架,有利于算法的实际推广应用。实验结果表明,该算法能够有效抑制不良QoS信息影响,提高服务调用成功率保证诚信服务的互操作性。

Research of Collaborative Filtering Recommendation Algorithm for Short Text

Short text, based on the platform of web2.0, gained rapid development in a relatively short time. Recommendation systems analyzing user’s interest by short texts becomes more and more important. Collaborative filtering is one of the most promising recommendation technologies. However, the existing collaborative filtering methods don’t consider the drifting of user’s interest. This often leads to a big difference between the result of recommendation and user’s real demands. In this paper, according to the traditional collaborative filtering algorithm, a new personalized recommendation algorithm is proposed. It traced user’s interest by using Ebbinghaus Forgetting Curve. Some experiments have been done. The results demonstrated that the new algorithm could indeed make a contribution to getting rid of user’s overdue interests and discovering their real-time interests for more accurate recommendation.

基于实时路况的top-κ载客热门区域推荐

为降低城市出租车的空载率,缓解路网交通拥堵压力,亟需设计有效的出租车载客热门区域推荐方法.针对传统的出租车相关推荐方法忽略实际路况导致推荐精度较低的现状,提出了一个两阶段的载客热门区域实时推荐算法.首先,离线挖掘阶段,基于出租车历史轨迹数据集提取基于时段属性的载客热门区域;随后,在线推荐阶段,根据出租车请求位置及时间,结合实时路况设计潜在空载时间开销函数T_(cost)对载客热门区域进行评测排序,继而发现Top-κ载客热门区域.基于出租车轨迹数据集的实验结果表明,结合实时交通状况的Top-κ载客热门区域推荐方法以确保较小潜在空载时间开销,相较于传统的出租车推荐方法具有较好的有效性与鲁棒性.

Association between Ambient Air Pollution and Outpatient Visits for Acute Bronchitis in a Chinese City

Objective To investigate the short-term association between outdoor air pollution and outpatient visits for acute bronchitis,which is a rare subject of research in the mainland of China.Methods A time-series analysis was conducted to examine the association of outdoor air pollutants with hospital outpatient visits in Shanghai by using two-year daily data（2010-2011）.Results Outdoor air pollution was found to be associated with an increased risk of outpatient visits for acute bronchitis in Shanghai.The effect estimates of air pollutants varied with the lag structures of the concentrations of the pollutants.For lag06,a 10 μg/m3 increase in the concentrations of PM10,SO2,and NO2 corresponded to 0.94%（95% CI：0.83%,1.05%）,11.12%（95% CI：10.76%,11.48%）,and 4.84%（95% CI：4.49%,5.18%） increases in hospital visits for acute bronchitis,respectively.These associations appeared to be stronger in females（P〈0.05）.Between-age differences were significant for SO2（P〈0.05）,and between-season differences were also significant for SO2（P〈0.05）.Conclusion Our analyses have provided the first evidence that the current air pollution level in China has an effect on acute bronchitis and that the rationale for further limiting air pollution levels in Shanghai should be strengthened.

User space transformation in deep learning based recommendation

Deep learning based recommendation methods, such as the recurrent neural network based recommendation method(RNNRec) and the gated recurrent unit(GRU) based recommendation method(GRURec), are proposed to solve the problem of time heterogeneous feedback recommendation. These methods out-perform several state-of-the-art methods. However, in RNNRec and GRURec, action vectors and item vectors are shared among users. The different meanings of the same action for different users are not considered. Similarly, different user preference for the same item is also ignored. To address this problem, the models of RNNRec and GRURec are modified in this paper. In the proposed methods, action vectors and item vectors are transformed into the user space for each user firstly, and then the transformed vectors are fed into the original neural networks of RNNRec and GRURec. The transformed action vectors and item vectors represent the user specified meaning of actions and the preference for items, which makes the proposed method obtain more accurate recommendation results. The experimental results on two real-life datasets indicate that the proposed method outperforms RNNRec and GRURec as well as other state-of-the-art approaches in most cases.

结合图卷积神经网络和集成方法的推荐系统恶意攻击检测

推荐系统已被广泛应用于电子商务、社交媒体、信息分享等大多数互联网平台中,有效解决了信息过载问题。然而,这些平台面向所有互联网用户开放,导致不法用户利用系统设计缺陷通过恶意干扰、蓄意攻击等行为非法操纵评分数据,进而影响推荐结果,严重危害推荐服务的安全性。现有的检测方法大多都是基于从评级数据中提取的人工构建特征进行的托攻击检测,难以适应更复杂的共同访问注入攻击,并且人工构建特征费时且区分能力不足,同时攻击行为规模远远小于正常行为,给传统检测方法带来了不平衡数据问题。因此,文中提出堆叠多层图卷积神经网络端到端学习用户和项目之间的多阶交互行为信息得到用户嵌入和项目嵌入,将其作为攻击检测特征,以卷积神经网络作为基分类器实现深度行为特征提取,结合集成方法检测攻击。在真实数据集上的实验结果表明,与流行的推荐系统恶意攻击检测方法相比,所提方法对共同访问注入攻击行为有较好的检测效果并在一定程度上克服了不平衡数据的难题。

分时段预约胸外科门诊迟到患者特征的回顾性分析

目的分析在分时段预约中四川大学华西医院胸外科门诊患者迟到的特征。方法从四川大学华西医院HIS中提取2019年胸外科患者人口信息及预约就诊信息,共获得有效数据9378例。以正常患者为对照组,采用SPSS 24.0软件对迟到患者进行分析。结果年龄、性别、来源地、就诊类型、就诊时间、预约方式、专家级别、就诊时段是患者迟到的影响因素,差异均有统计学意义(P<0.05)。患者在周一迟到率最高,达到34.6%。每天上午时段的>08:30—09:00迟到率最高,达到78.1%;下午时段的>13:00—13:30迟到率最高,达到65.7%。结论在预约就诊时段内迟到的患者具有一定的特征,需根据相关特征提出解决方案,以提高患者在预约时间段内就诊的依从性。

基于知识图谱的个性化音乐推荐系统设计与实现

本文基于知识图谱技术设计了个性化音乐推荐系统。该系统由用户端、管理端、服务端以及数据库组成。用户完成注册并登录系统后,可以通过输入关键词等方式检索对应的歌曲和歌单,并将其添加收藏,管理员可查看并编辑用户信息。当用户再次登录该系统时,系统会根据之前用户的操作行为向用户精准推送个性化的歌曲和歌单,并在网站主页上显示。从测试情况来看,该系统的音乐推荐、歌曲搜索、歌曲收藏等基本功能可以实现,系统各项操作的平均响应时间均在1 s以内,使用体验较好。

融合时序门控图神经网络的兴趣点推荐方法

现有的大多数兴趣点(point of interest,POI)推荐系统由于忽略了用户签到序列中的顺序行为模式,以及用户的个性化偏好对于POI推荐的影响,导致POI推荐系统性能较低,推荐结果不可靠,进而影响用户体验。为了解决上述问题,提出一种融合时序门控图神经网络的兴趣点推荐方法。运用时序门控图神经网络(temporal gated graph neural network,TGGNN)学习POI embedding;采用注意力机制捕获用户的长期偏好;通过注意力机制融合用户的最新偏好和实时偏好,进而捕获用户的短期偏好。通过自适应的方式结合用户的长期和短期偏好,计算候选POI的推荐得分,并根据得分为用户进行POI推荐。实验结果表明,与现有方法相比,该方法在召回率和平均倒数排名这两项指标上均有较为明显的提升,因此可以取得很好的推荐效果,具有良好的应用前景。

基于长短期偏好注意力网络的兴趣点推荐

兴趣点(Point-Of-Interest,POI)推荐是基于位置的社交网络(Location-based Social Networks,LBSNs)研究中最重要的任务之一。为了解决POI推荐中的空间稀疏性问题,提出一种用于位置推荐的长短期偏好时空注意力网络(LSAN)。首先,构建了签到序列的时空关系矩阵,使用多头注意力机制从中提取非连续签到和非相邻位置中的时空相关性,缓解签到数据稀疏所带来的困难。其次,在模型中设置用户短期偏好和长期偏好提取模块,自适应的将二者结合在一起,考虑了用户偏好对用户决策影响。最后,在Foursquare数据集上进行测试,并与其它模型结果进行对比,证实了提出的LSAN模型结果最优。研究表明LSAN模型能够获得最佳的推荐效果,为POI推荐提供新思路。

融合时间感知和多兴趣提取网络的序列推荐

针对序列推荐任务中的时间动态性和多重兴趣建模问题,提出一种时间感知的项目嵌入方法,用于学习项目之间的时间关联性。在此基础上,提出一种融合时间感知和多兴趣提取网络的序列推荐(time-aware multi-interest sequence recommendation,TMISA)方法。TMISA采用自注意力序列推荐模型作为局部特征学习模块,以捕捉用户行为序列中的动态偏好;通过多兴趣提取网络对用户的全局偏好进行建模;引入门控聚合模块将局部和全局特征表示动态融合,生成最终的用户偏好表示。实验证明,在5个真实推荐数据集上,TMISA模型表现出卓越性能,超越了多个先进的基线模型。

时间感知的用户长短期兴趣特征分离推荐算法

针对传统推荐算法未充分考虑行为序列间的动态时间间隔、语义不规则以及用户长短期兴趣相互纠缠的问题,提出一种时间感知的用户长短期兴趣特征分离推荐算法。利用用户个性化时间聚合间隔感知和时间位置多头注意力捕获长期兴趣特征,采用动态时间间隔感知和潜在意图注意力的Time-LSTM捕获短期兴趣特征,提出长短期兴趣特征分离获取方法,分别独立捕获两种时间尺度的用户兴趣,通过注意力机制自适应融合长短期兴趣特征,提高用户兴趣特征捕获准确率。实验结果表明,该算法在预测精度指标AUC和GAUC上较对比算法均有提升,消融实验也进一步验证了该算法的必要性。

宫颈环扎术治疗宫颈机能不全的研究进展

宫颈机能不全(CIC)常表现为反复发生妊娠中晚期无痛性宫颈管缩短、扩张,最终导致难免流产或早产。早产儿的出生严重影响人口质量和健康素质,在妊娠22~26周期间每延迟分娩1 d,新生儿可增加3%的存活率。宫颈环扎术可有效延长妊娠周数、降低流产与早产的发生率。因此选择合适的手术时机、进行规范的术后管理对提高宫颈环扎术的成功率、延长妊娠周数至关重要。本文结合国内外相关临床研究与最新指南,综述宫颈环扎术的手术时机、术后药物管理及术后随访等。