在施工场景中,正确有效的检测工人佩戴安全帽是工地安全重要的一环.基于人工检测安全帽佩戴情况,既耗时耗力,监管效率也不高.为了降低因为安全帽佩戴问题导致的施工问题,本文采用目标检测算法中的YoloV3算法进行安全帽检测,利用闸机处情境照片作为原始数据集,采用LabelImg对原始照片进行安全帽区域的人工标注,并确定边界框(bounding box)的个数以及位置,建立了人工标注的安全帽检测的训练数据集和测试集.利用上述的训练数据,训练安全帽检测算法,并进行检测.实验结果显示:本文的安全帽检测算法的mAP值达到98％,检测速率为20fps,该算法在取得了较高准确率的同时,也满足了实时性的要求.

证券行业存在着大量的标签信息,帮助投资者进行股票的标签分类、板块划分.如何挖掘证券的标签信息,对于帮助投资者更快、更准地了解证券市场、板块、股票的情况,具有重要意义.本文主要研究证券标签信息的挖掘分析技术,包含:标签信息获取、股票之间、标签之间的关联性分析、股票网络和标签网络分析,以及应用上述分析技术,实现了一种证券市场的标签分析技术.

OBJECTIVE:Atherosclerosis (AS) is the most dangerous factor for human death, which is responsible for coronary heart disease. Growing evidence has showed that long non-coding RNAs (lncRNAs) are involved in the development of AS. In this study, we mainly aimed at investigating the roles of FOXC2-AS1 in AS patients.PATIENTS AND METHODS:RT-PCR was performed to detect the expressions of FOXC2-AS1 and miR-1253 in serum samples of AS patients (n=35) and healthy volunteer (n=35). The correlation between FOXC2-AS1 and miR-1253 was further analyzed. Human vascular smooth muscle cells (VSMCs) were respectively treated with ox-LDL, IL-6, CRP, TNF-α and IL-8 to explore the affecting factors. P-FOXC2-AS1 was constructed and transfected into VSMCs. Cell proliferation abilities were measured by CCK-8 assay. Cell apoptotic rates were measured by flow cytometry (FACS) analysis. Western blot (WB) was performed to detect protein levels of FOXF1, Bcl-2, Bax and Cleaved Caspase3. Finally, luciferase gene reporter assay was performed to prove the relationships between FOXC2-AS1 and miR-1253, miR-1253 and FOXF1.RESULTS:We found that FOXC2-AS1 was significantly upregulated in AS patients, which could be induced by ox-LDL and IL-6 in VSMCs. MiR-1253 was decreased in AS patients, which was negatively correlated with FOXC2-AS1. Furthermore, FOXC2-AS1 overexpression promoted proliferation and inhibited apoptosis in VSMCs. Luciferase gene reporter assay showed that FOXC2-AS1 could bind to miR-1253 in VSMCs and 293 cells. Moreover, miR-1253 overexpression inhibited proliferation and promoted apoptosis of VSMCs. Luciferase reporter assay proved that miR-1253 could target at FOXF1 in VSMCs and 293 cells, which was reported to be associated with cell proliferation and apoptosis in some cancers. Additionally, miR-1253 mimic or GSK343, a FOXF1 inhibitor, was respectively transfected into VSMCs with p-FOXC2-AS1. Results showed that the promoted cell proliferation and inhibited cell apoptosis were reversed as well, confirming that FOXC2-AS1 promoted cell proliferation and inhibited apoptosis via miR-1253/FOXF1 signaling axis in AS patients.CONCLUSIONS:According to the results, we found that FOXC2-AS1 was upregulated in AS patients; furthermore, FOXC2-AS1 overexpression promoted cell proliferation and inhibited cell apoptosis via targeting miR-1253/FOXF1 signaling axis. Our results elucidated a potential mechanism underlying the role of FOXC2-AS1, which might be used as a promising marker and a potential target for AS patients.

In the present research, carbon nanotubes (CNTs) and graphene oxide (GO) were used to enhance the composites interfacial properties synergistically by the strategy combining electrodeposition and chemical grafting. It was found that the interface between CF and matrix can be tuned by the binary addition of CNTs and GO. The interfacial shear strength (IFSS) of GO/CNTs synergistic reinforced composites based on this physical and chemical combination method via the “bridge” of Ag nanoparticles (NPs) exhibits 42.99% and 4.61% enhancement compared to that of achieved by electrophoretic deposition (physical method) and chemical grafting (chemical method), respectively. In addition, the IFSS has 69.68% and 19.54% improvement due to the synergistic effect of GO/CNTs compared to the untreated CF and GO single-reinforced CF composites. Therefore, it is a promising technique for developing multifunctional CF with the excellent interfacial performance.

Online social networks mainly have two functions: social interaction and information diffusion. Most of current link recommendation researches only focus on strengthening the social interaction function, but ignore the problem of how to enhance the information diffusion function. For solving this problem, this paper introduces the concept of user diffusion degree and proposes the algorithm for calculating it, then combines it with traditional recommendation methods for reranking recommended links. Experimental results on Email dataset and Amazon dataset under Independent Cascade Model and Linear Threshold Model show that our method noticeably outperforms the traditional methods in terms of promoting information diffusion.

The development of well-ordered TiO2 nanotubes arrays (TNTAs) as a binder-free electrode in supercapacitors was hindered by their poor conductivity and low capacitance. Herein, the hierarchical structure of TNTAs/C/MnO2 (TNTCM) electrode was built via three-step process - typical anodization, carbon deposition and electrodeposition method. TiO2 nanotube arrays obtained from anodization provide a high surface area substrate for binder-free electrode, followed by carbon nanoparticles which could act as an electronic transfer medium penetrated into TNTAs using gas thermal penetration, the TNTAs/C electrode became high conductivity, then the two-dimension birnessite-type MnO2 nanoflakes were in situ growth onto TNTAs/C composites through scalable and easy electrodeposition method could storage energy by faradaic reaction. The resultant TNTCM hybrid composites manifest a remarkable specific areal capacitance of 492 mF/cm(2), 207 times than that of TNTAs/C electrode, the value is comparable or much higher than those of previously reported TiO2 nanotubes-based electrode for supercapacitor. The synthesized TNTCM electrode exhibit a high energy density of 465 mWh/m(2) at power density of 2.5 W/m(2). In addition, the excellent energy storage performance is well maintained with a capacitance retention of 98% during 3000 charge-discharge cycles, indicating its promising application in energy storage and conversion fields. (C) 2019 Elsevier B.V. All rights reserved.

An adaptive updating learning and evaluating method for user interest model is proposed. In order to provide users with more accurate search results,the user interest model is verified after adding adaptive adjustment algorithm. Through the analysis of user short-term interest and long-term interest in law,it becomes interested in model of users of the system. With the change of time,the user' s in-terests will change accordingly. We analyze the algorithm of user' s interest by the adaptive learning process,in which the rules change, so as to obtain the user' s interest points. We also research on interest learning technology and evaluate it. Main parameters like precision is calculated,and the evaluation shows the user interest mining precision rate is better,providing a well solution for the modern computer network shopping and network application and process of mining user behavior and interest,with aid to personalized recommendation ap-plication.

Learning distributed representations of symbolic data were introduced by Hinton[1], and first developed in modeling networks for learning the node vectors by Perozzi et al (2014). In this work, we proposed Dnps, a novel nodes embedding approach for acquiring distributed representations of large-scale dynamic social networks. Dnps is suitable for many types of social networks: dynamic/static, directed/undirected, and weighted/unweighted. Recently, several works of nodes embedding were proposed. However, they were designed for static networks, such as language networks. To address this problem, first, we develop a damping based positive sampling (DpS) algorithm to learn the hierarchical structure of social networks. Then, we devise a local search based DpS algorithm to obtain incremental information of network evolution. Finally, we show Dnps's potentials on future link prediction task for three real-life large-scale dynamic social networks. The results show that Dnps consistently outperforms all baseline methods and exhibits an improvement of 12%, 6%, 4% on Digg, Flickr and YouTube over the second-highest level, respectively. Moreover, Dnps is also scalable. For example, Dnps can speed up the training process in 2 ~ 36 times compared with benchmarks on Flickr network. The source codes of the project is available online 1 .

The rapid development of online social networks (e.g., Twitter and Facebook) has promoted research related to social networks in which link prediction is a key problem. Although numerous attempts have been made for link prediction based on network structure, node attribute and so on, few of the current studies have considered the impact of information diffusion on link creation and prediction. This paper mainly addresses Sina Weibo, which is the largest microblog platform with Chinese characteristics, and proposes the hypothesis that information diffusion influences link creation and verifies the hypothesis based on real data analysis. We also detect an important feature from the information diffusion process, which is used to promote link prediction performance. Finally, the experimental results on Sina Weibo dataset have demonstrated the effectiveness of our methods.

微博用户的兴趣分析和模型表示是用户关系分析的基础,而用户关系分析又构成了微博社会网络的生成和分析的基础.该文主要讨论微博的用户关系分析技术.作者将微博社会网络视为一个加权无向图,节点表示用户,边表示用户之间的关系,边的权值表示用户之间的关系强度.该文将用户关系强度定义为用户之间的相似度,分别给出了基于各种用户属性信息（背景信息、微博文本、社交信息）的用户相似度计算方法,并通过实验系统性对比了上述方法的优劣.实验结果显示:基于社交信息的用户相似度在用户关系分析方面取得了最好的效果.为了进一步验证上述用户相似度的实际性能,该文将它们应用于用户推荐的相关实验,基于社交信息的用户相似度又取得了最好的推荐效果.最后,该文应用基于社交信息的用户相似度生成了微博的社会网络（称作用户相似性网络）,在该社会网络上进行了团体挖掘的实验,实验结果显示了该相似度在团体挖掘上的有效性.

Influence maximization in social networks has been widely studied motivated by applications like spread of ideas or innovations in a network and viral marketing of products. Current studies focus almost exclusively on unsigned social networks containing only positive relationships (e.g. friend or trust) between users. Influence maximization in signed social networks containing both positive relationships and negative relationships (e.g. foe or distrust) between users is still a challenging problem that has not been studied. Thus, in this paper, we propose the polarity-related influence maximization (PRIM) problem which aims to find the seed node set with maximum positive influence or maximum negative influence in signed social networks. To address the PRIM problem, we first extend the standard Independent Cascade (IC) model to the signed social networks and propose a Polarity-related Independent Cascade (named IC-P) diffusion model. We prove that the influence function of the PRIM problem under the IC-P model is monotonic and submodular Thus, a greedy algorithm can be used to achieve an approximation ratio of 1-1/e for solving the PRIM problem in signed social networks. Experimental results on two signed social network datasets, Epinions and Slashdot, validate that our approximation algorithm for solving the PRIM problem outperforms state-of-the-art methods.

A growing number of people are paying attention to the microblog,which is an important tool in social media.Comparing with conventional network media,the information propagation on microblog presents some new features such as the big volume of data,rapidity,and timeliness.This paper mainly discusses the methods to analyze information propagation trees,and puts forward two algorithms to generate the trees:a generating algorithm based on reposting relations,and a fast generating algorithm.By experiments,the paper compares the performance of the two algorithms and analyzes the distribution of nodes in the propagation trees.

In order to solve the poor performance in text classification when using traditional formula of mutual information （MI）,a feature selection algorithm were proposed based on improved mutual information.The improved mutual information algorithm,which is on the basis of traditional improved mutual information methods that enhance the MI value of negative characteristics and feature’s frequency,supports the concept of concentration degree and dispersion degree.In accordance with the concept of concentration degree and dispersion degree,formulas which embody concentration degree and dispersion degree were constructed and the improved mutual information was implemented based on these.In this paper,the feature selection algorithm was applied based on improved mutual information to a text classifier based on Biomimetic Pattern Recognition and it was compared with several other feature selection methods.The experimental results showed that the improved mutual information feature selection method greatly enhances the performance compared with traditional mutual information feature selection methods and the performance is better than that of information gain.Through the introduction of the concept of concentration degree and dispersion degree,the improved mutual information feature selection method greatly improves the performance of text classification system.

In order to get information from the internet more quickly and accurately,this paper designs and implements a topic detection and tracking system over a vast amount of web pages.On the basis of the system it also proposes the algorithm selection strategy for topic detection in the internet and a topic tracking model based on multiple features.The topic tracking model can validly distinguish the same topic from the similar topic,and the accuracy of topic tracking achieves 85.7%.The experimental results show that the system in the paper has good performance of topic detection and tracking.

This paper proposes a method to build and analyze social networks of person entities on Wikipedia. Here each person entity is represented by a few attributes. Different estimation approaches of attribute similarity are used, on which we employ the Systematic Similarity Measure theory to compute the person entity similarity. On the basis of the similarity array of person entities, we build the social network for them. On Wikipedia data, we conduct some experiments on social network analysis, and the experimental results show our social network mining approaches are effective.

Automatic web site classification has a wide application prospect; however, there are few researches on it. Different from pure texts, web sites are the combination of a large number of web pages via hyperlinks, so text classification methods are not suitable to classify them directly. This paper proposes a web site classification approach based on its topological structure. Given a web site, firstly we represent its topological structure as a directed graph, and from which we extract a strongly connected sub-graph including the site’s home page. Secondly, we use an improved PageRank algorithm on the sub-graph to select some topic-relevant resources, and represent them as a topic vector of the site. Finally we use an SVM classifier to classify the site in term of its topic vector. Some experiments are conducted for web site classification. Experimental results show our approach achieved better performance than traditional super page-based web site classification approach.