基于网络访问行为的混合阶Markov预测模型

随着WWW的迅速发展和网络用户的急剧增加,准确预测Web用户的访问行为对减小用户的感知延时,实现个性化推荐等具有重要的作用.无论是Markov模型还是其任何一种变种,高阶模型具有较好的预测性能.然而,高阶模型通常有较高的状态空间复杂度.提出了一种新的混合阶Markov模型(HMPM),将前缀相同的序列共享存储,降低了状态空间复杂度.仿真实验结果表明,该模型在一定程度上提高了预测准确率,查全率也有所提升.     

QoS prediction for web services based on user-trust propagation model

There is an important online role for Web service providers and users; however, the rapidly growing number of service providers and users, it can create some similar functions among web services. This is an exciting area for research, and researchers seek to to propose solutions for the best service to users. Collaborative filtering (CF) algorithms are widely used in recommendation systems, although these are less effective for cold-start users. Recently, some recommender systems have been developed based on social network models, and the results show that social network models have better performance in terms of CF, especially for cold-start users. However, most social network-based recommendations do not consider the user’s mood. This is a hidden source of information, and is very useful in improving prediction efficiency. In this paper, we introduce a new model called User-Trust Propagation (UTP). The model uses a combination of trust and the mood of users to predict the QoS value and matrix factorisation (MF), which is used to train the model. The experimental results show that the proposed model gives better accuracy than other models, especially for the cold-start problem.     

基于层次随机图模型的脑网络链路预测

针对一般链路预测算法在具有层次结构的脑网络中计算效率低且复杂度高的问题,提出了一种基于最大似然估计的层次随机图模型。该算法首先利用脑网络数据建立层次随机图,然后通过改进的马尔科夫蒙特卡罗算法采样树状图空间,最后计算脑网络边的平均连接概率,且通过评价指标对算法进行评价。实验结果表明,利用该算法对脑网络和3种不同的层次结构网络进行链路预测比较, 脑网络的预测结果最好。此外,所提出的算法较之传统的基于相似性的算法,该算法效果明显,且具有理想的计算复杂度。     

Image recommendation based on a novel biologically inspired hierarchical model

Image recommendation has become an increasingly relevant problem recently, since strong demand to quickly find interested images from vast amounts of image library. We describe a biologically inspired hierarchical model for image recommendation. The biologically inspired model (BIM) for invariant feature representation has attracted widespread attention, which approximately follows the organization of cortex visuel. BIM is a computation architecture with four layers. With the image data size increases, the four-layer framework is prone to be overfitting, which limits its application. To address this issue, we propose a biologically inspired hierarchical model (BIHM) for feature representation, which adds two more discriminative layers upon the conventional four-layer framework. In contrast to the conventional BIM that mimics the inferior temporal cortex, which corresponds to the low level feature, the proposed BIHM adds two more layers upon the conventional framework to simulate inferotemporal cortex, exploring higher level feature invariance and selectivity. Furthermore, we firstly utilize the BIHM in the image recommendation. To demonstrate the effectiveness of proposed model, we use it to image classification and retrieval tasks and perform experiments on CalTech5, Imagenet and CalTech256 datasets. The experiment results show that BIHM exhibits better performance than the conventional model in the tasks and is very comparable to existing architectures.     

TL-FCM: A hierarchical prediction model based on two-level fuzzy c-means clustering for bike-sharing system

In recent years, shared bikes have gradually emerged into public life as a new way to travel and helped solve the last-mile problem of residents’ travel. While this development has brought convenient travel to users, a series of problems exist, a prominent one is the uneven distribution of bikes at each shared bike station. Accurately predicting bike usage in a bike-sharing system can help solve this problem. In this paper, we investigate how to improve the accuracy of predicting the usage of bikes in bike-sharing system. First, considering both geographic location information of shared bike stations and the migration trend of bikes between stations, we design a two-level fuzzy c-means clustering algorithm to cluster shared bicycle stations into groups, which can better capture the connection between shared bicycle stations and improve the clustering accuracy of shared bicycle sites, then, we combine the two-level fuzzy c-means clustering algorithm with the multi-similarity reference model to predict the usage of bikes, which can significantly improve the accuracy of the forecast. To evaluate the performance of our model, we validate our model in the New York Bike-Sharing System. The results shows that our model obtained significantly better results than other models.

     

Markov网页预测模型综述

介绍了基本的Markov浏览预测模型;讨论了扩展的Markov浏览预测模型,包括隐Markov模型、多Markov模型、混合模型、结构相关性模型;综述了各个模型的算法及其优缺点;分析了Markov浏览预测模型需要深入研究的问题.     

基于分层学习的易混淆法条预测

目前针对法条预测的相关研究大都采用文本分类的思想,但模型构建过程都未考虑不同法条之间的从属关系或相似程度,因此对于易混淆法条预测效果普遍较差。针对现有方法在易混淆法条预测中存在的不足,提出基于分层学习的易混淆法条预测方法。将法条分为易区分法条和易混淆法条,按法条内容将易混淆法条组合为不同易混淆法条集并分别训练易混淆法条集预测模型,运用分层学习完成易混淆法条预测。在刑事案件的数据上进行实验,实验结果表明,该模型能较好解决易混淆法条预测问题,提高法条预测准确率。     

基于分级置信度传播的立体匹配新方法

针对当前全局立体匹配算法效率低下以及在遮挡区域难以准确匹配的问题,提出了一种高效的基于分层置信度传播的立体匹配新方法。该方法首先采用高效局部匹配获得初 始匹配代价,并根据初始匹配的结果检测遮挡和误匹配像素;其次,使用平面化估计的结果对初始代价进行了修正,改善了遮挡区域的匹配效果;最后使用改进的分层置信度传播 快速估计最小化能量,降低了低纹理区域的误匹配,提高了整体的平滑度。实验结果表明,该方法在精度和速度上具有较高的性能。     

A novel visualization model for web search results

This paper presents an interactive visualization system, named WebSearchViz, for visualizing the Web search results and acilitating users' navigation and exploration. The metaphor in our model is the solar system with its planets and asteroids revolving around the sun. Location, color, movement, and spatial distance of objects in the visual space are used to represent the semantic relationships between a query and relevant Web pages. Especially, the movement of objects and their speeds add a new dimension to the visual space, illustrating the degree of relevance among a query and Web search results in the context of users' subjects of interest. By interacting with the visual space, users are able to observe the semantic relevance between a query and a resulting Web page with respect to their subjects of interest, context information, or concern. Users' subjects of interest can be dynamically changed, redefined, added, or deleted from the visual space.     

A novel bankruptcy prediction model based on an adaptive fuzzy k-nearest neighbor method

Bankruptcy prediction is one of the most important issues in financial decision-making. Constructing effective corporate bankruptcy prediction models in time is essential to make companies or banks prevent bankruptcy. This study proposes a novel bankruptcy prediction model based on an adaptive fuzzy k-nearest neighbor (FKNN) method, where the neighborhood size k and the fuzzy strength parameter m are adaptively specified by the continuous particle swarm optimization (PSO) approach. In addition to performing the parameter optimization for FKNN, PSO is also utilized to choose the most discriminative subset of features for prediction. Adaptive control parameters including time-varying acceleration coefficients (TVAC) and time-varying inertia weight (TVIW) are employed to efficiently control the local and global search ability of PSO algorithm. Moreover, both the continuous and binary PSO are implemented in parallel on a multi-core platform. The proposed bankruptcy prediction model, named PTVPSO-FKNN, is compared with five other state-of-the-art classifiers on two real-life cases. The obtained results clearly confirm the superiority of the proposed model in terms of classification accuracy, Type I error, Type II error and area under the receiver operating characteristic curve (AUC) criterion. The proposed model also demonstrates its ability to identify the most discriminative financial ratios. Additionally, the proposed model has reduced a large amount of computational time owing to its parallel implementation. Promisingly, PTVPSO-FKNN might serve as a new candidate of powerful early warning systems for bankruptcy prediction with excellent performance.     

Multi-user web service selection based on multi-QoS prediction

In order to find best services to meet multi-user’s QoS requirements, some multi-user Web service selection schemes were proposed. However, the unavoidable challenges in these schemes are the efficiency and effect. Most existing schemes are proposed for the single request condition without considering the overload of Web services, which cannot be directly used in this problem. Furthermore, existing methods assumed the QoS information for users are all known and accurate, and in real case, there are always many missing QoS values in history records, which increase the difficulty of the selection. In this paper, we propose a new framework for multi-user Web service selection problem. This framework first predicts the missing multi-QoS values according to the historical QoS experience from users, and then selects the global optimal solution for multi-user by our fast match approach. Comprehensive empirical studies demonstrate the utility of the proposed method.     

一种基于分层模型的TSP构建算法

提出了一种新算法,有效地减少了最近邻域法和贪婪算法在构建旅行商问题可行解过程中引入不合理长边的问题。该算法先借助一种由伪凸包算子所得到的分层模型对旅行商问题中的城市分布进行分析,之后通过将分层模型中相对外层的点逐个添加到内层的规则得到可行解。借助仿真实验求解TSPLIB标准库中的40实例,并与最近邻域法和贪婪算法进行对比,结果表明分层融合算法具有更高的精度,其平均求解质量达到8.47%。     

关于跨站脚本问题的研究

现在的网站包含有若干动态内容,方便了用户的操作,但是动态网站有一个危险,叫做”Cross Site Scripting”(跨站脚本),而静态网站没有。当前有一些关于跨站脚本漏洞的信息资料,但能真正给用户和管理者解释清楚的并不多。现就这一问题进行整理说明,给出这种隐藏危险的一个清楚的了解,并给出一些经验以便如何去发现和阻止它。     

一种基于语义Web环境的电子商务模型

语义Web不仅能为人类用户而且能为软件Agent提供从语法层次到语义层次上的互操作性。语义Web与Web-Services有很强的互补性,现有的技术条件使二者的统一成为一种可能,这给电子商务的发展带来了一次历史性的革命机遇。给出了语义Web环境下结合Web Services关键技术和智能主体技术的一种极具潜能的电子商务实现模型。     

一种基于QoS的Web服务发现模型

当前采用 OWL-S 描述的 Web 服务,主要从服务功能进行描述,缺乏对服务质量的明确刻画,导致服务请求者难以得到最佳服务.针对此问题,本文对 OWL-S 规范扩展,提出一个基于 QoS 的语义 Web 服务发现模型 SWSDM-QoS.该模型支持 QoS 的服务发布、查找和反馈,并在此基础上提出一种 QoS 约束的服务选择算法.本论文创新性之处在于在服务发现过程中加入了服务质量本体信息,使得Web服务在框架内支持QoS约束的语义发现,满足了用户的个性化需求.     

基于层次智能切割的IP分类算法

首先对规则数据库存在冲突的情况进行处理,生成了无冲突的规则数据库。然后基于层次智能切割算法,同时构造无冲突的哈希函数,提出了一种新的IP分类算法,即基于层次智能切割和无冲突哈希算法( HICNCH),并且提出利用最小二乘法改进了层次智能切割树分割域的判定准则,使分割更加高效。通过与经典算法的各项指标的比较,论证了该算法具有较小时间复杂度和空间复杂度,综合性能有了较大提高。     

A multi-category inter-purchase time model based on hierarchical Bayesian theory

Because of recent diversity in consumer demands and the decrease in popularity of mass media, one-to-one database marketing is being increasingly used by companies to increase their competitiveness. Many studies have addressed the issue of inter-purchase time, but few have considered the impact of multiple categories of products on inter-purchase time, which may vary for different products. The aim of the present study was to build a one-to-one multi-category inter-purchase time model using a hierarchical Bayesian model based on a generalized gamma distribution and multiplicative model formulations. Using a hazard rate function, the model was applied to derive a purchase probability for individual customers. To validate the proposed model, field data were collected from a local catalog company and prediction hit rates were compared for different models. The multi-category inter-purchase time model exhibited better prediction hit rates than a basic model. Using the multiplicative model, our multi-category model can estimate the influence of product category on customers’ inter-purchase time.     

基于AHP的高校网站模糊综合评价

本文建立了我国高校网站评价的指标体系,并用指数标度的层次分析法确定评价指标的权重,并结合实际用模糊综合评价法研究了高校网站的评价问题.     

A taxonomy of web prediction algorithms

Web prefetching techniques are an attractive solution to reduce the user-perceived latency. These techniques are driven by a prediction engine or algorithm that guesses following actions of web users. A large amount of prediction algorithms has been proposed since the first prefetching approach was published, although it is only over the last two or three years when they have begun to be successfully implemented in commercial products. These algorithms can be implemented in any element of the web architecture and can use a wide variety of information as input. This affects their structure, data system, computational resources and accuracy. The knowledge of the input information and the understanding of how it can be handled to make predictions can help to improve the design of current prediction engines, and consequently prefetching techniques.This paper analyzes fifty of the most relevant algorithms proposed along 15 years of prefetching research and proposes a taxonomy where the algorithms are classified according to the input data they use. For each group, the main advantages and shortcomings are highlighted.