HDP采样消息传递算法

分层狄利克雷过程是一种贝叶斯无参模型,用以分析海量数据的概率主题模型解决潜在狄利克雷分布无法解决的动态聚类的问题。本文从因子图的角度出发将消息传递算法与吉布斯采样算法结合用以解决贝叶斯无参模型后验概率推断问题,最终将该算法与LDA算法以及HDP算法在混淆度方面进行对比。实验结果表明该算法相比HDP采样算法收敛较快,最终也能收敛到LDA模型最优主题数目下的混淆度。     

朴素并行LDA

并行潜在狄利克雷分配(LDA)主题模型在计算与通信两方面的时间消耗较大,导致训练模型的时间过长,因而无法被广泛应用.提出朴素并行LDA算法,针对计算和通信分别提出改进方法.一方面通过加入单词影响因子以及设置阈值的方法来降低文本训练的粒度,另一方面通过降低通信频率来减少通信时间.实验结果表明,优化后的并行LDA在保证精度损失为1％的前提下,将训练速度提高了36％,有效提高了并行的加速比.     

一种结合主题模型的推荐算法

针对传统协同过滤推荐算法存在的冷启动、数据稀疏以及相似度度量的准确性问题,基于LDA主题模型对文本隐式主题挖掘的有效性和KL散度在主题分布相似性度量的准确性,提出了结合LDA主题模型的矩阵分解推荐算法。首先,利用改进的LDA算法输出项目-主题分布,并用困惑度作为主题数设置的修正函数;然后分别基于余弦相似度和KL散度计算得到项目相似度矩阵,将得到的相似度矩阵结合原评分训练集输出预评分,再将预评分填充到训练集;最后将训练集输入ALS矩阵分解算法得到推荐结果。通过MovieLens数据集的实验结果表明,该算法在不同隐式参数设定下均能得到比ALS推荐算法以及更小的预测误差,并且最优预测误差小于传统推荐算法。该实验说明了通过集成LDA主题模型的ALS算法效果要优于其他推荐算法。     

基于LDA主题相似度的SVM迁移学习

针对文本分类领域中的迁移学习方法,提出了一种基于LDA(Latent Dirichlet Allocation)主题生成模型相似度的支持向量机(SVM)迁移学习新思路。基于此思想,提出了迁移学习算法LDA-TSVM。本算法通过对目标域的主题进行分类,依据主题分类信息熵对训练数据进行筛选,分别计算每个训练样本的权重,使得训练集与目标集有很高的相似度,从而达到迁移学习的目的。本算法不仅未引入辅助集,而且还考虑了样本本身的差异,有效地提高了源域数据集与目标域数据集的相似性。实验结果表明了新迁移算法的有效性。     

一种基于LDA的在线主题演化挖掘模型

基于文本内容的隐含语义分析建立在线主题演化计算模型,通过追踪不同时间片内主题的变化趋势进行主题演化分析。将Latent Dirichlet Allocation(LDA)模型扩展到在线文本流,建立并实现了在线LDA模型;利用前一时间片的后验概率影响当前时间片的先验概率来维持主题间的连续性;根据改进的增量Gibbs算法进行推理,获取主题一词和文档一主题的概率分布,利用KullbackLeibler(KL)相对嫡来衡量主题之间的相似度,从而发现主题演化中的“主题遗传”和“主题变异”。实验结果表明,该模型能从互联网语料中找出主题的演化趋势,具有良好的效果。     

基于滑动窗口的主题模型

LDA(Latent Dirichlet Allocation)是一个分层的概率主题模型,目前被广泛地应用于文本挖掘。这种模型既不考虑文档与文档之间的顺序关系,也不考虑同一篇文档中词与词之间的顺序关系,简化了问题的复杂性,同时也为模型的改进提供了契机。针对此问题提出了基于滑动窗口的主题模型,该模型的基本思想是文档中的一个单词的主题与其附近若干单词的主题关系越紧密,受附近单词主题的影响越大。根据窗口和滑动位移的大小,把文档切割为粒度更小的片段。同时,针对大数据集和数据流问题,提出了在线滑动窗口主题模型。在4个数据集上的实验表明,基于滑动窗口的主题模型训练出来的模型在数据集上有更好的泛化性能和精度。     

AdBagging:自适应抽样参数在线装袋算法

通过对数据流分类中的概念漂移问题的研究,提出了一种在线装袋(Online Bagging)算法的改进算法——自适应抽样参数的在线装袋算法AdBagging(adaptive lambda bagging)。利用在分类过程中出现的误分样本数量来调整Online Bagging算法中的泊松(Poisson)分布的抽样参数,从而可以动态调整新样本在学习器中的权重,即对于数据流中的误分类样本给予较高的学习权重因子,而对于正确分类的样本给予较低的学习权重因子,同时结合样本出现的时间顺序调整权重因子,使得集成分类器可以动态调整其多样性(adversity)。该算法具有OnlineBagging算法的高效简洁优点,并能解决数据流中具有概念漂移的问题,人工数据集和实际数据集上的实验结果表明了该算法的有效性。     

基于用户兴趣模型的推荐算法

传统的协同过滤算法虽然可以很容易地挖掘出用户的兴趣爱好,但存在数据冷启动和稀疏性问题.针对这些问题,提出一种基于用户兴趣模型的推荐算法.首先通过LDA主题模型训练数据集得到物品-主题概率分布矩阵,利用物品-主题概率分布矩阵得到用户历史兴趣模型,然后结合用户历史行为信息和物品内容信息得到用户兴趣模型,最后计算用户与候选集之间的相似度,进行TOP-N推荐.在豆瓣电影数据集上的实验结果表明,改进后的推荐算法能够更好地处理稀疏数据和冷启动问题,并且明显提高了推荐质量.     

一种基于增量加权平均的在线序贯极限学习机算法

针对在线序贯极限学习机(OS-ELM)对增量数据学习效率低、准确性差的问题, 提出一种基于增量加权平均的在线序贯极限学习机(WOS-ELM)算法.将算法的原始数据训练模型残差与增量数据训练模型残差进行加权作为代价函数,推导出用于均衡原始数据与增量数据的训练模型,利用原始数据来弱化增量数据的波动,使在线极限学习机具有较好的稳定性,从而提高算法的学习效率和准确性. 仿真实验结果表明, 所提出的WOS-ELM算法对增量数据具有较好的预测精度和泛化能力.     

SVM增量算法及在旅游信息分类中的应用

在如何从海量的数据中提取有用的信息上提出了一种新的SVM的增量学习算法.该算法基于KKT条件,通过研究支持向量分布特点,分析了新样本加入训练集后,支持向量集的变化情况,提出等势训练集的观点.能对训练数据进行有效的遗忘淘汰,使得学习对象的知识得到了积累.在理论分析和对旅游信息分类的应用结果表明,该算法能在保持分类精度的同时,有效得提高训练速度.     

How to Archive an Event: Reflections on the Social Repository of Ireland

Reflecting on a feasibility study into archiving social media, this article traces how “events” are defined in various domains and contexts, and employs case studies to analyze key relationships between hashtags and events to provide a critical analysis of how archival events can be constructed out of social events. It provides an overview of the archival and curatorial considerations involved in defining and preserving a social media event, and outlines the technologies developed for the process of collecting, annotating, and preserving social media events. Overall, the article endeavors to reveal how pragmatic considerations, computational approaches and curatorial perspectives shape digital archives and historical narratives.     

BBS中信息传播模式的特征分析

通过比较传染病传播机制与信息传播机制,提出BBS中的信息传播机制模型。通过对BBS中帖子数量变化规律的建模,分析了BBS中信息传播模式的特征,并使用实际数据说明BBS中的信息传播模式。实验表明：BBS可以吸引大批的用户参与,但用户只对部分话题感兴趣并参与讨论;绝大多数话题（占94.9%）帖子数的增长率先增加再减小直至为0,而少量话题（占5.1%）帖子数的增长率直接减小至0。这些结论有助于认识BBS的信息传播机制,对控制和管理BBS的信息传播有启发意义。     

Community building and knowledge sharing by individuals with disabilities using social media

The use of social media to share information, enhance learning, and connect with an online community has grown rapidly over the past 10 years. As social media becomes a more common tool in both formal and informal education, it is imperative to understand how it is used by individuals with disabilities. Through a systematic study of the literature, 215 articles on social media used by individuals with disabilities were selected and 29 selected for in-depth thematic analysis. Six major themes were identified: community, cyberbullying, self-esteem, self-determination, access to technology, and accessibility. To confirm these six categories, we expanded our search, yielding an additional 30 articles, for a total 59 articles reviewed in-depth. Interactions between individuals with disabilities within online communities often had the goal of acquiring knowledge or learning new information. A communities of practice theoretical framework is used to discuss interactions among the elements of social media design, learning, and the building of community by individuals with disabilities.     

基于社会认证的网络身份模型

网络实名制的提出,是为了解决网络匿名性所带来的问题,却又面临实名信息泄露的诟病。造成信息泄露的根源在于实名认证依赖于实名信息。基于社会认证的网络身份模型,依赖社会关系进行身份认证,其利用OSN节点的社会关系构建网络身份,在发挥网络监管作用的同时,避免实名信息的泄漏。模型首先在OSN中依据一定策略选择根节点;然后,采用担保方式进行社会认证;最后,在不依赖实名信息的基础上,构建节点的唯一网络身份SANI。SNAI身份含节点的社会认证信息,具有身份认证和行为溯源的功能。     

Cultivating Social Resources on Social Network Sites: Facebook Relationship Maintenance Behaviors and Their Role in Social Capital Processes

This study explores the relationship between perceived bridging social capital and specific Facebook‐enabled communication behaviors using survey data from a sample of U.S. adults (N=614). We explore the role of a specific set of Facebook behaviors that support relationship maintenance and assess the extent to which demographic variables, time on site, total and “actual” Facebook Friends, and this new measure (Facebook Relationship Maintenance Behaviors) predict bridging social capital. Drawing upon scholarship on social capital and relationship maintenance, we discuss the role of social grooming and attention‐signaling activities in shaping perceived access to resources in one's network as measured by bridging social capital.     

An Efficient Two-Phase Model for Computing Influential Nodes in Social Networks Using Social Actions

在社会网络的影响的测量在数据采矿社区收到了很多注意。影响最大化指发现尽量利用信息或产品采纳的有影响的用户的过程。在真实设置,在一个社会网络的一个用户的影响能被行动的集合建模(例如,份额,重新鸣叫,注释) 在其出版物以后由网络的另外的用户表现了。就我们的知识而言,在文学的所有建议模型同等地对待这些行动。然而,它是明显的一工具少些比一样的出版的份额影响的一份出版物相似。这建议每个行动有它影响的自己的水平(或重要性) 。在这份报纸,我们建议一个模型(叫的社会基于行动的影响最大化模型, SAIM ) 为在社会网络的影响最大化。在 SAIM,行动没在测量一个个人的影响力量同等地被考虑,并且它由二主要的步组成。在第一步,我们在社会网络计算每个个人的影响力量。这影响力量用 PageRank 从用户行动被计算。在这步的结束,我们得到每个节点被它的影响力量在标记的一个加权的社会网络。在 SAIM 的第二步,我们计算一个新概念说出 influence-BFS 树的使用的有影响的节点的一个最佳的集合。在大规模真实世界、合成的社会网络上进行的实验在计算揭示我们的模型 SAIM 的好表演,在可接受的时间规模,允许信息的最大的传播的有影响的节点的一个最小的集合。     

Perceived social support moderates the relationships between variables in the social cognition model

Posting behaviour on social networking sites (SNS) has become a method enabling unsatisfied users to vent emotions. Based on social cognition theory (SCT), personal outcome expectations and self-efficacy affect posting behaviour for venting emotions on SNS. However, perceived social support (PSS) may alter the relationships within the SCT model. Thus, this study aimed to explore the moderating effect of PSS on the relationships between variables in the SCT model for venting emotions on SNS. In total, 310 unsatisfied customers in Taiwan were investigated, and structural equation modelling was performed to test the hypotheses. The results indicated that personal outcome expectations and self-efficacy were positively associated with posting behaviour which, in turn, increased venting emotions on SNS. Moreover, PSS moderated the relationships between variables in the SCT model.     

Specifying the dynamics of social technologies as social microworlds

This paper explores the affordances of social technologies for supporting the construction of a shareable artefact by a group of learners. A qualitative study that captures the use of five different types of social technologies (Facebook, blogs, wikis, Google Documents and Dropbox) in three different classroom settings sheds light on the potentials and challenges of these tools for supporting material exploration, artefact construction and evaluation. Qualitative content analysis of instructors’ field notes, students’ and instructors’ reflections, interviews and focus groups sheds light on the potential of social technologies to transform the activity of learning across a new culture of computational tools. The affordances of social technologies are discussed as well as design principles that need to be followed in these new arenas.     

Topology-Based Recommendation of Users in Micro-Blogging Communities

Nowadays,more and more users share real-time news and information in micro-blogging communities such as Twitter,Tumblr or Plurk.In these sites,information is shared via a followers/followees social network structure in which a follower will receive all the micro-blogs from the users he/she follows,named followees.With the increasing number of registered users in this kind of sites,finding relevant and reliable sources of information becomes essential.The reduced number of characters present in micro-posts along with the informal language commonly used in these sites make it difficult to apply standard content-based approaches to the problem of user recommendation.To address this problem,we propose an algorithm for recommending relevant users that explores the topology of the network considering different factors that allow us to identify users that can be considered good information sources.Experimental evaluation conducted with a group of users is reported,demonstrating the potential of the approach.     

Behavior‐Based Propagation of Trust in Social Networks with Restricted and Anonymous Participation

Increasing interactions and engagements in social networks through monetary and material incentives is not always feasible. Some social networks, specifically those that are built on the basis of fairness, cannot incentivize members using tangible things and thus require an intangible way to do so. In such networks, a personalized recommender could provide an incentive for members to interact with other members in the community. Behavior‐based trust models that generally compute social trust values using the interactions of a member with other members in the community have proven to be good for this. These models, however, largely ignore the interactions of those members with whom a member has interacted, referred to as “friendship effects.” Results from social studies and behavioral science show that friends have a significant influence on the behavior of the members in the community. Following the famous Spanish proverb on friendship “Tell Me Your Friends and I Will Tell You Who You Are,” we extend our behavior‐based trust model by incorporating the “friendship effect” with the aim of improving the accuracy of the recommender system. In this article, we describe a trust propagation model based on associations that combines the behavior of both individual members and their friends. The propagation of trust in our model depends on three key factors: the density of interactions, the degree of separation, and the decay of friendship effect. We evaluate our model using a real data set and make observations on what happens in a social network with and without trust propagation to understand the expected impact of trust propagation on the ranking of the members in the recommended list. We present the model and the results of its evaluation. This work is in the context of moderated networks for which participation is by invitation only and in which members are anonymous and do not know each other outside the community. Copyright © 2014 John Wiley & Sons, Ltd.