Error and attack tolerance of small-worldness in complex networks

Complex networks may undergo random and/or systematic failures in some of their components, i.e. nodes and edges. These failures may influence various network properties. In this article, for a number of real-world as well as Watts–Strogatz model networks, we investigated the profile of the network small-worldness as random failures, i.e. errors, or systematic failures, i.e. attacks, occurred in the nodes. In errors nodes are randomly removed along with all their tipping edges, while in attacks the nodes with highest degrees are removed from the network. Interestingly, in many cases, the small-worldness of violated networks increased as more nodes underwent an attack. This indicates an important role of the hub nodes in controlling the small-worldness of Watts–Strogatz networks. The profile of changes in the small-worldness as a result of errors/attacks was independent of network size, while it was influenced by average degree and rewiring probability of Watts–Strogatz model. We also found that the pattern of the changes of the small-worldness in real-world networks is completely different than that of the Watts–Strogatz networks. Therefore, although Watts–Strogatz model is often used for constructing networks with small-world property, the resulting networks have different properties compared to real-world ones in terms of robustness in the small-worldness index against errors/attacks.     

Predicting future influence of papers,researchers, and venues in a dynamic academic network

Performance evaluation and prediction of academic achievements is an essential task for scientists, research organizations, research funding bodies, and government agencies alike. Recently, heterogeneous networks have been used to evaluate or predict performance of multi-entities including papers, researchers, and venues with some success. However, only a minimum of effort has been made to predict the future influence of papers, researchers and venues. In this paper, we propose a new framework WMR-Rank for this purpose. Based on the dynamic and heterogeneous network of multiple entities, we extract seven types of relations among them. The framework supports useful features including the refined granularity of relevant entities such as authors and venues, time awareness for published papers and their citations, differentiating the contribution of multiple coauthors to the same paper, amongst others. By leveraging all seven types of relations and fusing the rich information in a mutually reinforcing style, we are able to predict future influence of papers, authors and venues more precisely. Using the ACL dataset, our experimental results demonstrate that the proposed approach considerably outperforms state-of-the art competitors.     

基于混合图的在线社交网络朋友推荐算法

针对在线社交网络朋友推荐问题,尝试融合多个社会网络为一个混合图模型,采用基于混合图模型的重启动随机游走算法,为用户提供个性化的朋友推荐,并通过参数调节多个网络的权重。实验表明,该算法提高了在线社交网络朋友推荐的准确性。     

A regression analysis of researchers’ social network metrics on their citation performance in a college of engineering

Previous research shows that researchers’ social network metrics obtained from a collaborative output network (e.g., joint publications or co-authorship network) impact their performance determined by g-index. We use a richer dataset to show that a scholar's performance should be considered with respect to position in multiple networks. Previous research using only the network of researchers’ joint publications shows that a researcher's distinct connections to other researchers, a researcher's number of repeated collaborative outputs, and a researchers’ redundant connections to a group of researchers who are themselves well-connected has a positive impact on the researchers’ performance, while a researcher's tendency to connect with other researchers who are themselves well-connected (i.e., eigenvector centrality) had a negative impact on the researchers’ performance. Our findings are similar except that we find that eigenvector centrality has a positive impact on the performance of scholars. Moreover, our results demonstrate that a researcher's tendency toward dense local neighborhoods and the researchers’ demographic attributes such as gender should also be considered when investigating the impact of the social network metrics on the performance of researchers.     

基于异构信息网络嵌入的专利技术主体间交易推荐模型

考虑专利技术主体间技术邻近、地理邻近、共申请关系、引证关系、经济圈效应、主体类型邻近、主体间从属关系这7种因素对交易的影响,构建由4类节点、10类关系组成的异构信息网络,设计基于元路径与元结构的异构关系遍历算法获取主体间关系序列。以关系序列为语料,构建基于网络嵌入的异构信息网络主体间交易推荐模型(PSR-vec),采用基于Huffman树的Skip-Gram方法进行网络嵌入训练,计算主体向量间相似度以实现交易推荐。通过2012-2018年电子信息领域专利数据的实证研究得出:第一,PSR-vec模型相比DeepWalk、node2vec与PathSim等方法,推荐精度大幅提高,达到82.4%;第二,融合多个元路径与元结构特征的推荐与单一特征相比,推荐精度大幅提高;第三,基于ρ2以及改进的元结构S4、S6、S8、S10、S12、S14的推荐结果均高于基于ρ1以及改进的元结构S3、S5、S7、S9、S11、S13的推荐精度,说明基于主体间转让技术邻近性的推荐精度更高;第四,在技术邻近元路径基础上分别融合主体间共申请、引证、从属、经济圈效应这4类邻近关系获得元结构并进行推荐,推荐精度均显著提高,而融合地理邻近、类型邻近2类关系后推荐精度有所降低,说明地理邻近、类型邻近对交易的促进作用不明显;第五,基于PSR-vec模型的推荐结果包括具有控股和供应等紧密关系的主体,也包括关系松散的主体,推荐结果具有多样性。本研究为专利技术主体间的有效对接提供了决策方法。     

基于二模复杂网络的隐性知识发现方法研究——以潜在药物靶点挖掘为例

[目的/意义] 通过构建二模复杂网络模型，揭示隐藏在海量文献中的隐性知识。[方法/过程] 通过NetworkX复杂网络工具包，依据任意两个节点的共现关系构建二模复杂网络模型；对网络模型中节点的共现关系进行加权，计算网络的拓扑信息并进行AP聚类，提取节点间的直接关系；采用AUC方法对AA、JC、加权改进的wAA和wJC等4种链路预测算法进行评价，遴选出最合适的预测算法，并对复杂网络的隐性关系进行预测分析。[结果/结论] 以潜在药物靶点挖掘为例进行的实证研究结果表明，wAA链路预测算法为最优的链路预测算法；二模复杂网络模型、指标和方法体系在美国化学文摘社数据库中的药物靶点挖掘中具有一定的有效性。下一步计划在其他数据库中或其他研究领域中进行尝试，以进一步验证该模型的通用性和有效性。     

Predicting scientific research trends based on link prediction in keyword networks

The rapid development of scientific fields in this modern era has raised the concern for prospective scholars to find a proper research field to conduct their future studies. Thus, having a vision of future could be helpful to pick the right path for doing research and ensuring that it is worth investing in. In this study, we use article keywords of computer science journals and conferences, assigned by INSPEC controlled indexing, to construct a temporal scientific knowledge network. By observing keyword networks snapshots over time, we can utilize the link prediction methods to foresee the future structures of these networks. We use two different approaches for this link prediction problem. First, we have utilized three topology-based link prediction algorithms, two of which are commonly used in literature. We have also proposed a third algorithm based on nodes (keywords) clustering coefficient, their centrality measures like eigenvector centrality, and nodes community information. Then, we used nodes topological features and the outputs of aforementioned topology-based link prediction algorithms as features to feed five machine learning link prediction algorithms (SVM, Random Forest Classifier, K-Nearest Neighbors, Gaussian Naïve Bayes, and Multinomial Naïve Bayes). All tested predictors have shown considerable performance and their results are discussed in this paper.     

Effect of class imbalance in heterogeneous network embedding: An empirical study

Network science has been extensively explored in solving various bibliometrics tasks such as Co-authorship prediction, Author classification, Author clustering, Author ranking, Paper ranking, etc. While majority of the past studies exploit homogeneous bibliographic network (consists of singular type of nodes and edges), in recent past there is a surge in using heterogeneous bibliographic entities and their inter-dependencies using heterogeneous information networks (HIN). Unlike homogeneous bibliographic networks, a bibliographic HIN consists of multi-typed nodes such as Author, Paper, Venue, etc. and corresponding relations. Thus bibliographic HIN is more complex and captures rich semantics of underlying bibliographic data as well as poses more challenges. Since a real-world HIN may have different number of instances for different node types, class imbalance is ubiquitous. Recent studies discuss class imbalance in brief and exploit meta-path-based strategies to address the issue. However, there is no work which quantitatively study the effect of class imbalance in regards to solving real-world bibliometrics tasks. Therefore, this paper first proposes a metric to estimate class imbalance in HIN and study the effects of class imbalance over two bibliometrics tasks, namely (i) Co-authorship prediction and (ii) Author's research area classification, using node features generated by network embedding-based frameworks for DBLP dataset. From various experimental analysis, it is evident that class imbalance in bibliographic HIN is an inherent characteristic and for better performance of the above-mentioned bibliometrics tasks, the bibliographic HINs must consider Author, Paper, and Venue as node types.     

An improved SAO network-based method for technology trend analysis: A case study of graphene

This paper proposes an improved Subject-Action-Object (SAO) network-based method for analyzing trends in technology development. It attempts to address shortcomings of the traditional SAO network approach, i.e., when setting Subject, Action and Object as nodes of the network, there may be errors in explaining the relationship between Subject Node and Object Node, and the strength of the relationship between subject and object also cannot be identified. The proposed improved SAO network-based method in this paper includes: (1) a new method for constructing an SAO network based on SAO links that calculate the intensity of the relationship between nodes; (2) a model for identifying technology development trends based on structural holes, changes in the distribution of node degrees, and shifts in network centrality. An empirical study on graphene technology is used to illustrate the validity and feasibility of the proposed method.     

融合多源数据的科研人员画像构建方法研究

[目的/意义]大数据时代需要将"人"数据化,科研人员也需要数据化。科研人员画像的建立,对于科研管理层全面了解科研人员的信息、客观评价其研究水平等有重要作用,可以作为分析科研人员研究行为或专家推荐的基础,提高科研管理效率。[方法/过程]首先提出科研人员画像的概念,认为其是描述科研人员信息的标签的集合。其次,以个人主页、知网、基金网等多个异构数据源的数据为基础,提出融合多源数据的科研人员画像构建方法,分别从科研人员的基础属性、科研偏好和科研关系三方面形式化描述了科研人员信息,并提取各个维度的标签,以可视化的方式展示其画像。最后,分别以国内外两位科研人员为例,说明了科研人员画像构建方法的可行性。[结果/结论]科研人员画像的构建适用于国内外的科研人员,能够全面描述科研人员信息并直观展示出来。     

Social network analysis: An approach and technique for the study of information exchange

Social network analysis is an approach and set of techniques used to study the exchange of resources among actors (i.e., individuals, groups, or organizations). One such resource is information. Regular patterns of information exchange reveal themselves as social networks, with actors as nodes in the network and information exchange relationships as connectors between nodes. Just as roads structure the flow of resources among cities, information exchange relationships structure the flow of information among actors. Social network analysis assesses information opportunities for individuals or groups of individuals in terms of exposure to and control of information. By gaining awareness of existing information exchange routes, information providers can act on information opportunities and make changes to information routes to improve the delivery of information services.     

Quantifying the online long-term interest in research

Research articles are being shared in increasing numbers on multiple online platforms. Although the scholarly impact of these articles has been widely studied, the online interest determined by how long the research articles are shared online remains unclear. Being cognizant of how long a research article is mentioned online could be valuable information to the researchers. In this paper, we analyzed multiple social media platforms on which users share and/or discuss scholarly articles. We built three clusters for papers, based on the number of yearly online mentions having publication dates ranging from the year 1920 to 2016. Using the online social media metrics for each of these three clusters, we built machine learning models to predict the long-term online interest in research articles. We addressed the prediction task with two different approaches: regression and classification. For the regression approach, the Multi-Layer Perceptron model performed best, and for the classification approach, the tree-based models performed better than other models. We found that old articles are most evident in the contexts of economics and industry (i.e., patents). In contrast, recently published articles are most evident in research platforms (i.e., Mendeley) followed by social media platforms (i.e., Twitter).     

Interpersonal problem solving: Problem conceptualization and communication use

To date, researchers have not studied the relationship between the mental representation of interpersonal problem situations and interpersonal communication used to manage problem situations. This report examined the relationship between four types of communication (i.e., integrative, distributive, indirect, and avoidance) and fourteen perceptions of problem situations (i.e., problem uniqueness and frequency; goal importance, complexity, and mutuality; uncertainty about the other, relationship, and goal‐path; causal attribution to the self, other, intent of the other, relationship, and environment; and feelings for the other). Two studies tested hypotheses. The first study used self‐reports based on personally experienced problem situations to examine the relationship between the perceptions and perceived use of the types of communication. The second study explored the relationship between communication stimuli likely to be used in problem situations and the likelihood of the conceptualization perceptions. Results generally confirmed the hypotheses in both studies and indicated that a relatively specific relationship exists between each type of communication and the set of perceptions for both personally experienced and undefined problem situations. Implications of these results for problem‐related knowledge structures, problem solving, conflict, relationship intervention, and interpersonal influence were discussed along with directions for future research.     

Toward collaborator selection and determination of data ownership and publication authorship in research collaborations

This study examined factors that might affect researchers' willingness to collaborate with a specific researcher and the priorities given to those factors. In addition, it investigated how researchers determined the ownership of collaborative project data and how they determined the order of authorship on collaborative publications in condensed matter physics. In general, researchers rated their intrinsic motivations the highest, such as the quality of ideas a potential collaborator might have and their satisfaction with a past collaboration, followed by their extrinsic motivations, such as the complementary knowledge, skills, or resources the collaborator could provide. In addition, researchers who had a greater number of collaborative projects and researchers who had served as a project PI or co-PI valued the deep-level, personality-related characteristics of a collaborator higher than did those who had not. Younger researchers were more risk averse and more concerned with a collaborator's reputation and the possible cost of a collaboration decision. Additionally, younger researchers indicated more often than older researchers that they did not know whether their project teams followed any rules or norms or engaged in negotiation to determine the order of authorship on collaborative publications.     

Predicting the citation counts of individual papers via a BP neural network

Predicting the citation counts of academic papers is of considerable significance to scientific evaluation. This study used a four-layer Back Propagation (BP) neural network model to predict the five-year citations of 49,834 papers in the library, information and documentation field indexed by the CSSCI database and published from 2000 to 2013. We extracted six paper features, two journal features, nine author features, eight reference features, and five early citation features to make the prediction. The empirical experiments showed that the performance of the BP neural network is significantly better than those of the six baseline models. In terms of the prediction effect, the accuracy of the model at predicting infrequently cited papers was higher than that for frequently cited ones. We determined that five essential features have significant effects on the prediction performance of the model, i.e., ‘citations in the first two years’, ‘first-cited age’, ‘paper length’, ‘month of publication’, and ‘self-citations of journals’, and the other features contribute only slightly to the prediction.     

Modeling users’ search sessions for high utility query recommendation

Query recommendation has long been considered a key feature of search engines, which can improve users’ search experience by providing useful query suggestions for their search tasks. Most existing approaches on query recommendation aim to recommend relevant queries, i.e., alternative queries similar to a user’s initial query. However, the ultimate goal of query recommendation is to assist users to reformulate queries so that they can accomplish their search task successfully and quickly. Only considering relevance in query recommendation is apparently not directly toward this goal. In this paper, we argue that it is more important to directly recommend queries with high utility, i.e., queries that can better satisfy users’ information needs. For this purpose, we attempt to infer query utility from users’ sequential search behaviors recorded in their search sessions. Specifically, we propose a dynamic Bayesian network, referred as Query Utility Model (QUM), to capture query utility by simultaneously modeling users’ reformulation and click behaviors. We then recommend queries with high utility to help users better accomplish their search tasks. We empirically evaluated the performance of our approach on a publicly released query log by comparing with the state-of-the-art methods. The experimental results show that, by recommending high utility queries, our approach is far more effective in helping users find relevant search results and thus satisfying their information needs.     

基于CNN-BiLSTM模型的在线医疗实体抽取研究

[目的/意义]在线医疗信息抽取是实现医疗信息检索、医疗信息推荐、个人医疗健康提醒及警示、疾病诊断、公众健康监控、药物不良反应挖掘等服务的基础环节，而医疗实体抽取则是在线医疗信息抽取的首要工作。本文拟解决传统医疗实体抽取严重依赖于人工特征提取且效率低的问题。[方法/过程]以网络文本为研究对象，首先对医疗实体类型和医疗实体抽取的目标进行描述。将在线医疗文本中的医疗实体抽取任务看作序列标注问题来解决，通过对CNN模型和BiLSTM模型基础理论的探讨，构建基于混合深度学习模型CNN-BiLSTM的医疗实体抽取框架。[结果/结论]通过三组对比实验，验证了本文所使用的CNN-BiLSTM模型在医疗实体抽取任务中的有效性。     

异质网络资源多维度推荐模式研究——以豆瓣网为例

[目的/意义]针对某些包含多级用户和多级资源的异质网络,在总结其各种异质模态的基础上提出一种多维度推荐框架MDRM,向目标用户进行好友和资源的推荐。[方法/过程]通过对用户、资源划分等级,在分析各种网络模态的基础上建立其异质关系;利用情感倾向分析得到二级用户——二级资源评分矩阵,借助协同过滤算法,实现同级用户和二级资源的推荐;基于异质关系,实现一级用户和一级资源的推荐,最终实现多维度推荐。[结果/结论]在以豆瓣网数据作为数据集的实验中取得了较好的效果,说明MDRM模型适合某些异质网络资源的推荐。     

基于表示学习的双层知识网络链路预测

当前,针对知识网络的链路预测主要是基于网络拓扑结构的相似性,很少考虑作者的研究领域,导致信息利用不充分等问题,因此本文提出了双层知识网络的链路预测框架hypernet2vec。双层知识网络,即作者合著关系网络和学术领域关系网络,利用网络表示学习,分别将两层网络中的节点映射到低维的向量空间,再输入到专门设计的卷积神经网络中计算并进行链路预测。与经典的链路预测指标如RA指标、LP指标和LRW指标等相比,hypernet2vec模型预测的AUC(area under curve)值取得了显著的提升,平均提升幅度达11.17%。文章还从情报产生层面和复杂系统层面,对模型发生作用的深层机理进行了探讨。     

Detecting and assessing statistical interactions with multiple regression

This paper focuses on the treatment of statistical interactions (i.e., non‐additivity) within the framework of multiple regression. The nature of statistical interactions in relation to other types of effects and the implications of interactions for analysis and interpretation are discussed. We argue that researchers using regression analyses (1) often either fail to test for interactions (2) or use inadequate methods for testing for interactions (3) and, consequently may make faulty conclusions about their data. We outline several methods of dealing with interactions with regression and discuss the strengths and weaknesses of each. Recommendations are made for the detection and treatment of interactions within regression.