of 13

International Journal of Wireless Network Security(IJWNS APPLICATION OF ACADEMIC

International Journal of Wireless Network Security(IJWNS APPLICATION OF ACADEMIC
23 views13 pages
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
  International Jou  A  PPLICATI  A  CAD K. Sobha Ra 1 Dept. of Information Tec 2 Anil Neerukonda 3 Department of CSSE, Col  ABSTRACT Social network is a group of indivi large scale and distributed due Quantitative analysis of networks is and in turn the society. Clustering dense social networks. We have co When a social network is represent mining would be suitable for statis nodesto simplify network analysi  similaritybetween unstructured dat  social network.  K   EYWORDS  : Social Network Analysis, Clusterin 41.   I NTRODUCTION   Different kinds of network exist attributes like being distribute interesting quantifiable measur connectivity, centrality, clusteri modeled as random graphs, scal being scale free, the social net relation amongst the nodes vary  probability of a node being infl attributes and interactions. The database; hence different appr network data. 1.1   A CADEMIC S OCIAL N ET  With the growth of the intern influential. One of the most effe of collaboration of colleagues employers’ recruiting and in wo nal of Wireless Network Security(IJWNS)Vol1, No1 N OF C LUSTERING TO  A  N MIC S OCIAL N ETWORKS   ni 1 , KVSVN Raju 2  and V.Valli Kumari 3 nology, MVGR College of Engineering, Vizia Inst. of Technology & Sciences, Visakhapatna   lege of Engineering, Andhra University, Visak   uals with diverse social interactions amongst them. Th to involvement of more people from different parts need of the hour due to its’rippling influence on the net helps us to group people with similar characteristics nsidered similarity measures for statistical analysis of d as a graph with members as nodes and their relation ical analysis. We have chosen academic social network  s. The ontology of research interests is considere a elements extracted from rofile pages of members o    , Graph Mining, RDF    viz. social, technological, business etc., all of which , continuously growing and of large scale. Th es that help analyze these networks, like numb g coefficient and degree distribution[14]. These ne -free networks and hierarchical networks. Due to t orks continuously expand with addition of new with time and frequency of interaction between t uential in the network is not uniform due to their ocial network data is not even suitable to be store aches are followed to store and analyze unstru ORKS   et and the World Wide Web, social networks ctive channels for obtaining information is the info and friends etc. The use of social networks is rkers’ job-seeking, pursuit of hobbies and building 9 LYZE agaram. . apatnam. network is of of the globe. work dynamics to analyze the ocial network.  s edges, graph and clustered d to measure an academic share similar re are some er of nodes, works can be e property of odes and the em. Also the difference in in relational ctured social ave become rmal network idespread in collaboration  International Jou within or between organization  profile with fields like homepa address, email, telephone and heterogeneous and distributed RDF [26] to represent propertie wide range of services provided content providers, which has lea Visual representation of social identification of its members a  between individuals in social n organizations, web sites or cit relative importance of nodes an widely used in disciplines like eigenvectors of the adjacency m Academic social networks provi share knowledge with their pe  profiles. These networks also contributor to the networ Arnetminer(www.arnetminer.or for researcher in Research(academic.research.micMicrosoft Research Asia[23]. conferences, journals and author Figure1. Snapsho 1.2   G RAPH R  EPRESENTATIO As stated by Han[18], informati data can be represented as a gra network and Edges(E) are links the relationship or interaction a w.r.t. traditional data mining structured. Our approach of clu selected node in a graph, also te of A, there is a probability of B method which helps in cluste  parameters are identified by   nal of Wireless Network Security(IJWNS)Vol1, No1 s. A person can have different types of informat ge, field of interest and hobbies, contact informat fax number. However, the information is usua eb pages. The web pages follow a standard frame s of members. With the development of social net  by them, members have transformed from content to vast amounts of data to be stored and processed. networks is important to understand the network nd their attributes. Social network analysis maps tworks. Such individuals are often persons, but m ations between scholarly publications. Informati d edges in a graph can be obtained through central sociology. For example, eigenvector centrality trix to determine nodes that tend to be frequently vi de a platform for scholars / researchers to publish t rs. In addition to this they can create and upd rank the researcher’s achievement based on thei . Some sample academic social netwo)[23], that provides comprehensive search and mi social networks, Microsoft rosoft.com) that is a free academic search engine hey provide many innovative ways to explore sci s, connecting millions of scholars, students. of authors’collaboration from an academic social n  N OF S OCIAL N ETWORKS   on of a social network is heterogeneous and the m h or network. Nodes/Vertices (V) are objects i.e. m which are either uni-directional or bi-directional ongst the nodes. The complexity of mining real is that the data is multi-relational, heterogeneo tering simply depends on the principle that adjace d to be adjacent to each other, i.e. if B and C are a and C to be adjacent. Sub-Graph pattern mining is ing analysis to simplify the dense social netw etwork Analyzer [20] as number of connected 10 ion: personal ion including lly stored in ork called as work and the consumers to . structure and relationships ay be groups, on about the ity measures, 13] uses the sited. eir work and te their own efforts as a ks include; ning services Academic developed by ntific papers, etwork ulti-relational embers of the hat represent orld datasets s and semi- nt nodes of a djacent nodes an important rk. Network components,  International Jou network diameter, path length, n derived from the above are degr  betweenness centrality, closenes 1.3   C LUSTERING Clustering, also referred as uns nodes in a network based on th the network topology which ma of nodes in the network results i intense analysis at individual no The important metric needed for a measure of degree to which coefficient of node quantifies h global clustering coefficient is the network is the average of loc The local clustering coefficient    Where   is the     ∶  ∈  , is t However, as the academic so collaboration amongst them,    i Different nodes in the network c Euclidean Distance, Cosine Simi Euclidean distance is the distanc dimensions. , ∑   Euclidean distance will be sm webpages. The reason behind th use pair wise similarity measure Cosine similarity is used to c coefficient computes the probabi k is a neighbor of either  x or  y . retrieval and it does not consider cos,.‖‖.‖‖  Hence we are comparing the val 1.4   FOAF   S PECIFICATION   Resource Description FramewoWeb Consortium to represent m  based schema to describe perso objective of FOAF is to allow  based on machine readable we   nal of Wireless Network Security(IJWNS)Vol1, No1 umber of neighbors, density etc. The complex para ee distribution, neighborhood connectivity, clusteri s centrality etc. upervised classification is the process of identifyi similarity of attributes of the nodes. This process es the network analysis simple, fast and efficient. identification of sub-graphs which will give us sco es and the network properties using graph metrics. our analysis, the clustering coefficient   is defined b nodes in a graph tend to cluster together. The lo ow its neighbors tend to form a complete graph, ased on triplets of nodes. The average clustering al clustering coefficients of all nodes in the network    is defined as  |  |      ∶  ,    ∈  ,   ∈   ut-degree of vertex i, and he set of neighbors of vertex i. cial network is treated as an undirected grap s normalized as   ′2   an be clustered based on frequently used similarity larity and Jaccard Coefficient.[22] e between any two points (x,y) in the sample space        ll for almost similar documents and very high fo is isthat all terms in the document may not be simil for matching tag values in the RDF format of web ompare two documentsthat are represented as v lity that two nodes  x and  y will have a common nei his metric is used to compute document similarity i term frequency and placement order. Jaccardx,y  xx  es for the selected attribute tags in RDF document. k(RDF) is a family of specifications maintained b eta data in XML format. FOAF (Friend of a Frie ns and their social network in a semantic way[15 integration of data across different applications. home pages for people, companies and other d 11 eters that are g coefficient, ng groups of also redefines he clustering  pe to have an [19], [21] as cal clustering where as the coefficient of . due to the measures like f multiple r non-similar ar. Hence, we document. ctors.Jaccard hbor k  , given n information ∩y∪y  World Wide d) is an RDF ]. The design he project is ta. As social  International Jou networks deal with events that o of a node contains different attr framework and the FOAF vocab object. For example, the sample data represents the  subject   of the RD the property as “Person” and it’ FOAF vocabulary definitions sein Table 1. For example  foaf:k  specified in those tag definition node. Table 1. Elements o foaf:Document foaf:Image foaf:Organizationfoaf:PersonalPro   Sample RDF Document from <rdf:RDF xmlns:rdf="http://www.w xmlns:rdfs="http://www. xmlns:foaf="http://xmlns. xmlns:admin="http://web<foaf:PersonalProfileDocumentrdf: <foaf:makerrdf:resource=" <admin:generatorAgent r <admin:errorReportsTordf </foaf:PersonalProfileDocument> <foaf:Personrdf:ID="me"> <foaf:name></foaf:name> <foaf:title>Professor</foaf:title> <foaf:homepagerdf:resource="http<foaf:phonerdf:resource="tel:(217) <foaf:interest>Natural Language Pr<foaf:interest>Machine Learning</<foaf:knows> <foaf:Person> <foaf:name>Wen <foaf:homepage </foaf:Person> </foaf:knows> --------------/* list of all persons */ <foaf:publicationsrdf:resource="Yiz Mining.: Mining Text Data: 259-295-------------- /* list of all publications </rdf:RDF> The above FOAF document ex authors and publications of a contact number. The co-authorsand <foaf:homepage> which ar child node of <foaf:knows>.   nal of Wireless Network Security(IJWNS)Vol1, No1 ccur between nodes of the network, the object orie ibutes and methods. All this data is standardized b ulary. An RDF dataset contains a triple of  subject, element is of the form <foaf:Person>John <  F data set as foaf(Acronym for Friend Of A Friend) s object   as the value of the property with “John”lected for our analysis of academic social network ows  attribute specifies that our selected node kno s,  foaf:interest   specifies the research interests of FOAF document considered to measure similaritiefoaf:Person foaf:knows foaf:Homepage foaf:mbox foaf:interest foaf:name ileDocument foaf:mbox foaf:publications FOAF project .org/1999/02/22-rdf-syntax-ns#" 3.org/2000/01/rdf-schema#" om/foaf/0.1/" s.net/mvcb/"> about=]""> #me"/> <foaf:primaryTopicrdf:resource="#me"/> f:resource="http://keg.cs.tsinghua.edu.cn/tj/cs/foaf_cr:resource="mailto:jery.tang@gmail.com"/> ://www.cs.uiuc.edu/~hanj/"/> 333-6903 "/> ocessing</foaf:interest> oaf:interest> min Li</foaf:name> df:resource="http://arnetminer.org/person/wenmin-li6houSun,HongboDeng,Jiawei Han: Probabilistic Models f "/> */ tracted from the academic social network[24] pre researcher along with his personal attributes like ’ names and homepage resources are specified in  presented as child nodes of <foaf:person> whic 12 ted treatment ased on RDF redicate and   foaf:Person>  ,  predicate  or .Some of the are presented w the person he respective s. ator"/>   44700.html"/> r Text sents the co- email id and <foaf:name> inturn is the  International Jou Rest of the document is organi description is given in Section3 Experimental data set and resu Section 6 with possible future sc 2.   R  ELATED W ORK    Subgraph discovery in relationa [1] is done by considering numb Relation extraction in social n similar entity pairs according to relation label. Peter Mika [3] FOAF documents and RDF fra also for analysis. Social Action using link structure and find patt Model based clustering propos depends up on distance betwee has described the different type and web usage mining.Informati is demonstrated by [7] based on searches for typical patterns of Peter Mika [9], Semantic ann Processing techniques. Alan Mi and analysis of the structure of in matrix form by [12] to clust content retrieval proposed by [1 Of all the works mentioned ab clustering techniques will not b  proposed model covers differe document and the retrieval pro categories of social networks, aoffered by the social networks.   3.   P ROBLEM D ESCRIPT Academic social networks an researchers and academicians k concerned domain. In some app etc. the representation of conn researchers as nodes and relati involved in the highly dense net a single domain. The present representation, which need lot o However, if the graph is clustere  profiles, it would make the proc nal of Wireless Network Security(IJWNS)Vol1, No1 zed as follows. Section 2 gives the related work, . Architecture of our proposed method is discusse lt analysisarediscussed in Section5. We conclude ope of research. l graphs to retrieve important subgraphs known as er of vertices and edge connectivity to identify freq tworks using similarity developed by [2], the pr the collective context in the web documents and emonstrated the application of semantic web tech ework to analyze social networks by considering rediction by [4] have used SNA measures to aggre erns across different links. d by [5] states that the probability of a tie betwe them in an unobserved Euclidean social space. I- s of web mining i.e. web content mining, web str on propagation in a social network based on strong clustering coefficient. Frequent pattern mining pr  structural change in dynamic social networks. In tation of Wikipedia is implemented using Natu slove[11] addresses the issue of large scale meas ultiple online social networks. Co-author relation r set of coauthors. User profile representation and ] uses semantic clustering to identify user clusters. ove, multiple similarity measures were not consi e compatible to different categories of social netw t similarity measures applied to the basic attribu cess involve graph mining techniques. This mod s the RDF frame work is standardized representati ON offshoot of social networks provide a channel t ow about the current trends of research and find lications like Arnetminer[24], Microsoft Academic ectivity highlights the relation of co-authorship b ns as edges in the graph. Due to the huge num ork, it is difficult to identify the group of research ffered services are limited to profile extraction f time to scan the entire database and produce the r d based on research interests and further on the sim ss of information retrieval faster and efficient. 13 and problem in Section4. our work in SkyGraph by uent patterns. cess clusters ssigns a new nology using   graph metrics gate networks en two actors HsienTing[6] cture mining and weak ties  posed by [8] the work by ral Language rement study s represented  personalized ered and the rks. But, our tes of FOAF el suits wide on of service rough which experts in the Research[23] y identifying er of people rs working in nd graphical quired result. larity of their
View more
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!