The International Journal of Engineering and Science (IJES)


Published on

The International Journal of Engineering & Science is aimed at providing a platform for researchers, engineers, scientists, or educators to publish their original research results, to exchange new ideas, to disseminate information in innovative designs, engineering experiences and technological skills. It is also the Journal's objective to promote engineering and technology education. All papers submitted to the Journal will be blind peer-reviewed. Only original articles will be published.

  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

The International Journal of Engineering and Science (IJES)

  1. 1. The International Journal of Engineering And Science (IJES)||Volume|| 1 ||Issue|| 2 ||Pages|| 260-268 ||2012||ISSN: 2319 – 1813 ISBN: 2319 – 1805 Framework for Analysis of Dynamic Social Networks Evolution  1 Karuna.C.Gull, 2 Akshata B. Angadi, 3Kiran B.Malagi 1, 2, 3 Department of Computer Science and Engineering, K.L.E. Institute of Technology, India.-----------------------------------------------------------------Abstract----------------------------------------------------------------------Recent years have seen an explosive growth of various online communities. The processes by whichcommunities come together, attract new members, and develop over time is a central research issue in the socialsciences—political movements, professional organizations, and religious denominations all provide fundamentalexamples of such communities. In the digital domain, on-line groups are becoming increasingly prominent due tothe growth of community and social networking sites such as MySpace, Twitter. However, the challenge ofcollecting and analyzing large-scale time resolved data on social groups and communities has left most basicquestions about the evolution of such groups largely unresolved: what are the structural features that influencewhether individuals will join communities, which communities will grow rapidly, and how do the overlaps amongpairs of communities change over time? So considering these, in this paper we present a framework for modelingand detecting community evolution in social networks. This framework allows tracking of events related tocommunities as well as events related to individual nodes. These events can be considered as building bloc ks forpattern detection in networks with evolving communities. This framework can be formalized by applying it to areal dataset consisting of emails. Keywords – Event based Framework, Group formation, Network Evolution, SNA (Social Network Analysis). ------------------------------------------------------------------------------------------------------------------------------------------ Date of Submission: 11, December, 2012 Date of Publication: 25, December 2012 ------------------------------------------------------------------------------------------------------------------------------------------ I. INTRODUCTION the challenge arises from the difficulty in identifying andT he tendency of people to come together and formgroups is inherent in the structure of society; and theways in which such groups take shape and evolve working with appropriate datasets: one needs a large,over time is a theme that runs through large parts of realistic social network containing a significantsocial science research [9]. The study of groups and collection of explicitly identified groups, and withcommunities is also fundamental in the mining and sufficient time-resolution that one can track theiranalysis of phenomena based on sociological data— growth and evolution at the level of individual nodes.for example, the evolution of informal close-knitgroups within a large organization can provide insight II. The Present Work: Analyzing Groupinto the organization‘s global decision-making Formation And Evolution:behavior; the dynamics of certain subpopulations In this paper we seek to address thesesusceptible to a disease can be crucial in tracking the challenges, exploring the principles by which groupsearly stages of an epidemic; and the discussions develop and evolve in large-scale social networks. Wewithin an Internet-based forum can be used to follow consider a number of broad principles about thethe emergence and popularity of new ideas and formation of social groups, concerning the ways intechnologies. The digital domain has seen a which they grow and evolve.significant growth in the scale and richness of on-linecommunities and social media, through the rise of We consider three main types of networking sites beginning with Friendster and • Membership. What are the structural features thatits relatives, and continuing to more recent systems influence whether a given individual will join aincluding MySpace, Facebook, and LiveJournal, as particular group?well as media-sharing sites such as Flickr. • Growth. What are the structural features that While abstract descriptions such as this — influence whether a given group will growof groups growing concurrently and organically in a significantly (i.e. gain a large net number of newlarge network —are clearly suggestive, the fact is that members) over time?it has been very hard to make concrete empiricalstatements about these types of processes. Much The IJES Page 260
  2. 2. Framework for Analysis of Dynamic Social Networks Evolution• Change. A given group generally exists for one ormore purposes at any point in time; in our datasets, III. RELATED WORKfor example, In the literature there has been a considerablegroups are focused on particular ―topics of interest.‖ amount of work done to detect communities in socialHow do such foci change over time, and how are these networks [1, 4]. A common issue in the previous workchanges correlated with changes in the underlying set is that the analysis of social networks was mainly aof group members? static investigation of the aggregated graph of the network across multiple snapshots. Hence, in the Conventional data is typically a set of noticeable effect of time was neglected. However, aobservations, which are considered independent of large number of social networks are continuouslyone another and identically distributed. In reality, data changing over time, thus they require a dynamiccould be highly dependent, with observations relating analysis. Recently there has been some work onto one other in a variety of relationships. The analysis, analyzing communities and their evolutions invisualizing, and interpreting of such relational data is dynamic social networks. Leskovec et al. [5] studiedknown as social network analysis (SNA). the patterns of growth for graphs based on various topological properties, such as the degree of Network input format is common to other distribution and small-world properties of largeSNA tools, using a simple pair of vertices to indicate networks. They also proposed a graph generationan edge. The Network file can optionally include model, called the Forest Fire model, to produce graphsexplicit lists of vertices, edge or vertex properties, and exhibiting the discovered patterns. Backstrom et al. [6]timeframe labels for dynamic networks. Meerkat‘s proposed using structural features of communities andfunctionality can be divided into four general individuals and then applying decision-trees tocategories: (1) interactive network visualization and approximate the probability of an individual joining anetwork metrics; (2) filtering and extraction; (3) community. They also tried to identify communitiescommunity mining; and (4) event analys is. These that are more likely to grow over time and predictedfeatures allow researchers to discover the importance the movements between communities based on theof entities in their domain networks, and make same features. Tantipathananandh et al. [7] presentedinferences about algorithmically discovered frameworks and algorithms to determine the evolutioncommunities. Being able to track entities and of communities in social networks. Although theycommunities over time, observing community assumed all groups are disjoint and explicitly defined,evolution, and performing analysis at different they tried to identify the notion of a community overhierarchical granularities allows for better leverage in all snapshots based on the changes in those exploration. They focused mostly on tracking the membership of an individual across all snapshots. Asur et al. [8] This idea of analysis over time, or dynamic analyzed the behavior of interaction graphs bynetwork analysis, is the final pillar on which Meerkat is defining critical events and computing them in anbuilt. Social networks in many domains are subject to efficient manner. They also introduced novelentities dropping in and out of interactions and thus behavioral measures such as stability, sociability,migrating across communities. To promote and influence and popularity for nodes and an incrementalfacilitate community mining across time, Meerkat way to calculate them over time. Falkowski et al. [9]offers event analysis functionality. analyzed the evolution of communities and studied their stability and fluctuation by defining similarity We propose an event-based framework to between them. Moreover, in order to identifycategorize and track how communities evolve in social persistent communities, they applied standardnetworks. Our framework takes the detected statistical measures. The use of on-line socialcommunities at consecutive snapshots as an input networking sites for data mining applications has beenand provides a mapping of how each community the subject of a number of recent papers; see [13,14]evolved at each snapshot. It also allows one to follow for two recent examples. These recent papers havethe events pertaining to individual nodes across each focused on different questions, and have not directlysnapshot. exploited the structure of the user-defined communities embedded in these systems. Studies of The rest of the paper is organized as follows. the relationship between different newsgroups onFirst in Section 2, we discuss related work. In Section Usenet [16,15] has taken advantage of the self-3, we describe our event-based framework in detail. identified nature of these online communities,Expected experimental results on sample email although again the specific questions are quitedatasets are discussed in Section 4. Finally in Section different.5, we discuss the conclusion and future The IJES Page 261
  3. 3. Framework for Analysis of Dynamic Social Networks Evolution According to Jiyang Chen et al. [4], IV. FRAMEWORK ELABORATION :Meerkat‘s community mining abilities are meant to 3.1. Event-based framework to detect evolutions insynergize with its other features. Although an analyst socialmight come to understand their networks through Networks:exploration of entities and through patterns visible Detecting the evolution of communities bythrough a particular layout, having automated monitoring when they form, dissolve, and reform cancommunity detection can give insight that is otherwise provide great insight into a dynamic social network.difficult to achieve. Given high quality community Asur et al. [8] proposed an event-based framework togroupings based on edge connections, a researcher capture and identify events on communities an dmay be able to alter social policies in government, individuals. Based on these events, the behaviouraladvise marketing plans in telecommunications, or patterns of communities over time can beunderstand what groups of proteins function together. characterized. Although they formulate the criticalTo facilitate this, Meerkat offers both existing and events for communities, and propose behaviouralnovel community mining algorithms. This includes measures for individuals, the presented events are tooseveral central algorithms, and slight variations of restricted to cover all of the changes that a communitythem, including Fast Modularity , Max-Min may experience.Modularity [12], TopLeaders , Clique Percolation ,Local Mining with or without hubs and overlap [11]. In this paper we present a framework for modeling and detecting community evolution in social The results of these algorithms are visualized networks. The framework allows tracking of eventsby colouring or labeling the nodes that belong to each related to communities as well as events related tocommunity, as well as listing community membership individual nodes. In order to define events that coverin tabular form as shown in Fig.1. After identifying all possible transitions of a community, a new termcommunities, Meerkat can compute common statistics called the community flag is defined. Base on thissuch as density, diameter, and cohesion for each concept, we propose event definitions that cover allcommunity. possible transitions of a community. Naturally, individuals in a community have mutual common interests and interact with each other around those interests. For example, members gather physically, or virtually, to share an idea or to discuss about a topic. This is exactly what identifies members from non-members. Although this is more sensible for human communities, artificial communities have the same patterns in their structure. Thus one can assume an independent identity for a community based on the interests that members share with each other. We call this identity the community flag, which shows characterization of the community and its members. A community flag is unique and cannot be divided or cloned. The life cycle of a community is defined asFigure 1.. Meerkat‘s hierarchy layout. The small follows. A community forms in a snapshot: Flag hascoloured circles are the nodes, and the edges depict been raised. It may be stable from a snapshot toan ‗owner‘ relationship in the hierarchy. Nodes within another: Flag is still there. It could attract newthe larger coloured circles belong to the community members or lose some members: Flag is waving. It mayindicated by its colour. The star-shaped objects incorporate another community: Dominant flag takesrepresent a community, with the grey objects being a control. It may divide into two or more smallercommunity made up of other communities and the blue communities, with each new part having its ownobject being the top-level super community. The independence: The most significant part carries thecommunity panel displayed on the right indicates flag with itself. Finally it can break apart into pieceswhich nodes belong to which communities while no piece preserves the identity of the community: Flag has been vanished. The identity of a community is defined by a significant portion of that community. However, this portion could be different in various The IJES Page 262
  4. 4. Framework for Analysis of Dynamic Social Networks Evolution Thus our new event definitions are k-form: A new cluster Cki+1 is marked as formed if atparametric based on this portion, denoted by k. In least k% of its nodes have not been a member of theorder to use our proposed framework, the social same community at the previous time. Thus Cki+1 isnetwork should first be converted into a time series formed ifgraph, where the static graph at each time captures the k jinformation at that specific moment. Then, based on a V i 1 V icommunity mining algorithm, the communities in eachsnapshot are obtained independently. Finally the transition of the communities C ik such that max V j i , V i 1 k   k%between two consecutive snapshots will be obtainedby the critical events defined in the framework. In the following, G = (V, E) denotes adynamic social network where V and E are the totalindividuals and total interactions respectively. Asnapshot Si = ( Vi , Ei ) of G represents a graph onlywith the set of individuals and interactions at aparticular time interval i. Each snapshot Si contains k icommunities Ci = {C1i , C2i , …, C k ii} where thecommunity Cji is also a graph denoted by (Vji , Eji ).For each two consecutive snapshots a total of 11events are defined with seven events involvingcommunities and fourother involving individuals in the network.3.2 Events involving communities: In order to categorize the changes ofcommunities that evolve over time, we consider sevenevents including form, dissolve, continue, split, merge,shrink, and reform. These events are based on the Figure 2. Community formation, merge and continue.relationship between communities and areparameterized based on the portion k. k-dissolve: A community Ck i is marked as dissolved if at least k% of its nodes will not be a member of the A community splits if it fractures into more same community in the next snapshot as depicted inthan one community and one of these communities Fig 3. Thus, the conditions for the dissolved iscarry the flag of the former community. In the case j kwhere it fractures into more that one community but V i 1 V i C i 1 such that V   k% knone of these communities carry the flag, a dissolve k jevent is occurred. A community continues if there max i , V i 1exists a community in the future that contains all thenodes of the former community. A community mayshrink or reform when it loses a portion of its membersbut this portion is not significant enough to bedetected as a split. In the case where new individualsjoin to the community, the community is marked asreformed, while it shrinks when no one has joined to it.Two or more communities are marked as merge if amajor portion of at least one of these communitiesinvolve in the merge. Furthermore at any snapshotthere may be newly formed community that does not Figure 3. Community Dissolvecarry the flag of any community at previous time. k-continue: A community Ck i is marked as continued if For two consecutive snapshots Si and Si+1 there existswhere Ci and Ci+1 denoting the set of their  i1 such that j acommunities respectively, the formal definitions of the C communityseven events involving communities are as The IJES Page 263
  5. 5. Framework for Analysis of Dynamic Social Networks EvolutionCji+1 that contains all the nodes of Ck i and at least k%of its nodes are belonging to Ck i. In other words, thetwo conditions for continue are as follows:  V i 1 k j 1. V i k V i  k% 2. j V i 1n-k-merge: A set of communities { C1i , C2i …, Cni} are There is a potential of raising the flag of Cj i in Cki+1marked as merged if there exists a community Cji+1 inthe next snapshot that for any community Ck i, thefollowing conditions are held: There is a potential of raising the flag of Cj i in C i+1 Also the flag of Cji has to be carried into one m of {C1i+1 , C2i+1 …, Cni+1} and it has to be dominant there: If the above condition is not held, the community Cji undergoes the dissolve event. A community may shrink or reform if it loses a portion of its members but this portion is not significant enough to be detected as a split. In the case where new individuals join to the community, the community is marked as reformed. On the other hand,The flag of Cm i has been moved into Cj i+1 it shrinks when no one has joined to it.Also at least one flag in Cji+1 has to be dominant inorder to distinguish this case and the case that a newcommunity has been formed from small pieces of someother communities as shown in Fig 2. Thus thefollowing condition should be held for {C1i , C2i …, Cni}: Figure 4. Community Splitn-k-split: A community Cji is marked as split (Fig 4)ifthere is a set of communities { C1i+1 , C2i+1 …, Cni+1} inthe next snapshot that for any community Ck i+1 thefollowing conditions are held: k-shrink: . A community Cki is marked as k-shrink(Fig 5) if there exists a community Cj i+1 that its set of nodes is a subset of the nodes in community Cki and also contains at least k% of the nodes from Cki . Thus the community is marked as k-shrink The IJES Page 264
  6. 6. Framework for Analysis of Dynamic Social Networks Evolution Disappear: A node v is marked as disappeared when it existed in the previous snapshot but it does not exist in the current snapshot i.e.Figure 5. Community Shrink Figure 7. Individual Node Appear and Disappeark-reform: A community Cki is marked as k-reform(Fig6) if there exists a community Cj i+1 that at least In Fig 7, Ci – Community Vi, Vi+1 – Vertices/Nodes Econtains k% of the nodes from Cki but its set of nodes – edges . Si , Si+1 – snap shots( current and nextis not a subset of the nodes in community Cki respectively.) Here V6 – Vertex V6 is said to be DISAPPEARED as it is present in Si snap shot, but not in the Si+1 snapshot. V6 – Vertex V6 is said to be APPEARED as it is present in the Si+1 sanp shot , which was not there in the S i snap shot. k-join: A node v joined to community Cji+1 if it exists in this community at snapshot i+1 but was not in Ck i in the previous snapshot where Cji+1 carries the flag of Ck i. Thus, the conditions for the join event are as follows:Figure 6. Community Reform Cj k %[ Ci Members] + set of nodes  Ci  c j is Reformed Community.3.3 Events involving individuals In order to analyze the behaviour of k-leave: A node v left community Ck i if it existed in thisindividuals in communities, four events involving community at snapshot i but it does not exist in Cji+1 inindividuals are defined. The taxonomy we use here is the next snapshot where Cji+1 is sufficiently similar tothe same as Asur et al. [8]. However, unlike [8] we Ck i. In other words, the conditions for the leave eventdefine the join and leave events parameterized based are as follows:on the portion k. For two consecutive snapshots Siand Si+1, the events involving individuals are definedas follows:Appear: A node v is marked as appeared when it is inthe current snapshot but it was not in the previoussnapshot The IJES Page 265
  7. 7. Framework for Analysis of Dynamic Social Networks EvolutionV. EXPECTED EXPERIMENTAL RES ULTS : means a community has been formed and similarly We have taken a sample dataset in order to disappearance of a colour shows the end of lifeshow the feasibility of the proposed events using our for that community.framework. To visually track the evolution of 2.Also tracking the community transitions such ascommunities, we have integrated our code into reformation, shrinkage, merger, and split areMeerkat [10].This tool enables us to preview the graph almost possible by looking at Table 2.of each timeframe and have the communities at eachtimeframe marked with different colours. In fact these Disadvantages of Asur Framework :colours are the notion of Community Flag and they 1.Since there is no notion of a community identity income from the results of our event-detection formulas. Asur framework, determining a map from communities in one snapshot to another is Without loss of generality, we will choose impossible.only one year data set to reduce the graph size, and 2.There is no way to keep track of a specificonly considered people who had sent at least one community and its transitions over time whenemail per day to filter out non-informative nodes. The using this framework.resulting graph will b having almost 250 nodes and Thus, in order to compare the results found by the1500 edges approximately. Lets set the snapshots to two frameworks, the number of events found bybe 1 month each and found the communities on each Asur and our framework are provided in Table 1month by a local community mining algorithm with no and Table 2 respectively.overlap between communities [4], provided inMeerkat. Using Asur framework, most of the communities are not marked by any event. On the other hand, our framework detects exactly one of the continue, reform, In order to evaluate our framework we have shrink, split, or dissolve events.also implemented the event-based framework by Asuret al. [8] which is the only framework that has an event Thus, the number of communities at eachbased approach similar to our framework. Fig 8 shows snapshot is the same as the total number of continue,the general view of communities in each snapshot. reform, shrink, split, and dissolve events. From TableThe area of each community in the figure is 2, we can observe that for the dataset, the reform andproportional to the number of its members. dissolve events far outnumber the other events. The high number of reform event indicates that most communities do not change greatly between two consecutive snapshots. However, the relatively high number of dissolve event denotes that most of the communities have short life cycles. So we can conclude that in the dataset most of the communities have a short life cycle and do not change drastically. Table 1. Number of events occurred for the dataset using Asur FrameworkFigure 8. shows the general view of communities ineach snapshot.Advantage of our Framework:1. By assigning colours to the different flags, one can easily make a map between communities through snapshots(the communities without any color are the ones that only exist for one snapshot). For example,appearance of a colour in a The IJES Page 266
  8. 8. Framework for Analysis of Dynamic Social Networks Evolution REFERENCES Table 2. Number of events occurred for the dataset [1] M. E. J. Newman and M. Girvan. Finding and using our Framework evaluating community structure in networks. Physical Review, E 69(026113), 2004. [2] Scott White and Padhraic Smyth. A spectral clustering approach to finding communities in graphs. In SIAM International Conference on Data Mining, 2005. [3] Kai Yu, Shipeng Yu, and Volker Tresp. Soft clustering on graphs. In Advances in Neural Information Processing Systems, 2005. [4] Jiyang Chen, Osmar R. Zaïane, and Randy Goebel. Local community identification in social networks. In International Conference on Advances in Social Networks Analysis and Mining (ASONAM), Athens, Greece, 2009. [5] Jure Leskovec, Jon Kleinberg, and Christos Faloutsos. Graphs over time: densification laws, shrinking diameters and possible explanations. In KDD ‘05: Proceedings of the 11th ACM SIGKDD international conference onI. CO NCLUSIO N AND FUTURE WO RK: Knowledge discovery in data mining, pagesVI. Conclusion: 177–187, 2005. In this paper, we presented an event-basedframework to analyze different types of dynamic social [6] Lars Backstrom, Dan Huttenlocher, Jonnetworks. Defining the concept of a Community Flag Kleinberg, and Xiangyang Lan. Groupallows us to capture all of the possible events among formation in large social networks: membership,communities. This includes tracing the formation, growth, and evolution. In KDD ‘06:continuation and dissolution of communities. Proceedings of the 12th ACM SIGKDDMoreover, it detects events involving individuals in international conference on Knowledgethe network and tracks their behaviour. Applying our discovery and data mining, pages 44–54, 2006.framework on the sample dataset, we visualized the [7] Chayant Tantipathananandh, Tanya Berger-Life-Cycle of all communities and the events that Wolf, and David Kempe. A framework foroccurred in Corporation‘s final year. Our results on the community identification in dynamic socialdataset indicate that most of the detected communities networks. In KDD ‘07: Proceedings of the 13thin the sample have short life cycle while having stable ACM SIGKDD international conference onmembers during their life. Knowledge discovery and data mining, pages 717–726, 2007.5.2. Future Work: [8] Sitaram Asur, Srinivasan Parthasarathy, and Most existing community mining algorithms Duygu Ucar. An event-based framework forfind separated set of communities, where every characterizing the evolutionary behavior ofindividual is a member of exactly one community. interaction graphs. In KDD ‘07: Proceedings ofHowever, in social networks individuals may belong to the 13th ACM SIGKDD internationaldifferent communities which results in highly conference on Knowledge discovery and dataoverlapping and nested communities. One possible mining, pages 913–921, 2007.future research direction is to analyze the evolutions [9] Tanja Falkowski, Anja Barth, and Myraof overlapping communities based on the proposed Spiliopoulou. Studying community dynamicsevents in a dynamic social network. Furthermore in our with an incremental graph mining algorithm. Inwork, we only consider the events between two Proceedings of the 14th Americas Conferenceconsecutive snapshots. However, it is possible to on Information Systems (AMCIS), 2008.detect events for any number of contiguous [10] Alberta Ingenuity Centre for Machine Learningtimeframes. Considering more than two snapshots at a (AICML) at the University of Alberta. Meerkat.time would enable us to detect communities that are fm, February 2010.inactive in a time frame which may reactivate again [11] Chen, J., Za¨ıane, O., and Goebel, R., ―Detectinglater on. Communities in Large Networks by Iterative Local Expansion,‖ International Conference The IJES Page 267
  9. 9. Framework for Analysis of Dynamic Social Networks Evolution Computational Aspects of Social Networks She is currently working as a Lecturer in (CASoN), Fontainebleau, France, June 24-27, K.L.E.IT, Hubli since 2011. 2009.[12] Chen, J., Za¨ıane, O., and Goebel, R., ―Detecting Mr. Kiran B. Malagi 3 .is from Hubli, Communities in Social Networks using Max- India. He has born on 8th July 1981. Min Modularity,‖ SIAM International He has received the B.E. degree in Conference on Data Mining (SDM’09), Sparks, Computer Science and Engineering Nevada, USA, April 30- May 2, 2009. from Visvesvaraya Technological[13] Lada A. Adamic, Orkut Buyukkokten, and University , India in the year 2005 and the M.Tech Eytan Adar, A Social Network Caught in the degree in Computer science and Engineering from the Web First Monday, 8(6), 2003. Visvesvaraya Technological University ,India in the[14] D. Liben-Nowell, J. Novak, R. Kumar, P. year 2010. Raghavan, A. Tomkins. Geographic routing in He has been working in the area of data social networks. Proc. Natl. Acad. Sci. USA, mining and social networking since 2010. He worked 102(Aug 2005). as Lecturer and head about 6 years .He is currently[15] F. Viegas and M. Smith. Newsgroup Crowds working as an Assistant Professor in K.L.E.IT, Hubli, and AuthorLines. Hawaii Intl. Conf. Sys. Sci. India since 2010. 2004.[16] C. Borgs, J. Chayes, M. Mahdian and A. Saberi. Exploring the community structure of newsgroups Proc. 10th ACM SIGKDD Intl. Conf. Knowledge Discovery and Data Mining, 2004. Biographies: is from Hubli, India. She has born on 7th June 1974. She has received the B.E. degree in Electronics and Communication from Karnataka University, India Karuna C. Gull1 in the year 1996 and the M.Tech degree in Computer science and Engineering from the Visvesvaraya Technological University ,India in the year 2008. She has been working in the area of datamining and social networking since 2009. She haspublished 1International paper on Data Mining (2009),2 International papers on Image Processing (2011). Shehas also published 3 National and 4 Internationalpapers in Conference Proceedings. She has alsoattended many of the workshops and conferencesheld in different places on High Impact TeachingSkills, Embedded System Using Microcontroller,Information Storage and Management (ISM) , DataMining, and many more. She worked as a Lecturer andSenior Lecturer for about 10 years .She is currentlyworking as an Assistant Professor in K.L.E.IT, Hubli,India since 2011.Akshata Angadi 2 received the BE degree in ComputerScience from Visvesvaraya TechnologicalUniversity,India in The IJES Page 268