Papers and posters
Assessing the Citizens’ Network Positions
Download 6.3 Mb.
|
- Navigate this page:
- Degree Egos Egos (%)
- Closeness Egos Egos (%)
- Betweenness Egos Egos (%)
- 5. A Procedure of Putting the Results into Actions
- Marketing options Explanation Selection criterion for profiles
- Cluster Number of profiles Attributes
- 5. Example: Marketing Playstation IV on MySpace
- Marketing Option Selected attributes/ catchword/topics Selected Structure
4. Assessing the Citizens’ Network Positions This paper faces the challenge of combining: information of attributes which are frequently specified according to pre-classification schemes with structure information about the network and the actors. As outlined at the end of Section 3, social networks spawn over time a couple of knots that have a high number of links and a large number of knots only have a small number of links (heavy tail). Consequently, it seems promising to identify the ones with many links to use the structural position of them as a marketing advantage. This has been done with degree, closeness and betweenness, three different centrality measures which have been proposed for the identification of “important” MySpace profiles (Everett & Borgatti, 1999). In addition, the concept of ego networks is introduced to understand the further computation. The degree of centrality  provides an impression of the structure of the network by considering the number of connections from one knot i (i = 1, … , I) to other knots j (j = 1, …, i-1; i + 1, … , I) of the network:
  (1)
with k denoting the knot and I denoting the total number of knots in the network. In this study, the degree of centrality is deemed to be the dimension of possible communication activity within the network. The more links a profile has, the higher is the probability of direct communication with other profiles. Thus, we assess how applicative profiles are to start canvassing on these profiles with a high degree of centrality (Klaus & Wagner, forthcoming). The closeness centrality  provides an impression of how central a profile is in relation to others:
 (2)
with d(ki, kj) denoting the number of edges between the knot pair (i, j). In our application domain, the closeness centrality is deemed to be the dimension of independence from other profiles because the closer the centrality of a profile is, the more direct connections are linked to it. So, a profile is less dependent on another profile if it has many others close by. Moreover, this measure is assessed as the efficiency of a profile in all the other knots within the network. Considering the distance from one profile to all other profiles in the graph, the closeness centrality indicates how fast a marketing communication measure could spread through the network, starting at profile i (Klaus & Wagner, forthcoming). The betweenness centrality considers the shortest distances within the graph:   (3)
with  denoting the number of geodesics and  denoting the number of geodesics through  . In this study, the betweenness centrality assesses the opportunities for controlling the communication process. If many shortest distances run over a profile, it has a high influence on the network communication, assuming the user usually uses the shortest way to communicate. In this way, communication from these profiles can be monitored and assessed by marketers with a view to influencing them as they wish (Klaus & Wagner, forthcoming).
Each profile is also assessed by its ego network. This comprises a single actor (ego), the actors that are connected to it (alters), and all the links among those alters (Everett & Borgatti, 2005). Thus, the larger an ego network is, the more alters it has – these alters do not know, or barley know, one another – and the more different alters are in relation to their attributes, the more powerfully this ego can distribute information. Because of computer power and time capacity limitation, the data sample of 19,477 profiles was cut down by a requirement of the profiles to have at least four or more links to other profiles. The evolved new data sample includes 1,315 profiles which were used for the following analyses. First, the degree-, closeness- and betweenness- centrality were calculated for every profile. In the next step, small proper intervals over the measures were built to find out how many profiles belong to each interval. In the last step, we calculated the ego networks for the cumulative sum for all intervals and analyzed what percentage of the profiles of the whole network was reached by the ego networks with direct links. Selected parts of the results are listed in Table 1 as examples.
Table 1: Selected results of the reach build by a number profiles selected according SNA measures Table 1 depicts three cut levels for each assessment. The upper part, referring to the degree of the profiles, shows that within the network exist 50 profiles with a degree greater or equal to 22. Using these 50 profiles as egos and analyzing their combined ego networks, these 50 profiles could reach 802 other profiles via just one direct link out of the whole 1,315 profiles. This would consist of 60.98% of the whole network. To reach near the same percentage of the network with the closeness measure, 49 profiles with closeness greater than or equal to 26 should be contacted. They reach 797 profiles with their ego networks which consists 60.61% of the whole net. Last, looking at the betweenness, it is noticeable that one could choose smaller intervals for the betweenness from 2 to 3. Here jumps the number of egos from 42 to 198 and the reached percentage of the net jumps with it from 59.05 % up to 77.11 %. However, this does not matter since we only look at a sample data analysis. For real data to be interpreted, it would be necessary to choose a smaller interval for the betweenness to have a closer look at the accession of the reached alters. Summarizing the above results, it is possible to reach up to 60% or 65% of the whole network via just one direct link to other profiles by advertising on only 3-4% of the profiles. To achieve this result, the profiles of net citizens need to be selected carefully using the above criteria. In addition to this direct effect, the interactions of the alters enforce the communication effect. The assumption for this approach is that profiles which are connected visit or interact frequently with one another.
In this section, we outline four different marketing options for advertising on MySpace, which are summarized in Table 2. In doing so, we do not consider the classic banner advertising as part of mass communication media. This is already common online marketing and is also well done by MySpace. We focus on groups, forums, company/product profile and leaving comments for direct marketing in the community. With these marketing options, each company has the possibility to build up a social community around their product and brands with direct advertising.
Table 2: The four marketing options for advertising on MySpace A basic step for the implementation of marketing campaigns is defining the target group. In our application domain, we propose to include attribute-values and keywords depending on the company, brand, or product to be featured. For the forum and group marketing activities, the marketer has to choose an appropriate topic according to their company, brand, or product. The next step is calculating the centrality measures for the chosen sub network as outlined above. In addition, the ego networks need to be identified. For marketing implementations on user profiles, all the centrality measurements and the corresponding ego networks are important. Marketers need to choose a suited set of profiles on the grounds of their degree, betweenness and closeness values. In a first step, they have to specify what percentage of the network they would like to reach. Similar to the examples depicted in Table 1, this information is used to calculate the number of egos needed to get involved to achieve the aim. In a second step, they need to specify which egos are suited for the campaign. Not only egos with a high betweenness, closeness or degree are in the relevant set, but also egos with a high betweenness, egos with a high closeness and those with a high degree are included. The adequate mix is important, but largely neglected in contemporary marketing practice. Naturally, the intersection of ego groups, scoring high in the individual criteria, makes up the relevant set. A promising avenue for future research is the adoption of classic media selection models and media budgeting procedures to this concept of marketing on online communities. This procedure is suited to overcome the major weakness of current practices of marketing and all other attempts at triggering communication processes in social communities: selecting the profiles on the basis of a limited set of keywords. The pre-classification of keywords restricts net citizens in the expression of their interests. Moreover, users frequently do not specify all attributes (e.g., material state and sex), which might be used to cluster the profiles. In this paper, clustering aims to identify those egos which are seldom selected on the grounds of unspecified attributes or missing keywords. In order to find structurally equivalent profiles in the sub-network, and therefore to find equivalent profiles, the MySpace dataset is segmented. Typical methods for this task include, for example, the CONCOR algorithm (Boormann & White, 1976) and an approach from Burt (1976), which is based on hierarchical clustering. However, partitional approaches like the bisected k-means (Decker & Scholz, 2007) are more suitable for the clustering of large sparse data sets because they need relatively low computational expense in contrast to the agglomerative approaches. The bisected k-means is executed with cosine similarity as a measure and the I2 criterion function, which outperforms other criteria functions discussed in the clustering literature (Zhao & Karypis, 2004). The number of clusters is determined by the stability measurement from Lange, Braun, Roth, and Buhmann (2004). Table 3 summarizes the clustering results and gives an overview of the number, cluster size and the attribute-values, which fit around 70% of the cluster members. The largest cluster, comprising 649 profiles, consists mainly of musicians, who join the community for networking. The other three clusters, with a size of between 345 and 130, consist of normal users, for which the home country, material status and age can be specified. The attribute “here for” can only refer to the majority of the members of cluster C.
Table 3: Clustering results for the sub-network Even though 17% of the profiles do not specify any attributes, the clustering results show that this approach makes it possible to assign profiles with no attributes to groups with explanatory attributes. These attributes within a cluster help to pre-identify target groups to with which to communicate. A combination of the cluster analysis with the structure analysis seems to have promise for individual marketing campaigns on MySpace to communicate with its net citizens. 5. Example: Marketing Playstation IV on MySpace The MySpace community offers companies the possibility to build up a social community around their company or brand and to get into direct dialog with the users. A fictive product launch of Playstation IV should substantiate the four marketing options for MySpace. The marketing concept is a Playstation for the whole family: Playstation 4 ALL – Girls, Boys, Mom and Dad. The new Playstation is also a very interesting product for musicians because it can be used as a hard disk recorder for singers. For this new product, five target groups are defined with the following product focus: Boys: Age14-21, focus: Gaming. Dad: Age 21-99, focus: Gaming, media center. Girls: Age 14-21, focus: Karaoke, fitness. Mom: Age 22-99, focus: Fitness, wellness. Singer: Age 14-99, focus: Hard disk recorder for singers. We focus on the four marketing activities summarized in Table 4. The first step is done by defining the target groups for the new Playstation campaign. Therefore, the attribute-values are known for the profile selection. Based on the clustering results at the end of Chapter 4, we take our sub-network of 1,315 profiles for an adequate pre-selection for the profiles which match the target groups. The next step is to define what percentage of the network should be reached by the marketing campaign. We define 60%. For the marketing actions “own profile” and “leaving comments”, we have to choose a suitable mix of betweenness, degree and closeness for the selection of egos and their ego network. First, we state that we want to reach 60% of the network with these direct marketing options and for each of the three centrality measurements. Table 1 shows how many egos are needed to achieve this aim: 50 egos with a degree greater than or equal to 22, 49 egos with a closeness greater than or equal to 26 and 42 egos with a betweenness greater than or equal to 2. The intersection set of these three groups of egos comprises 97 net citizens. This means that the direct marketing campaign with 97 profiles, or 7.37% of the network, can achieve more than 60% of the whole network because they are the connecters. In the course of this campaign, the mix of the three centrality measurements for the selection of the egos could be rearranged according to their main function. Campaign activity (degree centrality): If there is too little conversation about one’s campaign in the social community, marketers should increase the number of egos with a high degree. If there is enough conversation in the social community, marketers can decrease the number of egos with a high degree; but they must keep a critical mass and change your focus on efficiency and monitoring/controlling. Campaign efficiency (closeness centrality): If the size of the new social community around the company/product is growing too slowly, marketers should increase the number of egos with a high closeness. If the social community is growing very fast, marketers can decrease the number of egos with a high closeness; but they must keep a critical mass and change their focus on activity and monitoring/controlling. Campaign monitoring/controlling (betweenness centrality): in order to monitor the opinions of the members in the new social network, there have to be enough egos with a high betweenness. in a critical situation, marketers should increase the number of egos with a high betweenness. Forums and groups are selected by matching forum and group topics with suited catchwords: gaming, entertainment, music, family, hi-fi, wellness, fitness, music and recording. To promote a viral effect, a widget design – which every user could integrate in his MySpace side – would be used. It is thanks to the visual application of Playstation IV that the product is tested.
Table 4: Concretion of the four marketing options for a fictive product launch 6.Conclusions This paper introduces a new approach to advertising on online communities, using the example of the MySpace community. The basic concept is that the target group for online direct marketing is to select the net citizens by both their specified attributes, and by their structural position in the communication network. For this purpose, we extend the methodology to reveal the systematic patterns of influencers and recipients in online social networks. Estimating the power function of link frequencies provides us with an assessment of how a network might be suited for word-of-mouth communication and viral marketing activities in combination with content. It turned out that the MySpace network was well suited. Moreover, we outlined three measures for the degree of centrality of citizens and the related interpretation. In the case of the MySpace network, it has been found that involving about 4% of the members would be sufficient to bring about 60% of all members into contact with the marketing communication. This could be illustrated with the example of a product launch campaign on MySpace. We discuss four qualities of marketing communication actions and their impact. Finally, we propose using both the assessment of the individual’s position within the network and the net citizens’ demographic variables to identify archetypes of users on social networks such as MySpace. Future research should compare different social communities, use even larger datasets to validate our results and test different SNA measures and methods to combine the structural information with the content. Moreover, an analytical criterion to assess the minimal percentage of the net citizens that a communicator should infiltrate is needed. Furthermore, a challenging task is to analyze how the position of a member of the social community influences the celerity of diffusion of the marketing communication.
187-203.
Boormann, S.A. & White, H. C. (1976). Social structure from multiple networks II. American Journal of Sociology, 81, 1384-1446. Bundesministeriums für Wirtschaft und Technologie (2009): 11. Faktenbericht 2008. Burt, R.S. (1976). Positions in networks. Social Forces 55, 93-122. Caverlee, J. & Webb, S. (2008). A large-scale study of MySpace: Observations and implications for online social networks. Proceedings of the Association for the Advancement of Artificial Intelligence. Decker, R. & Scholz, S.W. (2007): Unsupervised topic detection in document collections – an application to marketing journals. International Journal of Business Intelligence and Data Mining, 2(3), 347-364. Dwyer, C., Hiltz, S.R., & Passerini, K. (2007). Trust and privacy concern within social networking sites. Proceedings of the Thirteenth Americas Conference on Information Systems.
Everett, M. & Borgatti, S.P. (1999). The centrality of groups and classes. J. Math. Soc 23(3), 181-201. Everett, M. & Borgatti, S.P. (2005). Ego network betweenness. Social Networks 27, 31-38.
Hinduja, S. & Patchin, J.W. (2008). Personal information of adolescents on the internet: A quantitative content analysis of MySpace. Journal of Adolescence 31(1), 125-146. Karandikar, A. (2007): Generative Model to construct Blog and Post Networks in Blogosphere. Thesis, University of Maryland. URL: http://ebiquity.umbc.edu/_file_directory_/papers/359.pdf. (retrieved, 2009-01-08) Klaus, M. & Wagner, R. (forthcoming). Exploring the Interaction Structure of Weblogs. In: Advances in Data Analysis, Data Handling and Business Intelligence, Berlin, Springer. Lange, T., Braun, M.L., Roth, V., & Buhmann, J.M. (2004). Stability-based validation of clustering solutions. Neural Computation 16(6), 1299-1323. Leskovec, J., McGlohon, M., Faloutsos, C., Glance, N.S., & Hurst, M. (2007). Cascading behavior in large blog graphs. SIAM Conference on Data Mining. Ma, H., Yang, H., Lyu, M.R., & King, I. (2008). Mining social networks using heat diffusion processes for marketing candidates selection, CIKM 2008: Proceeding of the 17th ACM Conference on Information and Knowledge Mining, New York: ACM, 233-242. Milgram, S. (1967). The small-world problem. Psychology Today, 60-67. Newman, M.E.J. (2006): Power Laws, Pareto distributions and Zipf’s law. URL: http://www-personal.umich.edu/~mejn/courses/2006/cmplxsys899/powerlaws.pdf. (retrieved, 2009-01-07). Pennok, D., Flake, G., Lawrence, S., Glover, E., & Giles, C. (2002). Winners don`t take all: Characterizing the competition for links on the web. Proceedings of the National Academic of Science 99(8), 5207-5211. Peris, R., Gimeno, M.A., Pinazo, D., Ortet, G., Carrero, V., Sanchiz, M., & Ibanez, I. (2002). Online chat rooms: Virtual spaces of interaction for socially orientated people. Cyber-Psychology & Behavior 5(1), 43-51. Stelter, B. (2008, June 16). MySpace might have friends, but it wants ad money. New York Times, http://www.nytimes.com/2008/06/16/business/media/16myspace.html (retrieved, 2008-11-26). Stone, B. (2007, September 18). Myspace to discuss effort to customize ads. New York Times, http://www.nytimes.com/2007/09/18/technology/18myspace.html?_r=1&scp=12&sq=myspace&st=cse (retrieved, 2008-11-26). Thelwall, M. (2004). Link analysis: An information science approach. San Diego: Academic Press. Thelwall, M. (2008a). Fk yea I swear: Cursing and gender in a corpus of MySpace pages, Corpora, 3(1), 83-107. Available at: http://www.scit.wlv.ac.uk/~cm1993/papers/MySpaceSwearing_online.doc (retrieved, 2009-01-07). Thelwall, M. (2008b). Social networks, gender and friending: An analysis of MySpace member profiles. Journal of the American Society for Information Science and Technology. Vazquez, A. (2003): Growing network with local rules: Preferential attachment, clustering hierarchy, and degree correlations. Physical Review, E 67, Available at: http://www.sns.ias.edu/~vazquez/publications/local.models.pre.pdf (retrieved, 2009-01-07). Zhang, J., Ackerman, M., & Adamic, L. (2007). Expertise networks in online communities: structure and algorithms. Proceedings of the International World Wide Web Conference (WWW). Zhao, Y., & Karypis, G. (2004). Criterion functions for document clustering: experiments and analysis. Machine Learning, 55, 311-331. Directory: content -> download download -> Appendix appendix 0 the florida emergency telephone act download -> Fia wtcc race of Portugal boosts Championship Battle download -> Définitions download -> Press release 6 July 2012 Financial year 2011/12: facc achieves record turnover download -> Curriculum vitae eleanor jane sterling download -> Practical Application of the casve cycle of Career Problem Solving and Decision-Making download -> What are Competencies? download -> Bruce almighty Download 6.3 Mb. Share with your friends:
|