Scholarly article on topic '“Brokering” Behavior in Collaborative Learning Systems'

“Brokering” Behavior in Collaborative Learning Systems Academic research paper on "Computer and information sciences"

CC BY-NC-ND
0
0
Share paper
Keywords
{Broker / Gatekeeping / "Knowledge networks" / "Social roles" / "Social learning" / "Web-based collaborative learning" / "Computer- supported collaborative learning (CSCL)" / "Social Network Analysis (SNA)" / "Communities of Practice (CoP)"}

Abstract of research paper on Computer and information sciences, author of scientific article — Cathleen M. Stuetzer, Thomas Koehler, Kathleen M. Carley, Gerhard Thiem

Abstract The meaning and the impact of social roles and their embeddedness in social systems have a long tradition in social sciences. Especially the identification of structural conditions of individuals is important for studying processes of innovation and diffusion. Prior research has shown that “mediating” actors acquire a special role in knowledge networks by forming dynamic chains of information flow in the process of knowledge transfer. Therefore, it is important to detect and describe the impact of these individuals in digital distance learning processes for a better understanding of the relationship between network structure and information flow. In a previous work five typical influential social roles of actors were extracted and examined in knowledge networks. This paper presents the special mediating role labeled as “Broker” in formal and informal online learning communities by using Social Network Analysis (SNA). In the research teachers and students in remote learning communities are examined by their embeddedness in different communication networks in macro and micro perspective via structural analysis. The data were collected from the most popular 120 discussion boards comprising834 users and 11030 articles in the distance learning system, called OPAL, that is actively used by 11 universities located in the state of Saxony, Germany. The social network data are represented as communication networks between individuals. Brokering behavior is characterized by examining role patterns and flow of information through the network of learners and educators in the distance learning system. The results of the study describe the impact of Brokers for the communication and collaboration process in digital knowledge communities.

Academic research paper on topic "“Brokering” Behavior in Collaborative Learning Systems"

Available online at www.sciencedirect.com

ScienceDirect PfOCSCl ¡0

Social and Behavioral Sciences

Procedia - Social and Behavioral Sciences 100 (2013) 94 - 107

8th Conference on Applications of Social Network Analysis - ASNA 2011

"Brokering" behavior in collaborative learning systems

Cathleen M. Stuetzerac, Thomas Koehlera, Kathleen M. Carleyb, Gerhard Thiemc

aUniversity ofTechnology Dresden, Weberplatz 5, 01217Dresden, Germany bCarnegie Mellon University, 5000ForbesAvenue, Pittsburgh, PA, 15213, USA cUniversity ofApplied Sciences Mittweida, Technikumplatz 17, 09648Mittweida, Germany

Abstract

The meaning and the impact of social roles and their embeddedness in social systems have a long tradition in social sciences. Especially the identification of structural conditions of individuals is important for studying processes of innovation and diffusion. Prior research has shown that "mediating" actors acquire a special role in knowledge networks by forming dynamic chains of information flow in the process of knowledge transfer. Therefore, it is important to detect and describe the impact of these individuals in digital distance learning processes for a better understanding of the relationship between network structure and information flow. In a previous work five typical influential social roles of actors were extracted and examined in knowledge networks. This paper presents the special mediating role labeled as "Broker" in formal and informal online learning communities by using Social Network Analysis (SNA). In the research teachers and students in remote learning communities are examined by their embeddedness in different communication networks in macro and micro perspective via structural analysis. The data were collected from the most popular 120 discussion boards comprising834 users and 11030 articles in the distance learning system, called OPAL, that is actively used by 11 universities located in the state of Saxony, Germany. The social network data are represented as communication networks between individuals. Brokering behavior is characterized by examining role patterns and flow of information through the network of learners and educators in the distance learning system. The results of the study describe the impact of Brokers for the communication and collaboration process in digital knowledge communities.

©2013TheAuthors.PublishedbyElsevierLtd.

Selection and/or peer-review under responsibility of Dr. Manuel Fischer

Keywords: Broker; Gatekeeping; Knowledge networks; Social roles; Social learning; Web-based collaborative learning; Computer-supported collaborative learning (CSCL); Social Network Analysis (SNA); Communities of Practice (CoP)

1877-0428 © 2013 The Authors. Published by Elsevier Ltd. Selection and/or peer-review under responsibility of Dr. Manuel Fischer doi: 10.1016/j.sbspro.2013.10.702

1. Introduction

The characterization of social behavior in social systems has a long tradition and it is regarded as a key topic in various approaches from sociology, psychology and communication sciences (Burt, 1992; Granovetter, 1978; Katz & Lazarsfeld, 1955; Lewin, 1947; Moreno, 1934; Simmel, 1890). Simmel (1890) highlighted the importance of individuals' participation in different social circles. Granovetter (1978, 1985) showed in his studies that different ways of relationships between individuals have different social impact (Granovetter, 1978, 1985). Burt discussed social capital of individuals as a resource of information access (Burt, 2000).

Gatekeeping is considered as one of the most popular terms in communication sciences and consists of answering the questions which, where, by whom, and with what effects information can flow. First ideas about opinion leaders in social systems derived from the social scientists Lazarsfeld et al. (1944). They developed a model of two-step flow of communication in which information diffuse through mass media in two steps. At first, the information is flowing to opinion leaders. Thereafter, the opinion leader acts as multiplier and transfers the information to others (Lazarsfeld, Berelson, & Gaudet, 1944). The gatekeeping theory is based on Lewin's studies about behavior in groups in which only a few actors have more specific function and gain greater influence in their social context than others (Lewin, 1947). Further gatekeeping models of communication were established by White (1950), Newcomb (1953), and Westley & McLean (1957). White (1950) introduced the metaphor of gatekeeping in mass media research. In his model he implemented gatekeeping as social construction of reality (White, 1950). Thereby he focused on the decision-making process of people. Newcomb (1953), influenced by Lewin and Heider, applied his Co orientation model in communication research (Newcomb, 1953). This model based on two communicators A and B and their object of communication X (as orientation) by which the information flow is influenced. Westley and McLean (1957), affected by Lasswell (Lasswell, 1948), expanded the Newcomb model and introduced the communicator as gatekeeper (Westley & MacLean, 1957). The definition of gatekeeping involves an "activity performed by a communication organization and its representatives" (Shoemaker, 1997). Furthermore Shoemaker defines gatekeeper as "in" or "out" decision points (Shoemaker, 1991, p. 2). Brokers in the online learning community OPAL are used synonymously with the terms gatekeeper and opinion leader. Brokers are in the position to play the role of decision point who bundles, filters, reduces and modifies information. Brokers evolve depending on their social environment and have the position to control the information flow (Freeman, 1977). This priority depends on their special property of embeddedness into their surrounding (Granovetter, 1978). The rapid evolution of technology in the last 35 years requires an increasingly rethinking of the education culture (Lattemann & Kohler, 2005). In particular, the demand for new advanced learning theories which explain the digital knowledge transfer processes moving to the foreground of modern education sciences.

So, this study aims to describe gatekeeping effects in online learning communities with the view of structural conditions from social network perspective. The question about the path of information in remote learning communities is emphasized to get an introduction about the impact of special roles like Brokers. The impact of the communication network on participants in distance learning systems during the knowledge transfer process is made clear by the interaction of emergent network roles of social actors during the collaboration process. To explore the network structure the methods and theories from Social Network Analysis (SNA) were used. This research focused on two research questions that address gatekeeping effects in online learning communities:

1. How can be brokering behavior in online learning communities characterized?

2. What is the impact of Brokers on the structure of online learning environments?

2. Background

In education sciences i.e. in the tradition of classroom learning, roles of both learners and educators were established. With the rise of the computer and the evolution of Web 2.0 technologies the understanding for the process of knowledge creation, information sharing, and knowledge management has changed fundamentally and the term collective learning was introduced as new paradigm in the context of education sciences (Dillenbourg, 1999). The scientific discussion about roles during the learning process changed increasingly. Currently, the research in this area focuses on the results in computer sciences, education sciences and psychology (Fischer, 2007). Latest used distance learning management systems are organized into traditional learning theories based on instructional designs so that learning models like the constructivism from classroom learning were applied to online learning communities (Kohler, Neumann, & Saupe, 2011).

In the discussion about learning behavior in online communities new terms occur which disclose the changing quality of online collective learning in comparison to learning in classrooms. The importance of online communities and the increasing need for improved digital-oriented knowledge management were first centralized in education sciences by the social scientists Lave and Wenger (1991). They assigned situated learning with their Communities of Practice (CoP) approach in the context of academic research. They discussed the phenomenon of group dynamics in digital knowledge networks (Lave & Wenger, 1998). CoP's are defined as "groups of people who share a concern or a passion for something they do and learn how to do it better as they interact regularly" (Lave & Wenger, 1998). A learning theory behind CoP's was introduced by Georg Siemens (2004). He highlighted the implementation of new learning theories in the age of Web 2.0 associating with social live, social communication and informal learning (Siemens, 2004). Learning is understood as process between actors and their embeddedness in a social system and it was introduced as Connectivism theory in the context of research in education sciences (Siemens, 2004). Stephen Downes (2006) emphasized in his scientific article about E-Learning 2.0 the popular position of the social integration of users in social networks during the knowledge transfer process (Downes, 2006). He argued: "a community of practice is characterized by 'a shared domain of interest' where 'members interact and learn together' and 'develop a shared repertoire of resources."'(Downes, 2006). Above all content analysis studies determined the current learning research in the area of education sciences i.e. to improve didactical concepts (Bruhn, 2000; Hammond, 1999; Hara, Bonk, & Angeli, 2000; Strijbos & Weinberger, 2010; Weinberger, 2005). For example, Hara et al. (2000) distinguished the participants in four roles: starter, wrapper, instructor and students. Schellens et al. (2007) established five participant scripted roles in his studies: starter, moderator, theoreticians, source searcher, and summarizer (Schellens, Van Keer, Valcke, & De Wever, 2007).

In recent years network studies are gaining in importance in education science. De Laat et al. (2004) implemented SNA in education research and identified five different roles in micro network perspective: discussion manager, process manager, content manager, knowledge manager and technical manager (De Laat, 2005; De Laat & Lally, 2004). Marcos et al. (2007) established the four participant roles in their network studies: teacher-guide, teacher-collaborator, isolated-learner, and coordinator-learner. Welser et al. (2007) distinguished two participant roles via SNA in macro network perspective: central and not central participants. Stegbauer (2009) examined in his network studies via SNA the conflict and cooperation behavior of participants within the knowledge network Wikipedia. He concludes that in spite

of the complexity of Wikipedia leadership elites emerge within the knowledge network (Stegbauer, 2009). Stuetzer et al. (2011) identified five emergent network roles of participants via structural analysis in macro network perspective in collaborative learning networks: alpha dog, broker, cosmopolitan, individualist, and sightseer (Stuetzer, Carley, Koehler, & Thiem, 2011). Based on this theoretical background, in this case study brokering behavior of participants in the collaborative learning network OPAL shall be discussed to get patterns of behavior of key actors in both macro and micro network perspective. In this paper the impact of the special position of a Broker is discussed to reveal a relational view to learning behavior in online learning communities.

3. Data

The relational data were collected from the most frequented discussion boards of the distance learning system OPAL. This learning management system is currently the most popular distance learning system in higher education in the state Saxony, Germany. OPAL combines collaborative blended-learning activities of 11 universities in the state Saxony, Germany, and is administered by the Educational Portal Saxony (BPS GmbH). The learning system supports the most popular forms of collaborative learning scenarios and is based on a core of modern AJAX technologies, such as discussion boards, learning groups, wikis, etc. The characteristics of the different participation roles are derived from relational analysis of the giant component in the discussion network with 834 participants in 120 discussion boards and 11030 articles. The data were analyzed in two ways.

At first, we extracted the communication network of participants. The communication network is defined as a two-mode network (UxF) in which the communication of users was extracted via posted articles from whom who has written at least one article in a discussion board. The participants, who did not post any contributions e.g. read-only participants, were not recorded. We analyzed the relationship between the participants and the discussion boards which is characterized as posted articles from users to get more information about the communication behavior of participants.

Second, we explored the network of participants within the learning network. The participant network is defined as one-mode network (UxU) in which the participation of users in discussion boards was extracted. We analyzed the participation network to get more information about the collaboration structure between participants in same discussion boards in order to extract the impact of Brokers.

4. Method

"One of the primary uses of graph theory in social network analysis is the identification of the most important actors in a social network."(Wasserman & Faust, 1994). This is the main topic for the following case study. In order to describe emergent network roles structurally, we use SNA as method for exploration. SNA is generally used as scientific research tool to describe social actors and their relationships. It seeks primarily to provide an explanation of how social behavior functions by the influence of social embeddedness. SNA is a body of methods and theories that supports people in exploring, describing and reasoning about social actors and their relationships (Scott, 1991). In general, social networks are represented as graphs which are characterized by nodes and edges. The advantage of this formal description is that it can be described various networks by using the same methods and algorithms so it can be integrated in the particular monitoring context (Jansen, 2006; Wasserman & Faust, 1994). In this case study the SNA software tool ORA is used (Carley, Reminga, Storrick, & Columbus, 2011) which was developed under the direction of Kathleen M. Carley at the CASOS Center at CMU,

Pittsburgh (USA). For identification of network roles standard network measurements for activity (Degree centrality (DC(U)), global centrality (Betweenness centrality (BC(U)), intensity (Weight centrality WC(U)), as well as local centrality (Eigenvector centrality EC(U)) for individual participants were introduced and established. The calculation of the values follows in general the definition from the relational statistics (Bonacich, 1983; Freeman, 1977, 1978/1979; Opsahl, Agneessens, & Skvoretz, 2010; Wasserman & Faust, 1994). See table 1 for a discussion of these measures.

Table 1. Measures

Measure Technical Name Basis Meaning

Degree Centrality Degree Centrality Number of other nodes this node is connected to A node high in degree centrality is generally in the know and has insight into who is doing what.

Global Centrality Betweenness Centrality Fraction of shortest paths between all pairs of nodes that go through this node A node high in global centrality has the power to connect disconnected groups and to broker opinions, and start or stop the flow of information.

Intensity Weight Centrality Cumulative weight of links to and from this node A node high in intensity has strong relations that may be frequently used, and will serve as an opinion leader with those to whom the node is connected.

Local Centrality Eigenvector Centrality Extent to which this node is connected to other nodes who are also highly connected A node high in local centrality is often part of a group that shares opinions, views, and can act cohesively. Further that node might be a leader of that group.

For explanation brokering behavior in the learning network we emphasize in this case study that not only one measurement is adequate to explain the role of participants in distance learning networks. We use a combination of four different measurements for activity, global centrality, intensity and local centrality from SNA and combine these to explore the variety of social behavior in learning communities. We underlie that brokering behavior is not only influenced by high activity in different discussion boards, but also by different intensity of article writing, and by central global and local positions of participants in the network. The following code book, table 2, shows the combination of these values for each participant acting as a Broker. If the value of DC(U), BC(U), WC(U) and EC(U) is above 5% from all the participants measurements it is coded with 1(+) otherwise with O(-).

Table 2. Coding book of brokering behavior

Activity (DC(U)) Global centrality (BC(U)) Intensity (WC(U)) Local centrality (EC(U))

0 (-) 1 (+) 0 (-) 0 (-) O(-) 1(+) O(-) 1(+) O(-) 1 (+) 1(+) 1(+) 1(+) 1(+) O(-) O(-) 1(+) 1(+) 1(+) O(-) 1(+)_1(+)_1(+)_1(+)

With the coding of brokering behavior of participants we extracted the properties of participants in the learning network. It can be characterized by high values of Betweenness centrality measurements (BC(U), UxU) of a node in the symmetric network. Combining the measurements for activity in different discussion boards (Degree (U), UxF), the strength of ties within discussion boards (Intensity (U), UxF) and local centrality in discussion groups (Eigenvector (U), UxU) the position of Broker could be explored with behavioral characteristics. Hence brokering behavior can be identified not only by a central position in the participation network, but also by diverse activity and different strength of ties in the communication network.

Brokers have several characteristics of interest. First they have the necessary connections to connect disconnected groups. Second, they are at points where they can control the flow of information and thus both start up information flows or shut down roomers, or conduct negotiations. Third brokers have access to wider ranges of information, often due to a wider range of connections, and as such can be the genesis of new information to the wider group.

5. Results

1.1. Communication network

The communication network (UxF) consists of 834 participants (U) in 120 discussion boards (F) with 1455 links between them. The links represents the written articles by participants in discussion boards. The network is weakly connected with the realization of only 1.5% realized links of all possible links. The diversity by writing in discussion boards is high with approx. 97% so that the network can be described as a heterogeneous network in which each participant writes in approx. 1.7 different discussion boards due to the use of the learning network. Summary statistics for this network are in Table 3.

Table 3. Statistics and visualization of the communication network (UxF)

Network-Level Measure Value

Count, Participants (IJ) 834

Count, Discussion boards (F) 120

Link Count 1455

Max 159

Link Weights 8 Avg 7.580

Stddev 14.592

Density 0.015

Diversity, Discussion boards 0.972

Load, Discussion boards 1.745

1.2. Participant network

The participant network (UxU) consists of 834 nodes which represent the participants within the learning community. The edges represent the relationship between the participants using the same discussion boards within the learning network.

The participant network is characterizable as component connected through 17709 links. In the network only 5% of all possible links are realized. The network can be characterized as weak connected network in which anybody can reach anybody else in only six steps with short distances between them. High Betweenness centralization measurements indicate that a few participants are more central placed on the shortest paths between many participants (Freeman, 1977; Wasserman & Faust, 1994). High Degree centralization indicates that participants are directly connected to disproportionately active participants in the learning network (Wasserman & Faust, 1994).

The participant network is especially marked by high Betweenness and low Degree centralization. That indicates that in the learning network the contacts are realized by a few central participants who are on the shortest path between each other. Approx. 24% of the links are realized by a few participants who are using many different discussion boards to keep in contact with each other. But the most participants

(approx.

in the learning network are using only one or two different discussion boards for

interaction so that the Degree centralization as activity measurement of the network is really low. The following map visualizes the analyzed participant network. The nodes represent the participants in the discussion boards. The edges represent the collaboration between participants by using same discussion boards. For additional details see Table 4.

Table 4. Statistics and visualization of the participant network (UxU)

Network-Lewi Measure Value

Count, Node 834

Link Count 17709

Component Count, Weak 1

Density 0.051

Diameter 6

Average Distance 2.933

Network Centralization, Betweenness Network Centralization, Degree 0.238 0.023

1.3. Brokering behavior in the communication network 1.3.1. Macro network perspective

Brokering behavior can be characterized by the distinctions in the communication and collaboration behavior of participants. First, we are characterizing the Brokers from a macro network perspective in the communication network (UxF). The following map, shown in table 5, shows the communication behavior of Brokers in the Participant-Discussion board network (UxF). The nodes represent the Brokers (colored orange circles) and their interaction in discussion boards (triangles colored black). The edges represent the communication via posted articles in the discussion boards. The strength of ties is visualized by the count of articles which was written by the Broker per discussion board.

Table 5. Characterization and visualization of brokering behavior in macro network perspective

Network-Lewi Measure Value

Count, Participants (U) 47

Count, Discussion boards (F) 119

Link Count 304

Max 159

Link Weights

Avg 10.421

Stddev 17.447

Density 0.054

Diversity, Discussion boards 0.971

Load, Discussion boards 6.468

Brokers fill only 6% (U=47) of the network and interact in 99% (F=119) of the discussion boards. Approx. 21% (1=304) of the realized edges evolve from the Broker. The density of his network is comparable to the others much less and only approx. 5% (Density=0.054) of all possible links are realized. So we characterize the Broker as participant who is weakly connected with others by sharing different resources. He is working very heterogeneous in different spaces but not in cliques, in which Brokers would use the same resources. By his diversity of postings in different discussion boards (Diversity=97.1%) a Broker gets the central position in the network to connect people and groups. The major contribution to discussion boards results from Brokers with maximum 159 written articles per panel. They are organizing their direct contacts by interacting in the most frequented discussion boards. By the property of diversity the Brokers get the position to connect everyone else.

1.3.2. Micro networkperspective

We analyzed the characteristics of communication behavior of Brokers within discussion boards in micro perspective. Comparable to the other participants Brokers write the most articles with 3168 in total and approximately 67 articles per participant. See table 6. In other words, the Broker can be characterized as the most active participant who is writing 29% of all articles in 99% of the discussion boards by only 6% of the participants.

Table 6. Statistics of brokering behavior

Characteristics of Brokers

Broker articles per discussion board 26.62

At the one hand he has low binding on discussion boards and writes comparable to the others approx. only 27 articles per discussion board. On the other hand he has a high diversity by posting articles in approx. 6.5 different discussion boards. As a result the Broker gets the central position by many weak ties and high diversity in discussion boards within the communication process. The Broker interacts as bridge

of information and holds the position to transfer knowledge and control the information flow. The following map, in Figure 1, shows the typical distribution of the strength of ties of Brokers within discussion boards.

The strength of ties of Brokers per discussion board

(U=47, F=23, W=3168)

56% .....................................................................

I ■ 23%

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

No of discussion boardsper Broker

Fig. l.Strength of ties of Brokers per discussion board

Brokers write in at least two different discussion boards. They have strong ties in a few discussion boards. The more they interact in different discussion boards the weaker are the ties within one discussion board. We characterize brokering behavior in two ways. First, the Broker has the position to get and share information by his diverse activity via weak ties in discussion boards. And second, Brokers transfer information to his strong ties in only a few discussion boards. So they can be translated as a connector in his weak tie environment and as an innovator in his strong tie environment.

1.3.3. Structure of communication

In the next step we analyzed the communication structure of Brokers via the way of writing contributions and his behavior in writing articles in formal and informal groups. If the participant only wrote introductory articles in discussion boards we characterized him as Initiator. In case the participants contributed only follow up posts we labelled the role of communication as Follower. Has the participant written both initiating and follow up articles depending on the discussion board, we characterized him as Role Switcher. Similar to the behavior of communication we made distinctions between the places where the Broker has written articles. Wrote the participant articles in discussion boards which were instructed by course leaders only, we characterized this as communication in formal groups. Has the participant written in discussion boards which were not depending on a course from teachers we characterized this as communication in informal groups. Participants writing articles in formal as well as informal groups are characterized as Group Switcher.

The Broker can be characterized as typical Role Switcher. He is changing his role of communication depending on the discussion board. Approximately 87% of all Brokers are switching their role of communication by writing follow up and initiating articles depending on the discussion board. Most of the Brokers can also be interpreted as Group Switchers (60%) who change their article writing between formal and informal learning groups. Consequently Brokers can not only be characterized by the different ways of communication but also by the variety of spaces where writing articles. Thus the interaction in

different social circles has a strong impact of the position in the communication network. Brokers get the central position by interacting in different formal and informal communities and get more information likewise more connections compared to the others due to the fact of changing the role of communication.

We summarize "brokering" behavior of participants in the communication process within the learning network depends on:

1. diversity of resources, which participants are using

2. activity of the participant, which is characterizable by his different strength of ties within discussion boards

3. Role switching in discussion boards via different type entries

4. Group switching via interaction in both formal and informal learning groups.

1.4. Impact of Brokers in the learning community

The first analyses have shown that the Broker has influential characteristics in the communication process due to his diversity, activity and his ability as role switcher and group switcher. In the next step we consider the Broker and his collaboration activities with other participants to show the impact of Brokers in the learning network. For this analysis we used the symmetric network of the discussion network and examine the position of Brokers via relational analysis in the one-mode network. The following map shows the participant network in which the nodes represent the participants and the edges demonstrate the participation in same discussion boards. The colored nodes (rectangles) represent the Brokers in the network. The colored edges describe the direct impact to the other participants in the process of information transfer.

Fig. 2. Visualization of the impact of Brokers in the participant network (UxU)

To illustrate the Brokers impact on the structure of the learning community we analyzed the network with Brokers compared to the same network without any Brokers. The results are summarized in table 7.

Table 7. Impact of Broker within the participant network (UxU)

Network-Level Measures Network with Brokers Change Network without Brokers

Count, Participants (U) 834 -5.64% 787

Count, Link (1) 17709 -22.63% 13701

Component Count, Weak 1 1500% 16

Average Distance 2.933 +19.34% 3.500

Density 0.051 -13.33% 0.044

Connectedness 1.000 -68.63% 0.314

Diffusion 0.998 -68.65% 0.313

Network Centralization, Betweenness 0.238 -69.22% 0.073

Network Centralization, Degree 0.023 -39.15% 0.014

The network without Brokers consists of 787 participants with 13701 links between them. The structure of the network without Brokers is high fragmented in 16 weak connected components. By the fragmentation of the network without Brokers the connectedness of the participants decreases by approx. 69%. The average distance between participants in the network without Brokers increases by approx. 20%. Since the Brokers acting as connectors are missing, the possibility that information can reach each other is very low (Diffusion) in comparison to the network with Brokers.

By missing the most central and most active participants in the network the Betweenness centralization and Degree centralization of the network without Brokers decreases immensely. So we summarize Brokers in the learning network have the most central position to connect people and offer the possibility for all others to keep in contact by using same discussion boards. The Broker is established as collaborator who sustainable connects groups by using different high frequented discussion boards. Not the Broker per se is influential for the structure of the network, but his network of collaboration to other Brokers influences the structure of the whole learning community.

6. Summary and Conclusion

The aim of information management in knowledge society of 21st century is characterized by innovation, distribution, and sustainability of information especially with the help of education technologies. The theoretical elaboration has shown that gatekeeping is an often addressed research field. The discussion about theories, models, and results from both communication and education sciences have given the possibility to characterize the behavior and the impact of Brokers in this learning environment from a micro and macro network perspective. As important sources for organizing information in the digital age pertain different communication spaces in which individuals interact (Simmel, 1890). The impact of individuals on the innovation and diffusion processes depends on their social embeddedness (Granovetter, 1978, 1985).

To learn more about the information transfer within online learning communities it is useful to get insights into the communication and collaboration structure of participants. In this study we emphasized that collaboration in learning communities was associated with communication before but we also pointed out that communication is not equal to collaboration. We analyzed the communication structure in the two-mode network and the collaboration structure of participants in the one-mode network via relational analysis. The communication network was defined as network between participants and discussion boards in which the relationship were characterized by the count of articles of participants within discussion boards. The collaboration structure between participants was extracted by the participants using of same discussion boards within the learning community. We aimed to detect the properties and the impact of key players labeled as Brokers in the digital learning system OPAL via SNA.

We discovered the patterns which are typical for the processes of diffusion of information and the impact of Brokers for the flow of information in learning networks. Brokers have a central position in the network. They are on the shortest path between all other participants and have the position to manage the information flow. The ability as connector between peoples and groups depends on his brokering behavior. Brokering behavior of participants in the presented study can be characterized by:

• heterogeneous interaction in different discussion boards (diversity)

• varied communication activity, which is characterized by his different strength of ties within discussion boards via article writing

• properties of both role and group switching behavior depending on the discussion board.

Brokers have more power than others in the communication process and are marked by their diversity of interaction. They are distinguished by high communication activities in the network in order to get the position as hub for the diffusion processes of information within the network. Based on the skill to switch roles the broker is in the position to share information. They get new information in their weak ties environment and transfer the new information in their strong ties environment. This is a typical characteristic for the process of diffusion of information (Granovetter, 1978). Brokering behaviour is not depending on formal or informal communication. So, Brokers are important for information transfer processes. They open the gate for information. Brokers are identified as information bridge in the learning network. They connect participants and groups and have the abilityto share information with each other. These results put us in the position to make patterns of brokering behavior a bit clearer respectively in collaborative learning communities.

The research has also highlighted the mechanisms of interactions between participants in collaborative learning networks and their impact on the structure of learning networks. These results are important by implementing new learning theories like the Connectivism theory. We showed that collaborative learning environments based on the evolution of social communities in which information transfer extracted via social interaction and social embeddedness. Information flow needs connections between participants. The higher the activity and diversity of participants is the more connected is the learning community and the more information can flow. The impact of brokering behaviour is influential for learning communities.

The limitation of the work is the analysis of the special instructed distance learning network OPAL. For further work data should be compared with other collaborative learning networks. Nevertheless, the research has given sociological, communicational, learning theoretical, and network analytical insights into the construction of collaborative learning communities. The impact of Brokers that are evolving in a

heterogeneous manner under the impact of structural conditions is considered to be a significant milestone in the development and transfer of learning management systems.

Acknowledgements

This research is funded by the European Social Fund (ESF), Germany, and the SAB Sàchsische Aufbaubank, Saxony, Germany. We are especially grateful Jens Schwendel and his team from the Educational Portal Saxony (BPS GmbH), Germany, for providing the data. The views and conclusions contained in this paper are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the European Social Fund (ESF), or the SAB Sàchsische Aufbaubank.

References

Bonacich, P. (1983). Representations forhomomoiphisms. Social Networks, 5(2), 173-192. doi: 10.1016/0378-8733(83)90024-2.

Bruhn, J. (2000). Förderung des kooperativen Lernens über Computernetze. [Facilitation of cooperative learning via computer networks]. Frankfurt a. M.: Lang.

Burt, R. S. (1992). Structural holes: the social structure of competition. Cambridge, Mass: Harvard University Press.

Burt, R. S. (2000). The network structure of social capital. Research in Organizational Behavior, 22, 345-423. doi: 10.1016/s0191-3085(00)22009-1.

Carley, K., Reminga, J., Storrick, J., & Columbus, D. (2011). ORA User's Guide 2011 Technical Report: Carnegie Mellon University, School of Computer Science, Institute for Software Research.

De Laat, M. (2005). Investigating group structure in CSCL: Some new approaches. INFORMATION SYSTEMS FRONTIERS, 7(1), 13-25.

De Laat, M., & LaUy, V. (2004). It's not so easy: Researching the complexity of emergent participant roles and awareness in asynchronous networked learning discussions. Journal of Computer Assisted Learning, 20(3), 165-171.

DiHenbourg, P. (1999). Collaborative Learning: Cognitive and Computational Approaches. Amsterdam: Pergamon.

Downes, S. (2006). E-Learning 2.0. eLearn Magazine. Retrieved from http://elearnmag.org/subpage.cfm? section=articles&article=29-1.

Fischer, F. (2007). Scripting computer-supported collaborative learningcognitive, computational and educational perspectives. New York: Springer.

Freeman, L. C. (1977). A set of measures of centrality based on betweenness. 35-41.

Freeman, L. C. (1978/1979). Centrality in social networks conceptual clarification. Social Networks, 1(3), 215-239. doi: Doi: 10.1016/0378-8733(78)90021-7.

Granovetter, M. (1978). The Strength ofWeak Ties. American Journal of Sociology, 1360-1380.

Granovetter, M. (1985). Economic Action and Social Structure: The Problem of Embeddedness. American Journal of Sociology, 91(3), 481-510.

Hammond, M. (1999). Issues associated with participation in on line forums: the case of the communicative learner. Education and Information Technologies, 4(4), 353-367.

Hara, N., Bonk, C. J., & Angeli, C. (2000). Content analyses of on-line discussion in an applied educational psychology course. Instructional Science, 28(2), 115-152.

Jansen, D. (2006). Einführung in die Netzwerkanalyse. Grundlagen, Methoden, Anwendungen. Wiesbaden: VS Verlag für Sozialwissenschaften.

Katz, E., & Lazarsfeld, P. F. (1955). Personal influence; the part played by people in the flow of mass communications. Glencoe, IH.: Free Press.

Köhler, T., Neumann, J., & Saupe, V. (2011). Organisation des Online-Lernens. In L. Issing & P. Klimsa (Eds.), Online-Lernen -Handbuch für das Lernen mit Internet. München: Oldenbourg Wissenschaftsverlag.

Lasswell, H. D. (1948). The analysis of political behaviour. London: Paul, Trench, Trubner & Co.

Lattemann, C., & Köhler, T. (2005). Multimediale Bildungstechnologien I. Anwendungen und Implementation. Frankfurt am Main [u.a.]: Peter Lang Verlag.

Lave, J., & Wenger, E. (1991). Situated learning: Legitimate peripheral participation. Cambridge: Cambridge University Press.

Lave, J., & Wenger, E. (1998). Communities of Practice: Learning, Meaning, and Identity: Cambridge University Press.

Lazarsfeld, P. F., Berelson, B., & Gaudet, H. (1944). The people's choice; how the voter makes up his mind in a presidential campaign (3d ed.). New York: Columbia University Press.

Lewin, K. (1947). Frontiers in Group Dynamics. Human Relations, 1(2), 145.

Marcos, J. A., Martinez, A., Dimitriadis, Y., & Anguita, R. (2007). A Role-Based Approach for the Support of Collaborative Learning Activities. e-Service Journal, 6(19), 40-58.

Moreno, J. (1934). Who shall survive? Die Grundlagen der Soziometrie: Leske+Budrich (4. Aufl., 1996).

Newcomb, T. M. (1953). An Approach to the Study of Communicative Acts. Psychological Review, 60, 393-404.

Opsahl, T., Agneessens, F., & Skvoretz, J. (2010). Node centrality in weighted networks: Generalizing degree and shortest paths. Social Networks, 32(3), 245-251. doi: 10.1016/j.socnet.2010.03.006

ScheHens, T., Van Keer, H., Valcke, M., & De Wever, B. (2007). Learning in asynchronous discussion groups: A multilevel approach to study the influence of student, group, and task characteristics. Behaviour and Information Technology - An International Journal on the Human Aspects of Computing, 26(1), 55-71.

Scott, J. (1991). Social network analysis : a handbook. London ; Newbury Park, Calif.: SAGE Publications.

Shoemaker, P. J. (1991). Gatekeeping. Newbury Park (Cal.); London; New Delhi: Sage.

Shoemaker, P. J. (1997). A New Gatekeeping Model In D. A. Berkowitz (Ed.), Social Meanings of News: A Text-Reader: SAGE.

Siemens, G. (2004). Connectivism. A Learning Theory for the Digital Age. eLearnSpace. Retrieved from http://www.elearnspace.org/Articles/connectivism.htm.

Simmel, G. (1890). Uber sociale Differenzierung. DigBib.Org. Retrieved from http://www.digbib.org/Georg_Simmel_1858/Ueber_sociale_Differenzierung.

Stegbauer, C. (2009). Wikipedia.Das Ratsel der Kooperation. Wiesbaden: VS Verlag.

Strijbos, J.-W., & Weinberger, A. (2010). Emerging and scripted roles in computer-supported collaborative learning. Computers in Human Behavior, 26(4), 491-494.

Stuetzer, C. M., Carley, K. M., Koehler, T., & Thiem, G. (2011). The communication infrastructure during the learning process in web based collaborative learning systems. Paper presented at the ACM-WebSci 2011, Koblenz, Germany.

Wasserman, S., & Faust, K. (1994). Social network analysis : methods and applications. Cambridge; New York: Cambridge University Press.

Weinberger, A., Ertl, B., Fischer, F., & Mandl, H. (2005). Epistemic and social scripts in computer-supported collaborative learning. Instructional Science, 33(1), 1-30.

Welser, H. T., Gleave, E., Fisher, D., & Smith, M. (2007). Visualizing the Signatures of Social Roles in Online Discussion Groups. Journal of Social Structure, 8(2).

Westley, B., & MacLean, M. (1957). A conceptual model for mass communication research. Journalism Quarterly, 34, 31-38.

White, D. M. (1950). The "Gate Keeper": A Case Studyin the Selection ofNews (Vol. 27, pp. 383-390): Journalism Quarterly.