CHAPTER 6 DETECTING CHANGES OVER TIME
6.4 APPLICATION OF ALGORITHMS DETECTING COMMUNITY CHANGES – STUDY
6.4.2 Change Patterns Detected
For each change pattern detected, illustrative examples will be provided pointing out how the detected pattern could be used to inform possible support that could have been provided.
Change Pattern 1: Members moving to periphery.
Detection: During the time period Month4 – Month5 three members, M9, M15 and M24, were detected to be shifting to the community periphery. M9 and M24 stopped downloading resources from the community, and thus their centrality dropped (Figure 6.1). On the other hand, M15 used to be one of the influential members who uploaded resources to the VC but completely stopped contributing or downloading. The detection between Month5 and Month6, showed that two other influential members, M2 and M31, shifted to periphery and stopped contributing resources during Month5 and Month6.
Figure 6.1 M9, M15 and M24 appear to have relationships with others in the community in Month4 (left). During Month5 all
members appear inactive with no connections with others.
Furthermore, among others, M9 and M24 appear to have their neighbourhoods shrinking forReadSim, InterestSim and the in-neighbourhood of ReadRes between Month4 and Month5. Similarly, M28 who also appears to be one of the cognitively central members is detected to have a neighbourhood shrinking for UploadSim the in-neighbourhood ofReadRes. This means that M28 reduced his relevant uploading and downloading to/from the VC. In the period between Month5 and Month6, member M31 who is one of the CCen members of the VC is detected to have reduced neighbourhood forInterestSim, and both in- and out-neighbourhoods ofReadRes. This denotes that this member has reduced his reading from the VC and also the resources he is uploading are not of interest to the others. M17 who is the CCen member of this VC appears to have a reduced neighbourhood forUploadSim during the same period of time.
Support & Benefits: The detected movement to the periphery indicates that people who could influence the VC could have been encouraged to continue contributing, which could have helped the VC remain active. Having influential members actively engaged often motivates others to engage in the VC too. In this way, the cycle of knowledge sharing could have been kept active to benefit all community members.
Support can be provided to the detected members moving to the periphery in the form of notifications, letting them know about the drop of their CCen and trying to motivate them by pointing out how popular the resources they previously uploaded were. This is a way to make members understand how important the knowledge they hold is to the rest of the VC. This can improve the TM of the community since members will become aware of who is interested in what they are interested, and may also promote collaboration.
Additionally shrinking of the neighbourhood of members is either due to their drop of activity (reading/uploading), or for ReadRes out-neighbourhood due to others not reading from them. This situation can cause the VC to stop functioning after a short period of time. There is a need to support the detected members in order for the VC to sustain and for them to continue benefiting from their membership.
Interventions can be used based on the data collected from this pattern in order to encourage members of the community to continue being active and contribute/benefit from the VC. By showing to them people with similar interests, or how popular a resource they uploaded is, may encourage them to engage with the VC in a more effective, for them and for the VC, way.
Change Pattern 2: Changes in the behaviour of members indicate unexplored relationships.
Detection: M5 and M23 were detected to have ReadResand ReadSim in Month4 but appeared to have only ReadSim in Month5. This situation can be seen in Figure 6.2Figure 6.3Figure 6.4.
Figure 6.2 ReadResMonth4
andReadSimMonth4 for both members
Figure 6.3 Represents the graph for ReadSimMonth5 whereM5 & M23 appear to
have a connection among them
Figure 6.4 M5 and M23 are not connected in the
Month5 ReadRes graph
In the period from Month5 to Month6, five pairs of members, (M7 and M23, M7 and M31, M23 and M17, M28 and M31, M31 and M17), have been identified to satisfy change pattern 2. These are illustrated in Figure 6.5 and Figure 6.6. Figure 6.6 show that the detected members indeed were reading similar resources and they were reading also resources uploaded from each other during Month5.
Figure 6.5 The five members and the ReadSimMonth5 relation they have with others
in the community
Figure 6.6 ReadResMonth5
relation between the detected members
Figure 6.7 represents the ReadSimMonth6where members detected above continue to read similar
resources. However, they appear not to have ReadResMonth6( Figure 6.8), thus they are not reading
and they have also similar interests during that time. In Month5, they stopped reading similar resources but still have similar interests. In the period Month5-Month6, M5 is detected again along with M19. The latter detection is important since both members are newcomers to the community and they need to be supported accordingly.
Figure 6.7 ReadSimMonth6: members
detected continue to read similar resources in Month6
Figure 6.8 ReadResMonth6: Members
appear not to read resources uploaded from each other.
Support & Benefits: Members detected in this pattern have to be supported and encouraged to continue reading from each other and also to read resources the others are reading. Ignoring what others have uploaded to the VC can lead to missing important resources. The members may not be aware what is happening in the community and may not be aware how their interests/expertise is related to the VC.
The members detected with this change pattern could have been informed of their similarity in terms of interests and reading, and could have been offered recommendations what similar members are reading. This can develop awareness and promote TM (Wegner, 1986). Since members can become aware of who is working on similar knowledge areas, they can discover further opportunities for collaboration.
Change Pattern 3: Members are not integrating effectively.
Detection: Applying this change pattern, M5 was detected as a newcomer for Month4 and appeared to have some ReadRes relationships but not any UploadSim relationships during Month4 and Month5. This member downloaded 21 resources when he first joined the community in Month 4 and 6 resources during Month5. He appeared not to have uploaded anything during these two months. M5 is also detected in this pattern between Month4 and Month6 when he downloaded 14 more resources without uploading anything to the VC. During Month5 and Month6, M19 was also detected as a newcomer with the above behaviour. M19 downloaded 4 resources in Month5 and 33 resources in Month6, but did not upload any resources in Month6.
Support & Benefits: The behaviour of excessive downloading that both members developed shows that they struggled to find their way to the resources that were interesting and relevant to them. Consequently, support could have been provided to these newcomers by notifying them who works in similar areas as they do, and recommending them important resources. This may help develop SMM in the VC along with possible collaboration between members
Detection: M28 appears to be one of the cognitively central members is detected to have a neighbourhood shrinking for UploadSim the in-neighbourhood ofReadRes. This means that M28 reduced his relevant uploading and downloading to/from the VC. In the period between February and March 2006, member M31 who is one of the CCen members of the VC is detected to have reduced neighbourhood forInterestSim, and both in- and out-neighbourhoods ofReadRes. This denotes that this member has reduced his reading from the VC and also the resources he is uploading are not of interest to the others. M17 who is the CCen member of this VC appears to have a reduced neighbourhood forUploadSim during the same period of time.
Support & Benefits: Interventions can be used based on the data collected from this pattern in order to encourage members of the community to continue being active and contribute/ benefit from the VC. By showing to them people with similar interests, or how popular a resource they uploaded is, may encourage them to engage with the VC in a more effective way, both for the individual members and for the whole VC.
6.5 Summary
This chapter explored the potential of defining and utilizing community change patterns to identify when intelligent support is needed to support a community to function better as an entity. Change patterns have been described in two categories: detecting changes to aid support to be triggered and detecting changes in members’ behaviour due to the support provided. Section 6.3.1 discussed three types of change patterns to aid interventions and provided rational for considering each pattern showing how it can be related to three main processes (TM, SMM, CCen) which are crucial for effective and sustainable VCs. The results extracted from the study with the BSCW VC (section 6.4) show how change detection can be used to identify interventions that may help a VC function better and sustain.
In the study presented here, we have used data from a closely-knit VC operating on the BSCW system. The approach though is generic and is applicable broadly to any closely-knit community for knowledge sharing - a relationship model suitable for the specific community has to be built in the
form of graphs and the evolution algorithms can be adapted accordingly. It is important to note that this PhD does not aim to provide an exhaustive list of change patterns that can be discovered in a VC. Change patterns can vary from one community to another according to the community’s topic, purpose, and members. Using the basic principle presented in this research, further patterns can be defined.
Chapter 7 will provide detailed description of the support that can be provided due to the changes discovered in this chapter.