Social Capital in Online Communities
From DML
Online communities are connecting hordes of individuals and generating rich social network data. The social capital that resides within these networks is largely unknown. We propose to create a general framework for measuring and leveraging social capital based upon explicit social networks, implicit affinities, and social resources. The resulting quantitative models are used to characterize social capital in several online communities.
Contents |
[edit] Framework Components
We find it useful to distinguish between types of connections among individuals, as follows.
- An explicit connection links individuals together based on a well-defined relationship, such as "is a friend of" or "collaborates with." Individuals thus linked are aware of the explicit connections among them.
- An implicit connection links individuals together based on loosely defined affinities, or inherent similarities, such as similar hobbies or shared interests. Individuals thus linked may not be aware of the similarities in attitudes and behaviors that exist among them.
We call explicit social networks (ESNs), social networks built from explicit connections and implicit affinity networks (IANs), social networks built from implicit connections, and focus on their complementary natures.
Additionally, social resources are an important component of measuring social capital.
[edit] Explicit Social Network (ESN)
ESNs can be created using a variety of techniques. Some online communities (e.g., Facebook, Twitter, LinkedIn) have incorporated functionality that allows you to specify who you are explicitly connected to (e.g., who your friends are). Other techniques have been used to derive ESNs by using social interaction data. For example, ESNs have been generated by observing the frequency of email correspondence among individuals within a corporation (See Diesner et al 2005). Others, have derived ESNs using links among blogs, assuming that blogs that are "friends" link often to one another (e.g., Kumar et al 2003, Adamic et al 2005). Figure 1a shows possible explicit connections that make up an ESN for a sample set of individuals.
[edit] Implicit Affinity Network (IAN)
IANs are built from individuals represented as collections of attributes and associated value sets, where links are created whenever two individuals share an attribute whose value sets overlap. For example, in the network pictures the IAN is represented by dotted orange lines. Whenever an individual X adds a value, say v, to one of its attributes, say A, some amount of affinity is automatically added between X's node and all existing nodes whose individuals have value v for A. IANs tend to be highly dynamic and are subject to a chosen similarity metric.
[edit] Social Resources
As theorized by Lin, personal and social resources can be characterized for Individual actors. These resources are defined as either material goods (e.g. land, houses, car, and money) or symbolic goods (e.g., education, memberships in clubs, reputation, or fame). Personal resources (i.e., human capital) are in the possession of the individual, while social resources (i.e., social capital) are accessible through social connections (see Lin 2001). Resources gained through bridging interactions are perceived to be of greater worth as they are more likely to be dissimilar than the resources already available.
Lin characterizes access and mobilization as theoretical approaches that describe how social capital is expected to produce returns (see Lin 2008). Access estimates the amount of social capital (known to be) available to an individual. This approach is based on the assumption that the amount of accessible social capital largely determines the returns, without regard to the particular actions taken to use the social capital. Alternatively, the theoretical approach of mobilization reflects "a selection of one or more specific ties and their resources from the pool for a particular action at hand" (see Lin 2008). For example, using a specific contact having certain resources (e.g., a highly trafficked blog, or domain-knowledge) to boost sales on an e-commerce site could be indicative of mobilized social capital.
[edit] Applications & Experiments
Our framework is being applied to a variety of problems within multiple domains including mobilizing social resources on Twitter, increasing bonding capital through blogging, and even an off-line linguistics study.
[edit] Twitter
Twitter is an open community that was estimated to have 4-5 million users in November of 2008 [1] and was ranked as the third largest social network behind MySpace and Facebook in February 2009 [2]. This relatively new community allows users to contribute short free-form status updates about themselves and follow the updates of others. The status updates, called tweets, are a rich source of data that can be used to build implicit affinity networks, while the following and followers information can be used for explicit social network links. Furthermore, rich status update information among individuals including web links and re-tweets that might be used to quantify mobilized social resources.
Measurable Social Resources
- click-thrus (e.g., status update links, profile link) (measured using trackable links, such as bit.ly)
- re-tweets
- responses to questions
[edit] Blogosphere
The Blogosphere refers to the growing social network among people that write blogs, or web logs. We explore the social capital found in these networks of bloggers. Little work has been done with regards to implicit relationships between bloggers. We are creating a blog entry database for analysis of both implicit and explicit networks within the blogosphere.
- BlogGrabber is the python package available for download that we created to extract large amounts of blog feeds.
- The Blog Database Creation Process details of how the enormous amount of blog data has been obtained for use in these studies.
Measurable Social Resources
- comments (e.g., count, frequency)
- page-views
- 3rd party rankings (e.g., Google PageRank, Alexa, PageMass)
[edit] Building Community around a Blog
The Blogosphere, or the social network of people that read and write blogs, has become an important network to participate in. Blogs, short for weblogs, are created by businesses and ordinary people to achieve goals like informing customers of new products, giving expert knowledge on a particular topic, or sharing life events with family and friends; blogs can communicate information that people wish to share. There has been plenty of ideas about how to build community around a blog, which has resulted in anecdotal evidence that would surely benefit from additional testing. The focus of this project has been on testing these ideas in order to show their effectiveness. The initial results provide further guidance to building community around a blog.
- DML Blog - the Data Mining Lab blog was started to (1) share the interesting work we are involved in, and (2) test anecdotal theories of how to build community around a blog
- DML Blog Log (Google Doc) - the schedule of actions that have been taken on the DML blog
- SRC Presentation (PDF) - Slides reporting on the first seven weeks of the the Data Mining Lab blog (presented at the 2008 Spring Research Conference)
[edit] Our Publications
Smith, M., Giraud-Carrier, C., and Purser, N. (2009). Implicit Affinity Networks and Social Capital. Information Technology and Management, 10(2-3):123-134, September.
Smith, M. (2008). Social Capital in Online Communities. In PIKM '08: Proceeding of the 2nd PhD workshop on Information and knowledge management, pages 17-24, New York, NY, USA. ACM. (Alternate Download Location)
Smith, M., Purser, N., and Giraud-Carrier, C. (2008). Social Capital in the Blogosphere: A Case Study. To appear in AAAI Technical Notes.
Smith, M., Giraud-Carrier, C. and Judkins, B. (2007). Implicit Affinity Networks. In Proceedings of the Seventeenth Annual Workshop on Information Technologies and Systems, 1-6.
Smith, M. (2007). Implicit Affinity Networks. Brigham Young University.
