Implicit Web Based Social Networks on Web Forums

[John Creighto]

Social Networks are ways of representing formal links between people. We often make it explicit using befriending mechanisms but their are also implicit methods, like if two people published a similar paper together. On a forum people post information from a variety of sources. The similarity in the sources people use forms an implicit link between them. Other types of links could be similarly categories and topics of interest.

Another type of similarity is the likelihood two people will comment on the same thread. This could form another method of creating implicit social networks and threads could perhaps be sorted by relevancy based on the number of people with similar interests who comment on it.

In the future it is suggested that social networking tools will be the primary method for us accessing information. Currently social networks are relatively primitive with news feeds being filled up with junk put out by apps and trivial comments. However, there is a wealth of network information that could help us better sort and filter which information we find relevant and which we don't.

To list these again:
like systems
rating systems
shared links of interest (similarities in common sites and such)
who started the thread
who commented on the thread
external links

Sorting though all this information to find an optimal method is an interesting challenge.

 

[berkeman]

Interesting thought. Almost like writing an application on top of vBulletin to catalog and visualize those relationships.

Two things will add noise to the data, though. First, there is inherent randomness in a forum like the PF, since folks get on at different times and may or may not see threads that they would respond to (and respond at different points in the discussion). Second, it seems like there is a bit more anonymity in a scientific discussion forum, versus the social networking sites like Facebook. That anonymity almost makes the contacts less "personal", but not fundamentally so.

 

[John Creighto]

Interesting thought. Almost like writing an application on top of vBulletin to catalog and visualize those relationships.

Two things will add noise to the data, though. First, there is inherent randomness in a forum like the PF, since folks get on at different times and may or may not see threads that they would respond to (and respond at different points in the discussion). Second, it seems like there is a bit more anonymity in a scientific discussion forum, versus the social networking sites like Facebook. That anonymity almost makes the contacts less "personal", but not fundamentally so.

Ideally I think that such tools are best if they extend across multiple sites. Some kind of open API would be good like RSS feeds. Tools could even be desktop based instead of web based. However, some things I would suggest to try to introduce these ideas into vbulitin are:

The implicit networks might be good as a way for recommend people to follow (as opposed to friend). Friending implies something more personal and therefore people are less likely to add people. It also requires that each party agree to add the other. However, the addon could have the option both befriend and follow.

A node does not need to correspond to a user. Rather it could be the combination of a user and a particular sub forum or even something more abstract like a topic with say three people from your network comment on. This also relates to more personal types of information. For instance you may or may not be interested in seeing someones status updates or personal pictures and they may or may not be interested in publishing them to you.

In news feeds, relevance is often done by sorting with respect to time but their should be methods to penalize people so their feed items get pushed to the bottom faster. Perhaps posters which you frequently rate with a dislike will get a greater time penalty and posts you like often may get a time penalty which actually works in their favor. Similarly frequent posters may get an implicit time penalty.

 

[brainstorm]

"Network" is a useful sociological concept insofar as it replaces "society" or "group" as super-individual object of study. Latour's writing on non-human delegates helped network theorizing a lot by showing how networks can be composed of various types of "agents." Discussing what can or cannot count as a network, a node, or an agent, is distracting imo. It's more effective to simply practice tracing interactive networks through chains of actions. You can do this by drawing something like a flow chart for a particular event. You could possibly label each node on the flowchart with whether it is a determined action or a moment of free-will, for example. The good thing about network-theorizing is that you can stick to the level of (inter)actions and events instead of getting into the demographics and institutional determinism of sociology that has "societies" as its objects. If education and economy are viewed as fixed structures of a society, how can you examine how specific individuals appropriate aspects of either institution to manipulate the other, etc? You need a network approach to analyze the specific complexes of individual interactions, including uses of institutions, technologies, etc. that make it possible for social projects to be achieved.

 

[aprillove20]

Social networking is defined as the grouping of individuals together into to specific groups, often like a small community or a neighborhood. Although social networking is possible in person, especially in schools or in the workplace, it is most popular online.