Profile cover photo
Profile photo
Scott Weingart
juggler, academic, nice guy
juggler, academic, nice guy
About
Scott's posts

Post has shared content

Post has shared content
A heat map analysis I did of the Democratic and Republican convention speeches using the word2word tool developed by Brent Kievit-Kylar at IU.  He has just added a "heatmap" feature to the software that lets you see the terms most diagnostic in a comparison. The "hotter" terms are the ones most diagnostic of a speech's being from the Democratic convention.  Hottest terms are "aarp, experience, affordable, abuelo, billionaires, afghanistan, beijing, biofuels, aliens".

The stats analysis is the BEAGLE holographic mental lexicon (a vector cosine similarity model that uses n-gram convolution to bind n-gram orderings) developed by Mike Jones.  The layout is MDS based on a "neutral" layout of a composite dataset consisting of all speeches from both conventions.

Another thing that's interesting about this visualization is the "triangular" shape, which holds up under many transformations of the data.  The X-axis seems interpretable as a "liberal->conservative" dimension, with liberal on the left and conservative on the right.  I interpret the triangular shape as evidence of high ideological focus on just a few issues by the Republicans (Obamacare, jobs, country, future, work) and a much more diffuse platform from the Democrats.
Photo

Post has shared content
Here's the network analysis of the same graph (see previous post).  Red links reflect high semantic similarity in Republican speeches, Blue links in Democrat.  The triangular shape is clearer here.

One thing I find really fascinating is the Democrats' emphasis on "know" and "knowledge" (lower left-hand corner).  
Photo

Post has shared content
Anatomy of a social network

Network researcher Ron Burt has identified two types of activities that create value in small-world networks: brokerage and closure.

Brokerage is about developing the weak ties: building bridges and relationships between clusters. Brokers are in a position to see the differences between groups, to cross-pollinate ideas, and to develop the differences into new ideas and opportunities.

Closure is about developing the strong ties: building alignment, trust, reputation and community within the clusters. Trust-builders are in a position to understand the deep connections that bond the people together and give them common identity and purpose.

These two kinds of activity, bridging and trust-building, demonstrate two very different ways that people and organizations can bring value to a network: Bridging leads to innovation and trust-building leads to group performance. The value that comes from these activities is known as social capital. Like every other form of capital, social capital represents stored value—in this case, relationship value—that can be translated into meaningful and tangible benefits.
The power of an individual node in any network can be considered along three dimensions: Degree, closeness and betweenness. 

Degree is the number of connections a node has to other nodes; for example the number of people in your family, or on your team at work, or the number of “friends” attached to your Facebook account. For an organization it could be the number of sales affiliates or business partners.

The value of a high degree is potential: the potential to connect and interact with a great number of other nodes in the network.

Closeness is a measure of how easily a node can connect with other nodes. For example you are probably very close to your team at work because it’s easy to connect to them: you can contact any person at any time. But you might be further away from other people in your company. Some you might be able to catch by walking down the hall or popping into their office, while to see others you might need an appointment, or you might need to be introduced by a mutual acquaintance. Anyone who has tried to make a connection on LinkedIn knows that the greater the distance, the harder it is to make a connection.

The value of closeness is ease of connection: The shorter the distance between you and other nodes, the fewer network “hops” you need to make, the easier it is for you to make connections when you need to. 

Betweenness indicates the degree to which a node forms a bridge or critical link between other nodes. For example, many executives are protected from distractions by executive assistants or secretaries who act as gatekeepers, who control access to the executive’s time and attention. 

The value of betweenness is the power you have to block or grant access to others. The more nodes that depend on you to make connections for them, the greater your potential value to them and thus the greater your power.

Thus, the most powerful person or organization in any network is one that has a high number of potential connections, all of which which are relatively close and thus easily accessible, while at the same time enjoying a position within the network such that it can choose to block or grant access to other nodes.
Photo

Post has shared content
After a very good start, the rate of signatures to the open access petition at the White House has slowed down to the point where, unless it speeds up again, the target of 25,000 will not obviously be reached, and ceratinly not surpassed by miles -- which was the hope. There have been just over 100 signatures since the beginning of today (which, by my calculations, means eight hours ago in Washington DC). If you haven't signed, then, unless you disagree with the petition, please take the short time needed to do so. And if you have, how about spreading the word?

https://wwws.whitehouse.gov/petitions/?#!/petition/require-free-access-over-internet-scientific-journal-articles-arising-taxpayer-funded-research/wDX82FLQ

Post has attachment
Wherein I write perhaps more than necessary about bias prevalent in scientific publication, and its relevance to digital humanities: http://www.scottbot.net/HIAL/?p=8832

Post has shared content
OFFICIAL White House petition to veto the #SOPA bill. I just signed it.

Go here: http://wh.gov/DfY and please re-share. And if anything deserves to be in the What's Hot Stream, this is it. Come on people. It's our internet.

EDIT: 8:54pm EST 10,246 signatures.
EDIT: 9:54pm EST 11,631 signatures.
EDIT:10:13pm EST 12,048 signatures.
Photo

Post has shared content
Elijah got to sharing this first (thanks!), but I suppose I should send it along the Google wave...
+Scott Weingart has done the world a service by giving a primer on the structure of networks and promises to write a series explaining networks and network analysis for scholars unfamiliar with the subject.

Post has attachment
Like my posts on networks and text? Maybe you'll like my post on networks and maps! (Like many of my posts, for some reason this one features +Elijah Meeks heavily. He gets everywhere first. Or I draw inspiration from him. Or maybe we just got there at the same time, and he's the Priestley to my Lavoisier.)

Post has attachment
Wherein I discuss combinations of topic modeling and network analysis.
Wait while more posts are being loaded