Comments
You must login to participate in this chat. Please login.

Feel free to like, link, friend, follow, forward, rate, review, reply, comment, or connect!

Prospector

Thanks a lot! 

Prospector

Marc, a very cool topic. Social networking / big data analysis / cool tools.   All good stuff. Thanks for the info...

Master Analyst

Awesome. Any other questions? If not, have a great afternoon everybody!

Blogger

Also: See: http://www.smrfoundation.org

And: http://connectedaction.net

And: http://netbadges.com

And: http://nodexlgraphgallery.org

Thanks!

 

Marc

Prospector

http://nodexl.codeplex.com

Prospector

Thanks, Marc. I believe you shared a link to the NodeXL download, correct?

Blogger

Thanks for sharing Marc!

Blogger

Thanks for all the interest.

"Think Link"!

Prospector

If I missed anyone's question, feel free to email me: marc@smrfoundation.org

Prospector

Thanks, Marc. Any other big questions from the community in trhe minutes remaining?

Blogger

Who has access to the All Analytics back end?  If they can generate an "edge list" with this format:

User ID #1, User ID #2, Relationship Type (EX: Replies), Date, Time, etc.

We can import, analyze, and visualize.

Prospector

Wow the stuff just makes my head swim.

Master Analyst

Most sites with social features contain a network.

Getting that network data out is fairly straight forward.

I like to say, whenever two GUIDs can be joined, a network is born.

Prospector

Netbadges will introduce new types of Netbadges.  Right now we only award the "Bridge Builder Badge" for being the most like a bridge (max value of the Betweenness Centrality score).  Next we may award the "Newcomer" or "Hub" award.

Prospector

I wonder what interesting observations one could make of the allanalytics postings... this site has got pretty large audience...  (I'm not sure how you'd get the raw data...)

Master Analyst

Looking at networks with a real-time element sounds very exciting, Marc!

Blogger

Thanks Marc. You've got ambitious plans. :-)

 

Blogger

Netbadges: we will expand to include data from beyond Twitter.  Facebook, YouTube, the WWW, G+ are all targets.

We are looking carefully at Activity Stream as a new "one data structure to rule them all".

Prospector

As we're winding down here with just a few minutes until the end of the hour, I'm wondering if their are any more burning questions out there.

Blogger

Next for us:

NodeXL: more content analysis (see recent maps for examples of summaries of top URLs, Hashtags, and @usernames mentioned in each group in the network).  We will focus on ease of use and access to data sources (G+ importer coming in a week or two, for example).  

NodeXL will focus more on Time.  Animation and better analysis of change and contrast between networks is coming.

Prospector

When NodeXL is set up to run automatically, I mostly just look at the contents of its report output directory (which is synched to all my machines via Dropbox).  I just look at the most recent maps and reports.  In minutes I can review 100+ maps.

Prospector

Sorry guys, lost my 'Net connection for a moment. Case that happens again, I want to be sure to sneak one more question in: Marc, so you've got Netbadges and NodeXL. What's missing? Or, maybe I should say, what's next?

Blogger

Since the goal of these tools is, at least in part, to get actionable results, I'm imagining all of these shapes require a different approach for engagement.

Blogger

Polarized: see: http://www.flickr.com/photos/marc_smith/4971926421/ which is the map of people talking about US Politics.  Two or more dense clusters with relatively little interconnection,

Prospector

Brand: mostly isolates, few connections, lots of turn over

Prospector

Community: few isolates, lots of density (many connections among the people), some clusters but lots of interconnection

Prospector

Odd, the URL will not appear.  You'll have to cut it and paste it.

Shapes:

Broadcast: a hub with many spokes, most of which do not connect to one another.  often centered on the brand account.

Prospector

Can you define those shapes a bit more for us, Marc?

Blogger

Forgot to wrap the URL:

Prospector

We have learned a lot about structure and social media.  The variations you can see in Graph Gallery are one example.  Also see the collection flickr (the images are higher resolution in flickr): http://www.flickr.com/photos/marc_smith/sets/72157622437066929/

The shapes we see often are "broadcast", "community", "brand", "polarized".

Prospector

Our current focus is on automating the process from end to end.  These maps are now automatically collected, analyzed, visualized and published with a text summary.  The results are what you can see in some of my maps in NodeXLGraphGallery.org.  We want to make it so that making a network map is as easy as making a pie chart.

Prospector

Marc, your tool also looks at YouTube, correct. After mapping these network for soe time, have you noticed a significant difference in how communities are formed based on network structure?

Blogger

Marc, with some of your own "connection" research, what are some of the more interesting that you've found? 

Blogger

KenAa: I do monitor these maps daily. I look for big changes in their structures.  I look for new people in the top 10 lists.  I look for new topics.

Prospector

LOL, re "influential," Marc. Hard to get around, huh?

Blogger

Shawn: If you can get data out of FourSquare, I have no doubt that we can get it into NodeXL!

Prospector

Marc, being a dashboard geek, I was wondering if you ever keep any on-going 'dashboard-style' observations of any particular social media sites? Something that you keep an eye on over a period of time?

Master Analyst

What should you do when you know the mayor of your hashtag?  Yes, in some cases it is at least a good idea to follow them.  They may follow you.  You may even want to judiciously retweet them.  They will almost certainly follow you then.  And you may find that over time your tweets get greater visibility among these influential (there I said it) folks.

Prospector

Marc, about Foursquare, just wondering. I suspect the data would be interesting.

Blogger

Shawn: No Foursquare importer for NodeXL yet, sorry.  But we do import data from a wide variety of formats that you may be able to get Foursquare data in: GraphML, CSV, etc.

Prospector

So Marc, this might not be something you're privy, to but when they do find out who the "mayor of the hashtag" is -- then what? Do they try to influence that person, and use further mapping to measure success, for example?

Blogger

Speaking of mayors and getting back to location-based networks, does NodeXL import from Foursquare?

Blogger

But also, how do we compare?

For example, see: http://nodexlgraphgallery.org/Pages/Default.aspx?search=foundation

For a collection of maps for "Foundations" in twitter.

They vary a great deal - contrast Gates Foundation with Ford Foundation.

Prospector

Brands want to know: who is the mayor of my hashtag?

Prospector

Brand examples:

Yes - lots of companies now request maps for various topics on a daily basis.

Prospector

Shawn: volatility is an issue that can be addressed by controlling the history window.  We like volatility if we want "instantaneous" data - who is the center of the conversation *NOW*.  But we may also want to know about the centrality of people over a longer window, to rule out infrequent fluctuations.  These measures are based on whatever you feed them.

Prospector

And networking is the heart and soul of this process. Great point, Marc.

Blogger

Collective action dilemma theory is a useful framework for describing phenomena like wikipedia, message boards, photo archives, where many contribute to assemble and author the material and even more people come and take from the resource (often without ever making a contribution).  The magic of the Internet is that it reduces the size of the "minimal contributing set" - the fewest number of people needed to generate a resource, while increasing the ability of a group to find each other, and simultaneously making the resources reusable and findable.  That is a lot to change for a social process!

Prospector

Marc, can you share some specific examples of the type of brand and/or customer research companies have used NodeXL to do?

Blogger

Marc. How permentant are these centralities? Google and Kout ranks take time to change. But exporting data from Twitter on NodeXL or badges on a topic on Netbadges, how different could my results be day to day?

Blogger

wow, this is impressive, i got to try it

 

Prospector

Network data importers for NodeXL: http://nodexl.codeplex.com/wikipage?title=Third-Party%20NodeXL%20Graph%20Data%20Importers

Prospector

Ning Song: 

  • Flexible Import and Export   Import and export graphs in GraphML, Pajek, UCINet, and matrix formats.
  • Direct Connections to Social Networks  Import social networks directly from Twitter, YouTube, Flickr and email, or use one of several available plug-ins to get networks from Facebook, Exchange and WWW hyperlinks.
Prospector

KenAa: Prediction is a often sought goal.  I have questions about the predictive value of social media.  I think social media is largely reactive.  But reactions can be predictive, so I do not rule out its potential there.  But many want to, for example, predict the US Presidential race with "Tweet" votes.

Prospector

Hi, Mr. Smith, I see in your video, NodeXL can easily extract information from Twitter, how about facebook and Google+?

Prospector

The number of bridges are a function of how divided the network is.  Some topics are highly divided, separate groups that have few if any links to one another.  Political topics often look like this.

Prospector

Marc, regarding a macro-socialogical topic such as 'Collective Action' in regards to social media analysis, have you encountered techniques that would enable one to predict events like 'flash mobs' (maybe there's an emergant metric that indicates this). This would probably be based on twitter data...

 

Master Analyst

Have a look at NodeXLGraphGallery, compare a number of the maps.  Note how they vary in terms of the ratio of isdolates.  The maps with many isolates represent topics that could be called "brands" or "pubic" topics.

Prospector

Network growth does require an continuous influx of new comers.

BUT, step #1 is then to connect them to someone already connected to the big clump of interlinked people.

Prospector

Marc, what have you learned about the hub/bridge relationship? For example, I'm wondering how many bridges might we expect to stem from each hub? (The map you linked to is astounding!)

Blogger

To your point, Marc, efforts to grow a network must almost always start outside it.

Blogger

Network analysis is used to generate the routes and directions in mapping tools.

Prospector

Network analysis is recommending books to you on Amazon and Friend to you on Facebook.

Prospector

Network analysis is everywhere.  Google is all network analysis - PageRank is a refinement of Eigenvector Centrality - which is a network measure of how connected the people you connect to are.

Prospector

Yes, when you want to change an entrentched belief about the product or service, the hubs are important.

But even there, the hubs may have less value than the bridges.

So, I am simply advocating for the use of more measures of network location and the inclusion of more locations in the network as having possible value - all depending on your goal.  It is just that the goal is not *always* get the big shot to talk about me.

 

Prospector

Marc. Not sure if we got to Ning Song's question, but wanted to insert it here. She asks: what are the main fields where social network analysis applied? in addition to customer sentiment analysis and fraud detection?

Blogger

We could say that current models of influence detection only look for mountain peaks.  But valleys have value.

Prospector

Marc, what if you goal isn't new business? What if it's forging better customer relationships or some such? Then I would think the hubs would be of higher value?

Blogger

BethSchultz: well said.  I like that.

Prospector

Here is an example:

http://nodexlgraphgallery.org/Pages/Graph.aspx?graphID=447

Prospector

Marc. From a marketing researc standpoint, this may be the most interesting part of the tool and there are obvious actionable approaches you could use with a social media capaign at that point.

Blogger

Marc, so I suppose another way of looking at that is to say the isolate is the most capable of being influenced?

Blogger

Have a look at http://nodexlgraphgallery.org

 

Prospector

Most influencer scores suggest that hubs with many links are the most valuable people.  And they are valuable.  If you want to get a new message out to your existing customers the hubs are great broadcasters.  But they are likley to already be a customer, and thus have little to add in terms of new business.

Prospector

Most "influencer" systems ignore the isolate.

But I will argue that for the strategic goal of "new customer acquisition" the isolate is the most "influential" person in the room.  They have no connections to your community but they just said your name.  

Prospector

It also seems to me that if one wanted to pay attention to geographical location, another option might be to use a place name as a keyword or phrase. This might be particularly usefuil with smaller communities where people using the name, say on Twitter, would likely be residents.

Blogger

Marc, you lost me there on the "influence" of the "isolate." Can you elaborate?

Blogger

So, I suggest that we consider the "influence" of the "isolate".  Isolates have no connections.

Prospector

Re: Location.  While sparse, the lat/long data that does exist is illuminating and remarkably precise.

Prospector

Yes, NodeXL can import a wide range of attribute data from Facebook (where available).  (http://socialnetimporter.codeplex.com/)

Prospector

Oh, OK, thanks Marc -- lat/long would be of limited value, then.

 

Blogger

Then how about Facebook's geographic info. Can this be imported as well?

Blogger

Re: Location.  Many tweets are from desktops, or from mobiles that have not enabled GPS.  So only a few tweets are authored with Lat/Long data.

Prospector

Ooops. Black box. Thanks, smkinoshita!

Blogger

Yes: NodeXL has a Facebook importer.  See: http://socialnetimporter.codeplex.com/

Prospector

Marc, I like that comparison of Klout vs. NodeXL. I get it!

Blogger

I would say that http://Netbadges.com is more directly competing with Klout.

And there is plenty of room for more than one: note the existence of FICO, Equifax, TransUnion, and Experian.

Prospector

Marc, can you explain about the lat/long? Does that mean you get coordinates from a Tweet but that Twitter users themselves do not generally mention where they're tweeting from? If you've got lat/long does the absence of location mention in a Tweet matter?

Blogger

@Shawn -- "Block Box"?

Master Analyst

Marc, NodeXL also can import data from Facebook?

Blogger

NodeXL <> Klout

Klout is a kind of FICO score for the web.

NodeXL is more like Excel or Google maps: its a tool for examing data.

You could build a Klout score with NodeXL (if you knew their ingredients).  But you could not build NodeXL out of Klout.

Prospector

Marc, in short, Klout is what social media experts like Marshall Sponder in a recent post here might call a "block box."

Blogger

Geo graphic data is an interesting dimension.

NodeXL does download lat/long data from Twitter, but only about 5% of tweets have location data on them that I see.

Prospector

So Marc, do you see NodeXL as competing with Klout or is there room for both types of tools?

Blogger

So, NodeXL (and my other [commercial] project, http://netbadges.com) are all about building a local network of people who share an attribute, for example, they all say a keyword. 

Prospector

Interesting question, Beth, since digital networks now have geographic components.

Blogger

Klout is useful, but I have some critiques.
> It is opaque: I do not know how they calculate their results

> It is global: I want a more focused measure of a person's importance within a defined topical area

 

Prospector

Marc, since you bring up "geography," I'm wondering if you can overlay physical maps on top of the virtual maps NodeXL creates, enabling you to pinpoint geographically where influence originates to where it spreads?

Blogger

Bridges may have fairly few connections, but their connections matter more since they have the few links to some other group.  Social media is a collection of islands of conversation.  If you want to jump from cluster to cluster, you need bridges.  

Prospector

Marc, measuring social influence is all the rage now with tools like Klout. Can you share with us how your work differs from that.

Blogger

I like to think of all the "strategic locations" that are in a network: who is the hub (traditionally recognized as an "influencer") - but also, who is the bridge?

Prospector

In network theory we imagine a landscape, a human topography.

Like geography, some points on the map are more "central" or "peripheral".

Prospector

We're started, fountainhead. Jump in!

Blogger

what are the main field social network analysis applied? in addition to customer sentiment analysis and fraud detection?

Prospector

Influence is more complicated than a single number.

I propose an alternative term for what we seek: "strategic locations".

Prospector

When will it start?

Prospector

It's seems Beth and I are on the track here. As a side discussion, Marc and I were talking about some of the tools we'll be talking about. Maybe you'd like to bring us up to speed on some of the tools you've developed as folks arrive, Marc.

Blogger

Data sets can be very large: billions of edges (edges = connections between two entities).

But in practice, many interesting data sets are in the low thousdands of connections.

The tool we have built, NodeXL, is able to handle about what Excel can handle (varys by hardware, on the order of 100Ks to 4M rows).

Prospector

Marc, to add to what KenAa asked, how well do others -- marketers, primarily -- in your opinion understand social media as "a set of collections of connections?"

Blogger

In sociology there are many traditions.  I was trained in three.
Collective Action (aka Dilemma Theory).
Interactionist sociology.
and
Socal Network Analysis

Prospector

Hi Marc. These webs of links you build... how big can these data sets be? 

 

Master Analyst

Hi, all, I recently see a SAS social network analysis tool used in Fraud detection, it is amazing how the power of social network analysis  

Prospector

I was trained as a sociologist.  For me the Internet is the greatest opportunity for the social sciences in 100 years.

Prospector

Marc. Just to get us started as others arrive, I wondered if you could start off with just the basics of some of your research leading up to some of the tools we'll be discussing today.

Blogger

I see social media as a set of collections of connections.  I use social network analysis tools to collect and analyze these webs of links.

Prospector

I'm curious Marc about your background. What prompted you to focus in on social media?

Blogger

I am a sociologist studying the social structure of social media.  I make tools to collect, analyze, and visualize these data sets.

Prospector

If anyone has already signed in and would like to start sharing some questions on some of the online networking tools shared in the post. You can start doing that now as we get the conversation started.

Blogger

Hi all. I'm looking forward to today's chat as well. Understanding social media influence and activity is such a challenge!

Blogger

Hi folks. Here's where we'll be chatting with Marc A. Smith of the Social Media Research Foundation beginning in just a few minutes.

Blogger

Here's where we will chat with Marc A. Smith, a sociologist specializing in the social organization of online communities and in other computer mediated interaction. He is co-founder of the Social Media Research Foundation. The organization develops tools to measure and analyze the connection and construction of social media communities.  

Blogger


Latest Blogs
Don't forget Datakind's Jake Porway will preview his SAS Global Forum keynote with a live interview from the show Tuesday morning at 10 am (EDT).
SAS opened its SAS Global Forum and Executive Conference with a half dozen announcements focused on analytics for cybersecurity, risk management, UN trade data, Visual Analytics, and the big data talent gap.
Cloud-based analytics tools promise not only to make use of big data more efficient for the enterprise, but pave the way for citizen data scientists to better utilize data.
Robert uses SAS to recreate a stop-smoking message from the Centers for Disease Control and Prevention.
The SAS Global Forum and Executive Conference will have analytics professionals discussing and learning about the inner workings of technology but also how data analytics can help their companies and the world run better.
Radio Show
Radio Shows
UPCOMING
James M. Connolly
Live Interviews From SAS Global Forum


4/28/2015   REGISTER   0
ARCHIVE
James M. Connolly
How to Hire Great Analytics Talent


4/23/2015  LISTEN   51
ARCHIVE
James M. Connolly
Sports Analytics Mean Fun and Business


3/24/2015  LISTEN   4
ARCHIVE
James M. Connolly
Secure Your Big Data in the Cloud


2/26/2015  LISTEN   114
ARCHIVE
James M. Connolly
Make It Big As a Data Scientist in 2015


2/11/2015  LISTEN   106
ARCHIVE
James M. Connolly
Big Data, Decisions & the Simulated Experience


2/3/2015  LISTEN   87
ARCHIVE
James M. Connolly
A Chat About Big Data, Machine Learning & Value


1/15/2015  LISTEN   125
ARCHIVE
Curtis Franklin Jr.
An Infrastructure for Analytics


12/18/2014  LISTEN   63
ARCHIVE
James M. Connolly
Prepare for the Internet of Things Data Blitz


12/16/2014  LISTEN   50
ARCHIVE
James M. Connolly
How Mature Is Your Analytics Program?


11/18/2014  LISTEN   148
ARCHIVE
James M. Connolly
Drive Big Decisions Using Data & Analytics


11/10/2014  LISTEN   73
Information Resources
Infographic
Infographic
It Pays to Keep Insurance Fraud in Check
While 97% of insurers say that insurance fraud has increased or remained the same in the past two years, most of those companies report benefits from anti-fraud technology in limiting the impact of fraud, including higher quality referrals, the ability to uncover organized fraud, and improve efficiency for investigators.
Follow us on Twitter
Follow us on Twitter
Quick Poll
Quick Poll
Like us on Facebook
Like us on Facebook
About Us  |  Contact Us  |  Help  |  Register  |  Twitter  |  Facebook  |  RSS