FAIR USE NOTICE

FAIR USE NOTICE

A BEAR MARKET ECONOMICS BLOG


This site may contain copyrighted material the use of which has not always been specifically authorized by the copyright owner. We are making such material available in an effort to advance understanding of environmental, political, human rights, economic, democracy, scientific, and social justice issues, etc. we believe this constitutes a ‘fair use’ of any such copyrighted material as provided for in section 107 of the US Copyright Law.

In accordance with Title 17 U.S.C. Section 107, the material on this site is distributed without profit to those who have expressed a prior interest in receiving the included information for research and educational purposes. For more information go to: http://www.law.cornell.edu/uscode/17/107.shtml

If you wish to use copyrighted material from this site for purposes of your own that go beyond ‘fair use’, you must obtain permission from the copyright owner.

FAIR USE NOTICE FAIR USE NOTICE: This page may contain copyrighted material the use of which has not been specifically authorized by the copyright owner. This website distributes this material without profit to those who have expressed a prior interest in receiving the included information for scientific, research and educational purposes. We believe this constitutes a fair use of any such copyrighted material as provided for in 17 U.S.C § 107.

Read more at: http://www.etupdates.com/fair-use-notice/#.UpzWQRL3l5M | ET. Updates
FAIR USE NOTICE FAIR USE NOTICE: This page may contain copyrighted material the use of which has not been specifically authorized by the copyright owner. This website distributes this material without profit to those who have expressed a prior interest in receiving the included information for scientific, research and educational purposes. We believe this constitutes a fair use of any such copyrighted material as provided for in 17 U.S.C § 107.

Read more at: http://www.etupdates.com/fair-use-notice/#.UpzWQRL3l5M | ET. Updates

All Blogs licensed under Creative Commons Attribution 3.0

Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 Unported License.

Saturday, June 15, 2013

Fom Metata to Person of Interest, Part one

Metadata

From Wikipedia, the free encyclopedia



Metadata (metacontent) are defined as the data providing information about one or more aspects of the data, such as:
  • Means of creation of the data
  • Purpose of the data
  • Time and date of creation
  • Creator or author of the data
  • Location on a computer network where the data were created
  • Standards used
For example, a digital image may include metadata that describe how large the picture is, the color depth, the image resolution, when the image was created, and other data.[1] A text document's metadata may contain information about how long the document is, who the author is, when the document was written, and a short summary of the document.
Metadata are data. As such, metadata can be stored and managed in a database, often called a Metadata registry or Metadata repository.[2] However, without context and a point of reference, it might be impossible to identify metadata just by looking at them.[3] For example: by itself, a database containing several numbers, all 13 digits long could be the results of calculations or a list of numbers to plug into an equation - without any other context, the numbers themselves can be perceived as the data. But if given the context that this database is a log of a book collection, those 13-digit numbers may now be identified as ISBNs - information that refers to the book, but is not itself the information within the book.
The term "metadata" was coined in 1968 by Philip Bagley, in his book "Extension of programming language concepts" [4] where it is clear that he uses the term in the ISO 11179 "traditional" sense, which is "structural metadata" i.e. "data about the containers of data"; rather than the alternate sense "content about individual instances of data content" or metacontent, the type of data usually found in library catalogues.[5][6] Since then the fields of information management, information science, information technology, librarianship and GIS? have widely adopted the term. In these fields the word metadata is defined as "data about data".[7] While this is the generally accepted definition, various disciplines have adopted their own more specific explanation and uses of the term.


Duke TODAY



Would Metadata Have Caught Paul Revere?

Sociologist shows how social network analysis can single out key conspirators
 
June 12, 2013 |

Kieran Healy's Network Analysis
Map of 1772 Boston social network. 
Durham, NC - Duke Sociologist Kieran Healey of the Kenan Institute for Ethics has written a brief primer on how one might use the sort of metadata available to National Security Agency snoops to figure out a burgeoning "terrorist" cell -- in 1772 Boston.

Written in a sort of pseudo-18th century grammar (with mostly modern spellings, thankfully) Healy walks through the steps one would take to turn the membership lists of various organizations into a matrix of connections showing who the key agitators might be.

Using  this "Social Networke Analysis," but knowing nothing else about these names,  he quickly fingers Sam Adams and Paul Revere as "persons of interest."

The clever and instructive post has taken off on social media in the last day -- reaching 100,000 views on Tuesday, and spreading even more strongly on Wednesday.

"I wanted to give non-specialists a sense of how the structural analysis of what's being called 'metadata' works, and to show in a fun but hopefully telling way how much you can get out of that approach," Healy wrote in his blog the next day. "So I tried to emphasize that I was using one of the earliest, and (in retrospect) most basic methods we have, but one that still has the capacity to surprise people unfamiliar with (social network analysis)."




Kieran Healy



Using Metadata to Find Paul Revere

London, 1772.

I have been asked by my superiors to give a brief demonstration of the surprising effectiveness of even the simplest techniques of the new-fangled Social Networke Analysis in the pursuit of those who would seek to undermine the liberty enjoyed by His Majesty’s subjects. This is in connection with the discussion of the role of “metadata” in certain recent events and the assurances of various respectable parties that the government was merely “sifting through this so-called metadata” and that the “information acquired does not include the content of any communications”. I will show how we can use this “metadata” to find key persons involved in terrorist groups operating within the Colonies at the present time. I shall also endeavour to show how these methods work in what might be called a relational manner.

The analysis in this report is based on information gathered by our field agent Mr David Hackett Fischer and published in an Appendix to his lengthy report to the government. As you may be aware, Mr Fischer is an expert and respected field Agent with a broad and deep knowledge of the colonies. I, on the other hand, have made my way from Ireland with just a little quantitative training—I placed several hundred rungs below the Senior Wrangler during my time at Cambridge—and I am presently employed as a junior analytical scribe at ye olde National Security Administration. Sorry, I mean the Royal Security
Administration. And I should emphasize again that I know nothing of current affairs in the colonies. However, our current Eighteenth Century beta of PRISM has been used to collect and analyze information on more than two hundred and sixty persons (of varying degrees of suspicion) belonging variously to seven different organizations in the Boston area.

Rest assured that we only collected metadata on these people, and no actual conversations were recorded or meetings transcribed. All I know is whether someone was a member of an organization or not. Surely this is but a small encroachment on the freedom of the Crown’s subjects. I have been asked, on the basis of this poor information, to present some names for our field agents in the Colonies to work with. It seems an unlikely task.

If you want to follow along yourself, there is a secret repository containing the data and the appropriate commands for your portable analytical engine.
Here is what the data look like.


1
2
3
4
5
6
7
8
9
10
11
12
                       StAndrewsLodge LoyalNine NorthCaucus LongRoomClub TeaParty Bostoncommittee LondonEnemies 
Adams.John                      0         0           1            1        0               0             0
Adams.Samuel                    0         0           1            1        0               1             1
Allen.Dr                        0         0           1            0        0               0             0
Appleton.Nathaniel              0         0           1            0        0               1             0
Ash.Gilbert                     1         0           0            0        0               0             0
Austin.Benjamin                 0         0           0            0        0               0             1
Austin.Samuel                   0         0           0            0        0               0             1
Avery.John                      0         1           0            0        0               0             1
Baldwin.Cyrus                   0         0           0            0        0               0             1
Ballard.John                    0         0           1            0        0               0             0

The organizations are listed in the columns, and the names in the rows. As you can see, membership is represented by a “1”. So this Samuel Adams person (whoever he is), belongs to the North Caucus, the Long Room Club, the Boston Committee, and the London Enemies List. I must say, these organizational names sound rather belligerent.

Anyway, what can get from these meagre metadata? This table is large and cumbersome. I am a pretty low-level operative at ye olde RSA, so I have to keep it simple. My superiors, I am quite sure, have far more sophisticated analytical techniques at their disposal. I will simply start at the very beginning and follow a technique laid out in a beautiful paper by my brilliant former colleague, Mr Ron Breiger, called “The Duality of Persons and Groups.” He wrote it as a graduate student at Harvard, some thirty five years ago. (Harvard, you may recall, is what passes for a university in the Colonies. No matter.) The paper describes what we now think of as a basic way to represent information about links between people and some other kind of thing, like attendance at various events, or membership in various groups. The foundational papers in this new science of social networke analysis, in fact, are almost all about what you can tell about people and their social lives based on metadata only, without much reference to the actual content of what they say.

Mr Breiger’s insight was that our table of 254 rows and seven columns is an adjacency matrix, and that a bit of matrix multiplication can bring out information that is in the table but perhaps hard to see. Take this adjacency matrix of people and groups and transpose it—that is, flip it over on its side, so that the rows are now the columns and vice versa. Now we have two tables, or matrices, a 254x7 one showing “People by Groups” and the other a 7x254 one showing “Groups by People”. Call the first one the adjacency matrix A and the second one its transpose, AT. Now, as you will recall there are rules for multiplying matrices together. If you multiply out A(AT), you will get a big matrix with 254 rows and 254 columns. That is, it will be a 254x254 “Person by Person” matrix, where both the rows and columns are people (in the same order) and the cells show the number of organizations any particular pair of people both belonged to. Is that not marvelous? I have always thought this operation is somewhat akin to magick, especially as it involves moving one hand down and the other one across in a manner not wholly removed from an incantation.
I cannot show you the whole Person by Person matrix, because I would have to kill you. I jest, I jest! It is just because it is rather large. But here is a little snippet of it. At this point in the eighteenth century, a 254x254 matrix is what we call “Bigge Data”. I have an upcoming EDWARDx talk about it. You should come. Anyway:

1
2
3
4
5
6
7
                       Adams.John Adams.Samuel Allen.Dr Appleton.Nathaniel
Adams.John                 -            2        1                  1
Adams.Samuel                2           -        1                  2
Allen.Dr                    1            1       -                  1
Appleton.Nathaniel          1            2        1                 -
Ash.Gilbert                 0            0        0                  0
Austin.Benjamin             0            1        0                  0
You can see here that Mr Appleton and Mr John Adams were connected through both being a member of one group, while Mr John Adams and Mr Samuel Adams shared memberships in two of our seven groups. Mr Ash, meanwhile, was not connected through organization membership to any of the first four men on our list. The rest of the table stretches out in both directions.

Notice again, I beg you, what we did there. We did not start with a “social networke” as you might ordinarily think of it, where individuals are connected to other individuals. We started with a list of memberships in various organizations. But now suddenly we do have a social networke of individuals, where a tie is defined by co-membership in an organization. This is a powerful trick.

We are just getting started, however. A thing about multiplying matrices is that the order matters. It is not like multiplying two numbers. If instead of multiplying A(AT) we put the transposed matrix first, and do AT(A), then we get a different result. This time, the result is a 7x7 “Organization by Organization” matrix, where the numbers in the cells represent how many people each organization has in common. Here’s what that looks like. Because it is small we can see the whole table.

1
2
3
4
5
6
7
8
                  StAndrewsLodge LoyalNine NorthCaucus LongRoomClub TeaParty BostonCommittee LondonEnemies
StAndrewsLodge              -         1           3            2        3               0             5
LoyalNine                    1        -           5            0        5               0             8
NorthCaucus                  3         5          -            8       15              11            20
LongRoomClub                 2         0           8           -        1               5             5
TeaParty                     3         5          15            1       -               5            10
BostonCommittee              0         0          11            5        5              -            14
LondonEnemies                5         8          20            5       10              14            -
Again, interesting! (I beg to venture.) Instead of seeing how (and which) people are linked by their shared membership in organizations, we see which organizations are linked through the people that belong to them both. People are linked through the groups they belong to. Groups are linked through the people they share. This is the “duality of persons and groups” in the title of Mr Breiger’s article.
Rather than relying on tables, we can make a picture of the relationship between the groups, using the number of shared members as an index of the strength of the link between the seditious groups. Here’s what that looks like.

"Group View"
And, of course, we can also do that for the links between the people, using our 254x254 “Person by Person” table. Here is what that looks like.
"Individual View"
What a nice picture! The analytical engine has arranged everyone neatly, picking out clusters of individuals and also showing both peripheral individuals and—more intriguingly—people who seem to bridge various groups in ways that might perhaps be relevant to national security. Look at that person right in the middle there. Zoom in if you wish. He seems to bridge several groups in an unusual (though perhaps not unique) way. His name is Paul Revere.

Once again, I remind you that I know nothing of Mr Revere, or his conversations, or his habits or beliefs, his writings (if he has any) or his personal life. All I know is this bit of metadata, based on membership in some organizations. And yet my analytical engine, on the basis of absolutely the most elementary of operations in Social Networke Analysis, seems to have picked him out of our 254 names as being of unusual interest. We do not have to stop here, with just a picture. Now that we have used our simple “Person by Event” table to generate a “Person by Person” matrix, we can do things like calculate centrality scores, or figure out whether there are cliques, or investigate other patterns. For example, we could calculate a betweenness centrality measure for everyone in our matrix, which is roughly the number of “shortest paths” between any two people in our network that pass through the person of interest. It is a way of asking “If I have to get from person a to person z, how likely is it that the quickest way is through person x?” Here are the top betweenness scores for our list of suspected terrorists:

1
2
3
4
5
6
7
8
round(btwn.person[ind][1:10],0)
     Revere.Paul     Urann.Thomas    Warren.Joseph      Peck.Samuel 
            3839             2185             1817             1150 
Barber.Nathaniel   Cooper.William     Hoffins.John       Bass.Henry 
             931              931              931              852 
    Chase.Thomas      Davis.Caleb 
             852              852
             
Perhaps I should not say “terrorists” so rashly. But you can see how tempting it is. Anyway, look—there he is again, this Mr Revere! Very interesting. There are fancier ways to measure importance in a network besides this one. There is something called eigenvector centrality, which my friends in Natural Philosophy tell me is a bit of mathematics unlikely ever to have any practical application in the wider world. You can think of it as a measure of centrality weighted by one’s connection to other central people. Here are our top scorers on that measure:

1
2
3
4
5
6
7
> round(cent.eig$vector[ind][1:10],2)
 Barber.Nathaniel      Hoffins.John    Cooper.William       Revere.Paul 
             1.00              1.00              1.00              0.99 
       Bass.Henry       Davis.Caleb      Chase.Thomas Greenleaf.William 
             0.95              0.95              0.95              0.95 
    Hopkins.Caleb    Proctor.Edward 
             0.95              0.90

Here our Mr Revere appears to score highly alongside a few other persons of interest. And for one last demonstration, a calculation of Bonacich Power Centrality, another more sophisticated measure. Here the lower score indicates a more central location.


1
2
3
4
5
6
7
> round(cent.bonpow[ind][1:10],2)
     Revere.Paul     Urann.Thomas    Warren.Joseph   Proctor.Edward 
           -1.51            -1.44            -1.42            -1.40 
Barber.Nathaniel     Hoffins.John   Cooper.William      Peck.Samuel 
           -1.36            -1.36            -1.36            -1.33 
     Davis.Caleb     Chase.Thomas 
           -1.31            -1.31 

And here again, Mr Revere—along with Messrs Urann, Proctor, and Barber—appears towards the top or our list.

So, there you have it. From a table of membership in different groups we have gotten a picture of a kind of social network between individuals, a sense of the degree of connection between organizations, and some strong hints of who the key players are in this world. And all this—all of it!—from the merest sliver of metadata about a single modality of relationship between people. I do not wish to overstep the remit of my memorandum but I must ask you to imagine what might be possible if we were but able to collect information on very many more people, and also synthesize information from different kinds of ties between people! For the simple methods I have described are quite generalizable in these ways, and their capability only becomes more apparent as the size and scope of the information they are given increases. We would not need to know what was being whispered between individuals, only that they were connected in various ways. The analytical engine would do the rest! I daresay the shape of the real structure of social relations would emerge from our calculations gradually, first in outline only, but eventually with ever-increasing clarity and, at last, in beautiful detail—like a great, silent ship coming out of the gray New England fog.

I admit that, in addition to the possibilities for finding something interesting, there may also be the prospect of discovering suggestive but ultimately incorrect or misleading patterns. But I feel this problem would surely be greatly ameliorated by more and better metadata. At the present time, alas, the technology required to automatically collect the required information is beyond our capacity. But I say again, if a mere scribe such as I—one who knows nearly nothing—can use the very simplest of these methods to pick the name of a traitor like Paul Revere from those of two hundred and fifty four other men, using nothing but a list of memberships and a portable calculating engine, then just think what weapons we might wield in the defense of liberty one or two centuries from now.

Note: After I posted this, Michael Chwe emailed to tell me that Shin-Kap Han has published an article analyzing Fischer’s Revere data in rather more detail. I first came across Fischer’s data when I read Paul Revere’s Ride some years ago. I transcribed it and worked on it a little (making the graphs shown here) when I was asked to give a presentation on the usefulness of Sociological methods to graduate students in Duke’s History department. It’s very nice to see Han’s much fuller published analysis, as he’s an SNA specialist, unlike me.

No comments:

Post a Comment