Visualizing THATCamp

THATCamps are quite popular. I’m throwing one myself. But who are the people talking about them on Twitter? What does the THATCamp look like on the Twitterverse?

I used NodeXL to retrieve the data – a search for tweets, people, and the links between them. I then visualized the data in Gephi, where colour = community (per Gephi’s modularity routine) and sized the nodes (individual Twitterers) using Pagerank, on the premise that this was a directed graph and one should follow the links (although there was little difference with Betweeness Centrality. Major players are still major, either way).

I found 233 individuals, linked together by 4435 edges. Some general stats on this directed network:

Top 10 Vertices, Ranked by Betweenness Centrality Betweenness Centrality
thatcamp 10493.93299
marindacos 3381.65598
amandafrench 2530.589717
openeditionsays 2491.27153
inactinique 2183.450362
briancroxall 2093.876857
piotrr70 2014.064889
brettbobley 1798.013658
miriamkp 1693.203103
melissaterras 1596.42585

 

Top Replied-To in Entire Graph Entire Graph Count
colleengreene 4
thatcamp 3
rosemarysewart 2
normasalim 2
janaremy 2
spagnoloacht 1
chuckrybak 1
lawnsports 1
ncecire 1
academicdave 1
Top Mentioned in Entire Graph Entire Graph Count
thatcamp 25
piotrr70 25
briancroxall 17
ncecire 16
spouyllau 14
thtcmpfeminisms 10
marindacos 8
dhlib2012 8
thatcamprtp 7
goldstoneandrew 6
Top URLs in Tweet in Entire Graph Entire Graph Count
http://leo.hypotheses.org/9506 26
http://bit.ly/RyrPvA 19
http://tcp.hypotheses.org/609 19
http://tcp.hypotheses.org/programme 15
http://rtp2012.thatcamp.org/apply/ 12
http://bit.ly/w1IFmR 11
http://dhlib2012.thatcamp.org/register/ 10
http://goo.gl/qJ185 10
http://dhlib2012.thatcamp.org/ 8
http://bit.ly/RNHLKO 8
Top Hashtags in Tweet in Entire Graph Entire Graph Count
thatcamp 137
mla13 27
dh 22
tcp2012 17
thatcampsocal 12
dhlib2012 9
unconferences 7
thatcamptheory 6
digitalhumanities 6
tcny2012 6

And now the visualization. You can download the zoomable pdf here.

As I look at the modularity in this graph, at first blush, you can see quite a North America / European divide, with various satellite outposts. This could be of course because there’s a THATCamp Paris coming down the pipe (lots of French in the tweets).

Artefact Networks Analysis – Thinking Out Loud

What follows is a process of me thinking out loud whilst working with data related to the Tiber Valley Brick Industry.

If we look at the distribution of stamped bricks across the Tiber Valley (as collected by the BSR), we can work out some of the implications for production and consumption by considering the way location, stamp type, and fabric of the brick intersect. I looked at every combination for every site in the South Etruria Survey, having done archaeometric analysis on the same collection to the point where I could say that ‘these two bricks share the same fabric’.

Here is the table (where 1 implies same, 0 implies different):

Combination Findspot Stamp Fabric
A 1 1 1
B 1 1 0
C 1 0 0
D 1 0 1
E 0 0 0
F 0 0 1
G 0 1 1
H 0 1 0

Which we may collapse, as some of these become functionally the same.
1. A&G: imply a common origin and distribution
2. B&H: imply a geographically dispersed production
3. D&F: imply a single clay source exploited by different figlinae.

Finally, there is situation ‘C’, which suggests that builders at a site had access to a variety of sources.

Since we can closely date many of the bricks, we can look at how these modes play out over time (remembering that this is a tally of relationships):

Julio-Claudian Flavian Nerva-Hadrian Antoninus Pius-Commodus Severan Late Total
Mode 1 (A/G) 10 2 2 6 2 22
Mode 2 (B/H) 54 14 4 4 6 82
Mode 3 (D/F) 13 13 7 7 2 42
Combination C (“Consumer Choice”) 17 7 6 2 1 2 35
Total 94 36 19 19 11 2 181

We can graph that:

After Fig. 4.2, Graham 2006 Ex Figlinis

Thus, by simply categorizing the relationships and graphing their evolution over time, we develop a complex view of the changing dynamics of production and consumption in the Tiber Valley.

However, we can also represent these relationships as a multi-modal network graph. The nodes in this graph are bricks and locations. Some of those locations, like the findspots, are known in space. Other locations, like the precise x,y, of the clay bodies is unknown or can be suggested to be roughly in a particular area. I create a directed graph where clay sources are ‘source’ and bricks are ‘target’, and where bricks are ‘source’ to findspots. I also categorize the linkages according to the relationships determined above. The result is below (Fig 1), where colours = communities and size = betweeness:

Fig 1. All relationships in evidence in Tiber Valley Roman Bricks

all relationships in tv-bricks  (zoomable pdf of the png image)

One can then filter this graph, looking for instance for relationships of type B (Fig 2). What I would really like is to be able to lay these graphs out so that my findspots and claysources can be pinned to geographic space, while allowing the bricks themselves and their relationships to float freely. This would require a layout plugin that understood how to apply a different layout algorithm to nodes with null latitude/longitude values. That at the moment is rather beyond me.

Fig 2. Type B relationships

Finally, one can collapse this graph into its component communities, and look at those relationships. In this PDF you can zoom in and see not just the bricks, but also the kinds of relationships, and the clay body that seems to go with this group (B6, somewhere in the neiborhood of Fiano Romano-ish).

Fig 3. Community 1

Here we collapse the communities into single nodes, and run betweeness on this graph (fig 4). We end up with community 6 being ‘most between’; this community seems to source its clay from somewhere in the upper Sabina in the Forum Novum area (see chapter 3 of Ex Figlinis). Next most between is community 1, which sources in South Etruria in the Fiano Romano area and in the lower Sabina. Then we have community 8, which is pulling from either side of the Tiber in the lower reaches of South Etruria and the Sabina.

Community 6 has within it two sites and six bricks -TVP3592 and TVP2304; bricks se18, se13, se116, se104, se156, and se152. These bricks are first century bricks from the Tonneiana/Viccianae, PL (which might mean ‘Portus Licini’), someone named CASPR and a Q. Sulpicius Sabinus. What does ‘betweeness’ mean in this context here? Obviously, a question with a lot riding on it, but off the top of my head, if the association of PL with Portus Licini, which was a known brick warehouse is suggestive.

Community 1 is a much more complicated group with six sites and 15 brickstamps. Figlinae mentioned are Tonneina/Viccianae again, some Domitianae Veteres, an explicit Portus Licini stamp, a few tied to the Domitiana family, some late ‘officinae’ stamps (including ‘Boconiana’, which implicitly suggests the town of Bocignano).

There is directionality implicit in this diagram, but I’m not entirely sure yet what to make of it.

Fig 4. Communities in Tiber Valley Brick

The Tiber Valley Brick Industry as Two Mode Network

From the first to third centuries AD, the brick manufactories in the Tiber Valley around Rome made millions of bricks. A proportion of these carry makers’ marks, with many different dimensions of information – the brick maker; the estate; the landowner; the year; and various decorative features which may or may not be signficant. I worked on this material for my PhD thesis (which these days makes me cringe when I look at it). I’ve been reconsidering that work lately, especially in light of recent work by Malkin, Knappet, Brughmans, and of course the interest in digital humanities in networks more generally.

I discovered on an old and nearly defunct laptop a file I’d created as part of my PhD work, which listed the text of every stamp in the CIL XV.1, and gave the original neo-latin description of the figurative device on these stamps. (CIL is written in 19th century academic Latin). So, I decided to see what I could find if I translated this into a Gephi two-mode network, and ran some descriptive statistics on it – how I wish I’d had these tools a decade ago!

As I wrote in 2002 there are 46 different types of signa, across 428 types of stamp, totalling about 3000 examples listed in CIL. The file I’m working here is a bit more messy, with many variants of the typical motifs. So, what do we have? It’s a two mode network, which can be filtered into a single giant component of 570 nodes and 608 edges connecting it. For the brick afficonados amongst you, you can pick out the big name figlinae like Terentianae or Domitianae, and see how they interconnect based on shared useage of signa. If you calculate modularity on this network, you get 18 very strong sub-communities. That jives very nicely with the archaeometry I did on the brick, finding about 12 groups in a cluster analysis using Ward’s Method on the XRD results from the South Etruria Survey collection of bricks.

Next thing: split this graph up by time, space, clay sources, and usage patterns at sites.

Here’s the pdf version with labels. Keep in mind, this is rather rough.

Towards the computational study of the Roman economy: draft

I’m contributing to a volume on  ‘Land and Natural Resources in the Roman World’. Below is my draft, on which I welcome comments and questions.

Towards the computational study of the Roman economy

Shawn Graham, Carleton University, Ottawa Canada

“Economies are complicated systems encompassing micro behaviours, interaction patterns, and global regularities. Whether partial or general in scope, studies of economic systems must consider how to handle difficult real-world aspects such as asymmetric information, imperfect competition, strategic interaction, collective learning, and the possibility of multiple equilibria. Recent advances in analytical and computational tools are permitting new approaches to the quantitative study of these aspects. One such approach is Agent-based Computational Economics (ACE), the computational study of economic processes modelled as dynamic systems of interacting agents.”[1]

Most models of the Roman economy do not take into account this idea that micro-behaviours, feedback, and local interaction provide the circumstances out of which emerge those larger issues about which we are typically concerned.[2]  Before we can ask questions about growth, or market integration, or the degree to which Rome was ‘primitive’ versus ‘modern’, we have to focus on individual decision making. In this paper, I outline an agenda for how we might be able to do this. It comes down to this: we have to draw out and understand networks of individuals at all geographical scales and then use those networks as the substrate for computationally simulating individuals’ economic activities. Thus, archaeology and ancient literature become united through computation.

It is not necessary, I think, to rehash the historiography of studies of the Roman economy; the broad outlines of the debate are well known.[3] What I find exciting though is the emergence of the New Institutional Economics of Douglass North[4]. These works all draw attention to ideas around the consequences of individual decision making from incomplete knowledge, of how ‘good enough’ or satisficing decisions push actors and economics towards path-dependence (where because of past decisions one is locked into a particular mode).[5] The idea that network relationships (and the institutions that emerge to promote these) are the mechanism through which ancient economies deal with incomplete knowledge[6]  is a powerful one because we can find and outline the traces of these networks through archaeology.

Bang takes this idea further and through cross-cultural comparison with Mughal India pushes our attention to the bazaar: “a stable and complex business environment characterised by uncertainty, unpredictability and local segmentation of markets’.[7] Bang shows how the bazaar helped shield the individual from (while at the same time encouraging) fragmented information and instability. For Bang, one of the key mechanisms for exploring and understanding a bazaar-like economy is social networks, the way they form, and how information flows through them.[8] We need to focus on the social differences between actors in a market situation.[9]

Even the Emperor can be incorporated into such a perspective. In considering the Emperor as both the embodiment of the state and a private person intervening in its economy as one more player amongst other private citizens, Lo Cascio draws attention to the interplay between the Emperor’s euergetism and private markets, that he solved the problem of feeding Rome “by leaving unchanged the market mechanisms at work” and by “establishing and enforcing the rules of the game”.[10] He does so to ensure that individuals neither engage in rampant speculation to force up prices, but also to ensure that prices do not become set to low, that is, to ensure an adequate profit.[11] The success or failure of the Emperor to do this should depend on his position in the network.[12]

This is a very complex view, too complex for any one individual to hold in one’s head and to be able to understand the non-linear outcomes of so many interacting parts.[13] We need the computer. To understand the Roman economy, we need to simulate it. The best way to do this is to ground our simulation in what our evidence actually gives us: the actions of individuals in the past, whether we find that evidence in the archaeology or the ancient literature. We have to build up from individuals before we can begin to understand what ‘the Roman economy’ might actually mean.

Roman Economics Needs to be Networked

Networks and network analysis are currently in vogue in many areas of research. Tom Brughmans has neatly summarised the historiography and main issues surrounding networks and network analyses as applied to archaeological research.[14] In essence, he argues that all of the varied approaches to network analyses are united by the idea that networks are ubiquitous and influence decision making; through networks both tangible objects and intangible influences spread (and in turn, promote or hinder network growth). Through studying networks, we are able to bridge the study of parts through reductionism to the emergent whole.[15] The methodological advantages enumerated by Carl Knappett[16]  can be tied explicitly to the advantages that the New Institutional Economics bring. A network perspective forces one to consider relations between entities; they are spatial whether considered in social or physical terms; they work across scales; they can incorporate people, objects, and time.

Schortman and Ashmore make the argument that a by-product of social networks is the emergence of power, through collaborative or cooperative action.[17]  Debates around structure and agency in the social sciences revolve around questions related to the concentration of power and the generation of hierarchies across multiple spatial scales. It is from this contending for assets that politics becomes necessary and structures for the same emerge. [18]

Achieving power over others involves monopolizing some aspect of the production, distribution or use of materials. On the other hand, others can contest this by using their own social networks to gather resources (whether material or social). In this way, action through social networks results in political structures which are themselves the aggregate of flows of materials and ideas through social networks.[19]

Perhaps the most well-known term in network studies is the idea of the ‘small-world’, first coined by Stanley Milgram.[20] A small-world is not just a metaphor, but rather a precise concept in network terms, where a randomly connected network with mostly local connections has a few long distance connections which allow the entire network to be spanned in only a few steps.[21] This is a crucial concept and one we should look for archaeologically or as a by-product of our models. Brughmans writes

“   …   Such a specific topology has direct implications for the processes underlying it, like the transportation of materials, the spread of religious ideas or the enforcement of political power. These processes would largely take place between the highly connected nodes [actors of whatever kind] and they would only reach the larger number of less connected nodes through these [linkages]. In a small-world network, on the other hand, nodes within the same small-world are more often directly connected to each other, while only processes involving other small-worlds (e.g. long-distance trade) would go over the bridging nodes.”[22]

How can we draw out networks from archaeological materials?  Objects carry the resonances of who make them. They mark out membership and social identity. “Exchanging these items thereby manifests and extends crucial social linkages… Networks, therefore, come alive, in part, through the transfer of items that partake of their members’ social essences”.[23] Objects therefore could be taken as proxies for social actors.[24] Artefacts are the result of human, individual, decisions. They influenced other individuals at the time through their complex resonances of thing and place and object life-history. By considering artefacts and their relationships explicitly in terms of social network analysis, we reconnect with the individual in the past, and we obtain a perspective that allows us to see what kinds of actions were possible in the past, patterns of agency and structure, that those actors themselves could not see.

Anytime one can discern a relationship, a network is possible. Fiona Coward reminds us that the archaeological record is not a passive by-product but rather is in fact social relationships: “The patterning of material culture is a direct result of the social relationships between individuals and groups in which these objects were caught up. A network perspective provides a much more realistic picture, not only of objective sociality, but also potentially of individuals’ subjective experience of their worlds”.[25]

But as Scott Weingart warns us, we also have to take into account the dangers of methodology appropriation.[26] Network analysis comes to us from graph theory, from statistics, from computer science. The methods, philosophies and concerns of those disciplines are not necessarily congruent with archaeology. Brughman’s recent work will go a long way to addressing the potentials and perils of drawing out networks from archaeological materials.[27]

Economic Simulation

If we can draw out networks from the ancient literature or the archaeological record, what then?  We re-animate these networks using agent based (or ‘individual based’) modelling. This is a technique where a population of autonomous, heterogeneous ‘agents’ are created within a computerized environment. They are given rules of behaviour which they implement given a particular situation (whether when interacting with other agents, or with the environment). They are goal-oriented and to a degree, self-aware. That is, if we are interested in something called ‘the Roman economy’, we simulate the agents that compose that economy and their behaviours: not the economy itself (in contrast to traditional economic models with their equations and ‘rationalizing’ assumptions; we simulate at one level of complexity below our ‘target’.[28]

Lea Tesfatsion discusses a number of ways in which ‘agent-based computational economics’ can be used, and classifies different studies according to their objectives. The one which should concern us here is what she calls, ‘qualitative insight and theory generation’: “how can economic systems be more fully understood through a systematic examination of their potential dynamical behaviours under alternatively specified initial conditions?”[29]  A key feature of agent modelling that differentiates it from other approaches is that once the starting conditions are specified, all subsequent events are driven by agent interactions. The researcher then is not so concerned with the final results of the simulation as its evolution over time, its history. Thus, the focus is on the process.[30] That is to say, we study the model’s history to generate insight into real history.

Research into agent based models of economies has found that the topology of interactions matters. It is not just the pattern of social ties that matters,[31] but also the environment in which these actions take place.[32] We can explore the environmental aspect by setting the simulation on top of a cellular-automaton, a chess-board like arrangement of squares, where each square represents a unit of land and its holdings, and which responds to rules about growth, climate, geology, and so on. In many ways, a cellular automaton represents a dynamic geographic information system and which could obviously be drawn from archaeological GIS. Combined with an agent-based model representing the decision making agents, we then have a powerful tool.

A number of researchers have applied ABM and cellular automata to studying problems of structural change in agricultural societies, exploring everything from government intervention,[33] to the diffusion of innovation in agriculture,[34] to the beginning of European-style agriculture in Indiana in the early 1800s.[35] These studies point to ways in which economic path-dependence emerges, under what circumstances innovation may diffuse, and decision making processes at different spatial scales.

Archaeology and Simulation

Simulation has a long history in archaeology.[36] Agent based modelling is an outgrowth from several different fields, primarily game theory and complex systems studies.[37] John Barret, in considering the relationship between agents and society, drew attention to how agents both form and are constrained by, the social structures that emerge from their interactions. He argued therefore that the level of ‘society’ should not form the basic unit of archaeological analysis, but rather the individual.[38] It is social learning that creates a society (or an economy, for that matter).

There are now many agent-based models of past societies.[39] Many of these are quite complicated, with many moving parts, which can make it difficult to understand what the model results may actually be telling us. I advocate instead for extremely simple models, exploring only a limited aspect of the phenomena in which we are interested in. After building a series of these, we can consider their results in aggregate.

Building a Simulation

How do we translate our arguments over the Roman economy into an agent framework? Tesfatsion suggests a four-step method for recasting that understanding in a way that can be modelled and explored:

“• Select as your benchmark case an equilibrium modelling of an economy from the economic literature that is clearly and completely presented and that addresses some issue you care about.

• Remove from this economic model every assumption that entails the external imposition of an equilibrium condition (e.g., market clearing assumptions, correct expectations assumptions, and so forth).

• Dynamically complete the economic model by the introduction of production, pricing, and trade processes driven solely by interactions among the agents actually residing within the model. These procurement processes should be both feasible for the agents to carry out under realistic information limitations and appropriate for the types of goods, services, and financial assets that the agents produce and exchange.

• Define an “equilibrium” for the resulting dynamically complete economic model.”[40]

Tesfatsion remarks that when she tries this exercise with economics students, they find it difficult to understand the economy at this level working with individual agents. The challenge of this method is that it foregrounds survival: that the needs of subsistence, of surviving over time (death is always a possibility in these models) are the bedrock on which everything else is based.

Thus, we build our rule-sets that our computational agents will follow from our understanding of how an individual [Roman; collegium; military unit; family; city] acts in particular situations. While every agent might have the same suite of variables, each agent is heterogeneous. Its particular combination of variables is unique. We might all be playing basketball by the same rules, but my abilities are different than yours. We model the appropriate underlying environment. We specify the initial starting conditions, and then set the simulation in motion. We run the simulation over and over again to explore the complete behaviour space for all of the particular starting conditions, to see what emerges when and how.

To know that we have found something new about the Roman economy from such a process, I would suggest that we ground one of the model’s behaviours in social networks. There are various ways this might be accomplished. We might draw from an understanding of how patronage worked in the Roman world.[41] Or, we could draw from Bang’s arguments about Rome-as-bazaar.[42] Then, when the simulation has run its course, we can measure the emergent network and compare its features to real networks known from the archaeology. When the two correspond, then the model settings for that particular run have something important to tell us about ancient society. Alternatively, we could reverse that and specify networks found in the archaeology as our starting point. Do the model outcomes, when set from a starting point known archaeologically, make any sense according to what we believe to have been true about the Roman economy? If not, then perhaps our rule sets are flawed.

Doran  underlines some of the main impediments to simulation building amongst archaeologists,  limitations in computing power, and the disciplinary boundaries that create barriers to knowledge building.[43] As Doran points out, the computing power issue is largely no longer an issue. Quite complicated simulations may be built on a desktop computer without too much trouble in terms of hardware. The second problem is more difficult. One way we can break those boundaries down may lie in the choice of simulation environment: the simpler and more intuitive the framework, the easier to use, develop, and communicate results. In my own work, I use the open-source Netlogo modelling environment, which I recommend to anyone interested in exploring the possibilities of this approach.[44] While Netlogo has its genesis in efforts to educated school children in complex systems thinking, it is now in its fifth major release and is quite powerful. The learning curve is not overly steep, and much can be accomplished through tweaking the many models that come pre-packaged with the software (more on this below).

A simulation is an argument in computer code about the way the world works, and so represents a kind of ‘procedural rhetoric’.[45] It is object-oriented, meaning that each individual behaviour exists as its own object. One then arranges the objects (which can be conditional) in the order the agents should carry them out, given a particular position in the environment or social position vis-à-vis other agents. One can have quite sophisticated simulations running after a day of working through the included tutorials. This should not be taken as a sign that Netlogo is simplistic. Quite powerful models have been built in it, including modelling the emergence of cities in the third millennium BC.[46]

An excellent place to begin is with the included model, ‘Wealth Distribution’.[47] In this model, a ‘world’ is simulated where grain is distributed randomly. In some places it grows thick on the ground; in other places it is sparse. A population of agents are introduced to this world. Their goal is to find enough grain to keep living, and to reproduce when conditions are right. The agents have ‘vision’ or knowledge of the world, to differing degrees. They have ‘metabolism’, or a preset amount of food they must consume with each time-step or they shall die (which again varies by individual). The amount of food collected above this metabolic rate becomes ‘wealth’. When this simulation is run, it becomes apparent that the differential distribution of resources in this world is sufficient in itself to create a partition of the world into classes where there are a few extremely wealthy individuals and a vast mass of others who are in constant danger of ‘dying’ from not having enough food.

We could then extend this model to represent something of the Roman world. We could give the agents a way of looking for help, of becoming a client of someone a bit wealthier than themselves. In return we could imagine that these ‘clients’ could offer support to their patron in turn. Perhaps the number of followers, and their relative ‘wealth’, could be translated into a score for ‘prestige’ which in turn affects the ‘patron’s’ ability to extract wealth from the world. What kind of artificial society results? In my ‘Patronworld’ model,[48] which had its inspiration in the Wealth Distribution model, chains of connected individuals (that is, networks) do emerge from this dynamic, but they are fragile. One result seems to be that extremely high levels of gift-giving seem to go hand in hand with network collapse. It seems to do this by destabilizing the networks: too many people outside particular chains of patrons-clients are shut out of the system.[49] Given the competitive building that characterizes the late Republic, this result is intriguing.[50] It is a very simple model, to be sure, but one that foregrounds an important element of new models of the Roman economy: networks and social life. This kind of modelling also has the virtue that if one disagrees with the assumptions of the model, the code can be easily modified and adapted. In this way, model building is not an end point of research but rather a first step of a larger conversation (my own models may be found at the digital data repository Figshare.com and I welcome their use, adaptation, and improvement).

Moving forward

We need more and better networks drawn from historical and archaeological data. It is not enough, in contrast to Malkin,[51] to use ‘networks’ as simple metaphor. Particular network topologies have different implications for the actors which make them up. If we are to make progress on the Roman economy, we need to explore the multiple networks of individuals and objects at multiple social and spatial scales. We need to turn our archaeological geographic information systems into computing environments in which agents can interact: at a stroke, we will have unified landscape archaeology, ancient history, and the study of ancient economics. Once we have this data, we can take our current understandings of the ancient economy, whether ’consumer city’, ‘primitive’; ‘modernising’, ‘bazaar’, NIE, or something else and translate them into an agent based simulation. If we can generate analogous networks to the ones we know archaeologically, then we might just have the wherewithal to argue that we have a model that tells us something useful, something new, about the past.

References

Agar, M. (2003).  ‘My kingdom for a function: modelling misadventures of the innumerate’, Journal of Artificial Societies and Social Simulation 6.3. http://jasss.soc.surrey.ac.uk/6/3/8.html (accessed 28/05/2012).

Bang, P. (2006). ‘Imperial Bazaar:  towards a comparative understanding of markets in the Roman Empire’, in P. Bang, M. Ikeguchi, H. Ziche (eds.), Ancient Economies, Modern Methodologies: Archaeology, Comparative History, Models and Institutions. Bari, 51-88.

Bang, P. (2009). ‘The ancient economy and New Institutional Economics’, Journal of Roman Studies 99: 194-206.

Barabási, A.-L. and R. Albert. (1999). ‘Emergence of scaling in random networks’, Science 268: 509-12.

Barrett, J. (2000). ‘A thesis on agency’, in M. Dobres and J. Robb (eds.), Agency in Archaeology. New York, 61-8.

Barrett, J. (2001). ‘Agency, the duality of structure, and the problem of the archaeological record’, in I. Hodder (ed.), Archaeological Theory Today. Cambridge, 141-64.

Bentley, R. and H. Maschner (2003). ‘Preface: considering complexity theory in archaeology’, in R. Bentley and H. Maschner (eds.), Complex Systems and Archaeology. Salt Lake City, 1-8.

Berger, T. (2001). ‘Agent-based spatial models applied to agriculture: a simulation tool for technology diffusion, resource use changes, and policy analysis’, Agricultural Economics 25: 245-260.

Bogost, I. (2007). Persuasive Games. Cambridge, MA.

Bourdieu, P. (1977). Outline of a theory of practice (trans. R. Nice). Cambridge.

Brughmans, T. (2010). ‘Connecting the dots: towards archaeological network analysis’, Oxford Journal of Archaeology 29.3: 277-303.

Brughmans, T. (2012). ‘Thinking through networks: a review of formal network methods in archaeology’, Journal of Archaeological Method and Theory 19.2 Online version: DOI: 10.1007/s10816-012-9133-8 http://www.springerlink.com/index/10.1007/s10816-012-9133-8 (accessed 28/05/12).

Buchanan, M. (2002). Nexus: Small Worlds and the Groundbreaking Science of Networks. New York.

Clarke, D. (1968). Analytical Archaeology. London.

Clarke, D. (1972). Models in Archaeology. London.

Costopoulos, A and M. Lake. (2010). (eds.), Simulating Change: Archaeology Into the Twenty-First Century. Salt Lake City.

Coward, F. (2010). ‘Small worlds, material culture and Near Eastern social networks’, Proceedings of the British Academy 158, 449-479. http://www.fcoward.co.uk/Cowardsmallworlds.pdf (accessed 28/05/12).

Davis, J. (1998). ‘Ancient economies: models and muddles’, in H. Parkins and C. Smith (eds.), Trade, Traders and the Ancient City. London, 225-56.

Davis, J. (2005) ‘Linear and non-linear flow models for ancient economies’, in J. Manning and I. Morris (eds.), The Ancient Economy: Evidence and Models. Stanford, 127-56.

Dean, J. S., Gumerman, G. J., Epstein, J. M., Axtell, R. L., Swedlund, A. C., Parker, M. T., and McCarroll, S. (2006) ‘Understanding Anasazi culture change through agent-based modeling.’ in Epstein, J. (ed.), Generative Social Science: Studies in Agent-Based Computational Modeling  Princeton, 90–116.

Deffuant, G., S. Huet, J.P. Bousset, J. Henriot, G. Amon, G. Weisbuch. (2002). ‘Agent based simulation of organic farming conversion in Allier Département’, in M. Janssen (ed.), Complexity and Ecosystem Management: The Theory and Practice of Multi-Agent Systems. Cheltenham, 158–187.

DeLaine, J. (2002). ‘Building Activity in Ostia in the Second Century AD’ in C. Bruun and A. Gallina Zevi (eds.). Ostia e Portus Nelle Loro Relzaioni con Roma (Acta Instituti Romanae Finlandiae 27). Rome, 41-101.

Dibble, C. (2006). ‘Computational laboratories for spatial agent-based models’, in L. Tesfatsion and K. Judd (eds.), Handbook of Computational Economics, Vol. 2: Agent-Based Computational Economics. Amsterdam, 1511-1550.

Dobres, M. and J. Robb (2000). Agency in Archaeology. New York.

Doran, J. (2011). Review of A. Costopoulos and M. Lake, Simulating Change: Archaeology into the Twenty-First Century Salt Lake City. http://jasss.soc.surrey.ac.uk/14/4/reviews/2.html

Dornan, J. (2002). ‘Agency and archaeology: past, present, and future directions’, Journal of Archaeological Method and Theory 9: 303-29.

Evans, T.P., and H. Kelley. (2004). ‘Multiscale analysis of a household level agent-based model of land-cover change’,  Journal of Environmental Management 72.1, 57–72.

Frier, B. and D. Kehoe. (2007). ‘Law and economic institutions’, in W. Scheidel, I. Morris, R. Saller (eds.), The Cambridge Economic History of the Greco-Roman World. Cambridge, 113-143.

Gardner, A. (2007). An archaeology of identity: soldiers and society in late Roman Britain. Walnut Creek.

Giddens, A. (1984). The constitution of society: outline of the theory of structuration. Berkeley.

Gilbert, N., and K. Troitzsch. (2005). Simulation for the Social Scientist. Second edition. Maidenhead, Berkshire.

Graham, S. (2006a). ‘Networks, Agent-Based Modeling, and the Antonine Itineraries’, The Journal of Mediterranean Archaeology 19.1: 45-64.

Graham, S. (2006b). Ex Figlinis: The Complex Dynamics of the Roman Brick Industry in the TiberValley during the 1st to 3rd Centuries AD. British Archaeological Reports, International Series 1486: Oxford.

Graham, S. (2009). ‘BehaviourSpace: Simulating Roman Social Life and Civil Violence’, Digital Studies/ Le champ numérique, 1(2). http://www.digitalstudies.org/ojs/index.php/digital_studies/article/view/172/214 (accessed 29/5/12)

Graham, S. and G. Ruffini. (2007). ‘Network Analysis and Greco-Roman Prosopography’, in K.S.B. Keats-Rohan (ed.), Prosopography Approaches and Applications: A Handbook. Oxford: 325-36.

Graham, S. and J. Steiner. (2008). ‘Travellersim: Growing Settlement Structures and Territories with Agent-Based Modelling’, in J. Clark and E. Hagemeister (eds.), Digital Discovery: Exploring New Frontiers in Human Heritage. CAA 2006. Computer Applications and Quantitative Methods in Archaeology. Proceedings of the 34th Conference, Fargo, United States, April 2006. Budapest: 57-67.

Hoffmann, M., H. Kelley, and T. Evans. (2002). ‘Simulating land-cover change in south-central Indiana: an agent-based model of deforestation and afforestation’, in M. Janssen (ed.), Complexity and Ecosystem Management: The Theory and Practice of Multi-Agent Systems. Cheltenham, 218–247.

Jansenn, M. and E. Ostrom (2006). ‘Governing Social-Ecological Systems’ in L. Tesfatsion and K. Judd (eds.), Handbook of Computational Economics, Vol. 2: Agent-Based Computational Economics. Amsterdam, 1466-1510.

Knappett, C. (2011). An archaeology of interaction. Network perspectives on material culture and society. Oxford.

Kohler, T., G. Gumerman, and R. Reynolds. (2005). ‘Simulating Ancient Societies’, Scientific American 293.1: 76-84.

Lehner, M. (2000). ‘Fractal house of pharaoh: ancient Egypt as a complex adaptive system, a trial formulation’ in Kohler, T. A. and Gumerman, G. J. (eds.), Dynamics in Human and Primate Societies: Agent-Based Modeling of Social and Spatial Processes . Oxford, 275–353.

Lo Cascio, E. (2006). ‘The role of the state in the Roman economy: making use of the New Institutional Economics’, in P. Bang, M. Ikeguchi, H. Ziche (eds.), Ancient Economies, Modern Methodologies: Archaeology, Comparative History, Models and Institutions. Bari, 215-236.

Malkin, I. (2011). A Small Greek World: Networks in the Ancient Mediterranean. Oxford.

Manning, J. And I. Morris. (2005). The Ancient Economy: Evidence and Models. Stanford.

Milgram, S. (1967). ‘The small world problem’, Psychology Today. 2:60–67.

North, D. (1990). Institutions, Institutional Change and Economic Performance. Cambridge.

Ourednik, A. and P. Dessemontet. (2007). ‘Interaction maximization and the observed distribution of urban populations: an agent based model of humanity’s metric condition’. http://ourednik.info/urbanization_mc (accessed 28/5/12).

Premo, L. S. (2006) Agent-based models as behavioral laboratories for evolutionary anthropological research. Arizona Anthropologist 17:91-113.

Scheidel, W. (2010). ‘Approaching the Roman economy’ version 1.0. http://www.princeton.edu/~pswpc/pdfs/scheidel/091007.pdf (accessed 28/5/12).

Scheidel, W. and S. von Reden (2002). The Ancient Economy. Edinburgh Readings on the Ancient World. New York.

Schortman, E. and W. Ashmore. (2012). ‘History, networks, and the quest for power: ancient political competition in the Lower Motagua Valley, Guatemala’, Journal of the Royal Anthropological Institute 18.1: 1-21.

Symons, S. and D. Raine. (2008) ‘Agent-Based Models Of Ancient Egypt’, in N. Strudwick (ed.), Proceedings of Informatique et Égyptologie. Piscataway, NJ. http://www.physics.le.ac.uk/ComplexSystems/papers/AgentBasedModelsEgypt2008.pdf (accessed 28/5/12).

Tesfatsion, L. (2006). ‘Agent-based computational economics: a constructive approach to economic theory’, in L. Tesfatsion and K. Judd (eds.), Handbook of Computational Economics, Vol. 2: Agent-Based Computational Economics. Amsterdam, 831-880.

Vriend, N. (2006). ‘ACE models of endogenous interactions’, in L. Tesfatsion and K. Judd (eds.), Handbook of Computational Economics, Vol. 2: Agent-Based Computational Economics. Amsterdam, 1047-1080.

Wallace-Hadrill, A. (1989) (ed.) Patronage in Ancient Society. London.

Watts, D. (1999). Small Worlds. The Dynamics of Network Between Order and Randomness Princeton.

Watts , D., and S. Strogatz. (1998). ‘Collective dynamics of ‘small-world’ networks’, Nature 393: 440-42.

Weingart, S. (2012). ‘Demystifying networks’, the scottbot irregular http://www.scottbot.net/HIAL/?p=6279 (accessed 28/5/12).

Wilensky, U., and M. Resnick. (1998). ‘Thinking in Levels: A Dynamic Systems Approach to Making Sense of the World’ Journal of Science Education and Technology 8.1: 3-18.

Wilensky, U. (1999). Netlogo . Evanston, IL. http://ccl.northwestern.edu/netlogo (accessed 28/5/12).

Wilhite, A. (2006). ‘Economic activity on fixed networks’, in L. Tesfatsion and K. Judd (eds.), Handbook of Computational Economics, Vol. 2: Agent-Based Computational Economics. Amsterdam, 1013-1046.


[1] Tesfatsion 2006: 832.

[2] Indeed, many economic models fail to recognize that the spatial context of economic action undermines many of the basic assumptions of economic theory, Dibble 2006: 1515.

[3] The discussion is usefully treated in for example Scheidel 2010, Manning and Morris 2005, and Scheidel and von Reden 2002.

[4] North 1990, and its application to the ancient world in Frier and Kehoe 2007; Bang 2009; the various papers in Manning and Morris 2005; Scheidel’s forthcoming Cambridge Companion to the Roman Economy

[5] Frier and Kehoe 2007.

[6] Frier and Kehoe 2007: 119.

[7] Bang 2006: 79.

[8] Bang 2006: 80-4.

[9] Bang 2008: 197.

[10] Lo Cascio 2006: 225.

[11] Lo Cascio 2006: 231.

[12] Graham 2006b: 111-113.

[13] cf.  Davis’ diagrams in 2005; 1998.

[14] Brughmans 2012; 2010.

[15] Brughmans 2012; Bentley and Maschner 2003:1.

[16] Knappett 2011: 10.

[17] Schortman and Ashmore 2012: 3.

[18] Schortman and Ashmore 2012: 2-3 citing Barrett 2000, Bourdieu 1977, Dobres and Robb 2000; Dornan 2002; Garnder 2007; Giddens 1984.

[19] Schortman and Ashmore 2012: 3-4.

[20] Milgram 1967.

[21] Buchanan 2002; Watts and Strogatz 1998; Watts 1999.

[22] Brughmans 2010.

[23] Shortman and Ashmore 2012: 4.

[24] Graham and Ruffini 2007: 325-331.

[25] Coward 2010.

[26] Weingart 2012.

[27] Brughmans 2012; 2010.

[28] cf.  Wilensky and Resnick 1998: 4; Agar 2003; Gilbert and Troitzsch 2005: 199-202. Gilbert and Troitzsch is an excellent resource for learning how to build simulations.

[29] Tesfatsion 2006: 840.

[30] Tesfatsion 2006:843.

[31] Watts and Strogatz 1998; Barabasi and Albert 1999.

[32] Janssen and Ostrom 2005: 1496 citing Dibble 2006; Wilhite 2006; Vriend 2006.

[33] Berger 2001.

[34] Deffuant et al., 2002.

[35] Hoffmann et al 2002; Evans and Kelley 2004.

[36] Some of the earliest work being in Clarke 1968, 1972; see Graham 2006a: 53-4.

[37] The premier journal is the Journal of Artificial Societies and Social Simulation, jasss.soc.surrey.ac.uk.

[38] Barrett 2001:155.

[39] for instance Lehner 2000; Kohler, Gumerman and Reynolds 2005; Dean et al 2006; Graham 2006a; Premo 2006; Graham and Steiner 2008; Symons and Raine 2008; Graham 2009; Costopoulos and Lake 2010.

[40] Tesfatsion 2006: 852-3.

[41] cf. Wallace-Hadrill 1989.

[42] Bang 2006; 2008

[43] Doran 2011.

[44] Wilensky 1999.

[45] Bogost, 2007: 28-44.

[46] Ourednik and Dessemontet 2007.

[47] Wilensky, 1998.

[48] Published in Graham 2009.

[49] Graham 2009: 11.

[50]  cf.  DeLaine 2002 on patronage in building projects at Ostia.

[51] Malkin 2011: 18-9.


Converting 2 mode networks with Multimodal plugin for Gephi

Scott Weingart drew my attention this morning to a new plugin for Gephi by Jaroslav Kuchar that converts multimodal networks to one mode networks.

This plugin allows multimode networks projection. For example: you can project your bipartite (2-mode) graph to monopartite (one-mode) graph. The projection/transformation is based on the matrix multiplication approach and allows different types of transformations. Not only bipartite graphs. The limitation is matrix multiplication – large matrix multiplication takes time and memory.

After some playing around, and some emails & tweets with Scott, we determined that it does not seem to work at the moment for directed graphs. But if you’ve got a bimodal undirected graph, it works very well indeed! It does require some massaging though. I assume you can already download and install the plugin.

1. Make sure your initial csv file with your data has a column called ‘type’. Fill that column with ‘undirected’. The plugin doesn’t work correctly with directed graphs.

2. Then, once your csv file is imported, create a new column on the nodes table, call it ‘node-type’ – here you specify what the thing is. Fill it up accordingly. (cheese, crackers, for instance).

3. I thank Scott for talking me through this step. First, save your network; this next step will irrevocably change your data. Click ‘load attributes’. Under attribute type, select your column you created for step 2. Then, for left matrix, select select Cheese – Crackers; for right matrix, select Crackers – Cheese. Hit ‘run’. This gets you a new Cheese-Cheese network (select the inverse to get a crackers – crackers network).  You can then remove any isolates or dangly bits by ticking ‘remove edges’ or ‘remove nodes’ as appropriate.

4. Save your new 1 mode network. Go back to the beginning to create the other 1 mode network.

Review of Malkin, “A Small Greek World: Networks in the Ancient Mediterranean”

A Small Greek World

Networks in the Ancient Mediterranean

Irad Malkin

OUP USA
304 pages | 21 illustrations | 235x156mm
978-0-19-973481-8 | Hardback | 24 November 2011
Price: £40.00
http://ukcatalogue.oup.com/product/9780199734818.do

I was excited to obtain this book.

Unfortunately, this is a book about social network analysis in antiquity that does not, in point of fact, contain any social network analysis. Rather, Malkin uses concepts drawn from networks and theories of evolving networks as metaphors to reframe centre-periphery arguments about the emergence of the Greek world around the Mediterranean as a ‘small world’.

There is much that is good with this book, in terms of its description of colonization and the emergence of the Greek world. He offers up a theory of ‘backwards-propagation’ to explain how the colonies could often be more Greek than Greece. As a Canadian educated in the UK, I know that this is a very real – and timeless- phenomenon; it is indeed a useful concept to bring into the discussion. However, Malkin need never have invoked any network theories in order to use that concept. Peregrine Horden and Nicholas Purcell’s The Corrupting Sea achieved much by focusing on the idea of connectivity of micro-regions without invoking social network analysis; this work is essentially a deeper exploration of that idea.

The broad strokes of Greek colonization are well known; there is ample material there for network studies of many different kinds (including work that seeks to generate likely networks, for example as in Rihll & Wilson 1991, the work of Tim Evans, Ray Rivers, and Carl Knappett, my own Travellersim). Malkin lays out the groundwork of the relevant concepts in chapter 1. But on p18 and 19 one reads,

Graphic illustrations of wide-ranging Mediterranean networks in the form of connective graphs usually prove to be unhelpful. Two-dimensional representations of connectivity mostly turn out to be messy “spaghetti monsters” with very long verbal explanations that are needed to accompany them… I have opted for the larger canvass of what seems to me highly probable at the risk of not presenting statistics and formulae that I am incapable of offering due to the state of our sources of knowledge”

Malkin takes pains on p16 to distinguish between ordinary (whatever that may mean) ‘networks’ and the ‘networks’ of network analysis, as if the two were distinguishable. They are not. The only difference is that in one we are using a metaphor, and in the other, we have taken pains to try to outline as fully as possible the connections relevant to the question we are asking, to understand the implications of the topology (yes, the statistics) for what they might mean for history. To invoke a small-world (a precise concept in network terms) without actually measuring to see if small world conditions are fulfilled does damage to the concept and to the analysis.

Elijah Meeks has recently written about developing conventions for the representation of network data, drawing on the long history of cartographic literacy. As far as Malkin’s critique of the visualization of networks go, it’s well founded: but the visualization has never been the endpoint, the raison d’etre, for the exercise. It’s the statistics. If you don’t know the shape of the thing, the important nodes (cities, individuals, extra-urban sanctuaries, what have you), how can you claim to be doing any sort of network analysis?

Scott Weingart has written about ‘halting conditions’, about knowing where to draw the linewhen your data are necessarily complex and in practice, infinite:

The humanities, well… we’re used to a tradition that involves very deep and particular reading. The tiniest stones of our studied objects do not go unturned. The idea that a first pass, an incomplete pass, can lead to anything at all, let alone analysis and release, is almost anathema to the traditional humanistic mindset.

Herein lies the problem of humanities big data. We’re trying to measure the length of a coastline by sitting on the beach with a ruler, rather flying over with a helicopter and a camera. And humanists know that, like the sandy coastline shifting with the tides, our data are constantly changing with each new context or interpretation. Cartographers are aware of this problem, too, but they’re still able to make fairly accurate maps.

It is not acceptable for ancient historians to bemoan the incompleteness of our sources – to use that as a crutch for not doing something – and then to go on and write another 224 pages. We’ve been studying the ancient world for several centuries now. Surely we’ve got enough material to be able to draw a line, to map something out, yes?

Should you buy this book? By all means, yes. Its roundup of the major themes and ideas in network studies in chapter 1 is valuable, and will no doubt be useful for those wishing to do formal network analysis, insofar as it establishes network theory in the broader classicist conversation. I’ve focused primarily on the first chapter in this review, since much of my teaching and research at the moment explicitly concerns network analysis in antiquity. The materials presented in subsequent chapters (Island Networking and Hellenic Convergence; Sicily and the Greeks; Herakles and Melqart; Networks and Middle Grounds in the Western Mediterranean; Cult and Identity in the Far West) do indeed move the conversation on from stale center-periphery models, and should be lauded. It would perhaps have been better though if it had not been framed in terms of a theoretical/methodological framework that is not, in fact, used.

[disclosure: I asked Oxford UP for a review copy when I saw it in the catalogue, in the hopes that I could use it as a text in an upcoming course on digital antiquity. I do not think I will be doing so, given my issues indicated here.]

Artefacts as Nodes in a Network II

Places connected to places by virtue of the use of the same stamped brick type, overlaid against latitude and longitude

4th – 6th century stamped bricks connected by virtue of use in the same location, around Constantinople

Places connected to places, by virtue of use of the same stamped 4th to 6th century bricks, around Constantinople

In these two one-mode networks (generated from Jonathan Bardill’s study of the brickstamps of Constantinople), colour indicates modularity class, while size indicates betweeness centrality. What network measures are most appropriate for understanding archaeological networks?

Converting 2-mode networks to 1-mode networks

April 4th: There is now a plugin for Gephi which will convert from multi-modal to 1-mode networks: https://gephi.org/plugins/multimode-networks-transformations/

Say you’re interested in patterns of communication between individuals who are members of multiple organizations (like for instance historical societies), or artefact types across multiple sites. You might like to map the network between these individuals and those organizations to understand something of how information flows in that world, how social norms permeate, or ideologies of consumption or display map across space (as Tom Brughmans does here and I’ve done in other places).

You’ll need Gephi and Sci2. Download and install these. Registration is required for Sci2. Any work you do with these tools, if published, requires the citation of the tool. Don’t forget!

1. Make a list. Every time you encounter an individual mentioned as a member of a group, write it out. Two columns: Source, Target. Shawn Graham, Carleton University. You might include a third column called ‘weight’ which gives some measure of the importance of that connection.  Why ‘source’, why ‘target’? Because we’re going to import that list into Gephi, and that’s how Gephi requires the information. However, when we do any sort of metrics, we’ll always treat this network as undirected; that is, we’re making no claim to know anything about the direction of the relationship (in a directed network, Alice’s connection to Bob is different than Bob’s connection to Alice). Save that list as a .csv file. If you graphed this right now, you’d have a network where there are two kinds of nodes; hence a two-mode network. Network statistics envision a network where the modes are all of the same kind, which is why we’re doing this tutorial.

2. Import the list into Gephi. Open Gephi, start a new project. Click on the ‘data laboratory’ lab. Under ‘data table’ click on ‘edges’ (this is important; if you click on ‘nodes’, this doesn’t work correctly). Click on ‘import spreadsheet’. Select your csv file, and make sure that ‘as table:’ is set to ‘edges table’. Click next. Click Finish.

3. Go to File ->> export ->> graph file. Save as file type .net (Pajek).

4. Open Sci2; click on File >> Load and select your .net file.

5. Click ‘data preparation’ >> extract reference co-occurence (bibliographic coupling) network. (see also 5.b, below under ‘variations’)

6. Click ‘preprocessing’ >> networks >> delete isolates.  You’ve now collapsed your two-mode network into a one mode network where your nodes under ‘target’ are now all connected to each other. If your source, target was ‘site’, ‘ware’, you’ve got a one mode network where wares are connected to each other by virtue of being listed at the same site, ie, the linkage implies the site. (If you’ve done step 5.b, your one mode network would be sites connected to each other by virtue of sharing the same wares; the linkage implies the ware.)

7. At this point, go to file >> view and your notepad application will open, displaying a table where each node in your network has its own unique id, and ‘label*string’, which is your original label. You’ll Save this from notepad as a txt file. You might call it ‘ware to ware index’ (following our example in 6).

8.This is where things get a bit tricky. Click ‘File’ >> ‘save’. Select ‘pajek .net’ as your file type.(see also 8.b below, under ‘variations’)

9. You can then go back to Gephi, start a new project, click ‘open’ and select the .net file you just created. Your one mode network will load up. HOWEVER Gephi won’t recognize the original node lables anymore. This is why you need the index you saved in step 7, so that when you run metrics on this one mode network, you’ll know that ‘node 342 is actually Stamped Brick CIL XV.1 841.d (for instance). (see also 9.b below, under ‘variations’)

Variations

5.b In step 5, you created a one mode network based on your ‘target’ column. To create a one mode network based on your ‘source’ column, click ‘data preparation’ >> extract document co-citation network. Resume at step 6.

8.b If you want to preserve the node labels in Gephi, instead of step 8.a, click on Visualization >> Networks >> GUESS. This is a small visualization tool (that allows you to do some network metrics; but if your network is v. big, > 1000 nodes, this might not be a good idea). In GUESS, click File >> export graph. Give it a file name that makes sense, and don’t forget to type in the extension .gdf; otherwise it won’t export. Go to step 9b.

9.b Go to Gephi, start a new project, click on ‘open’, and select the .gdf file you just created. Your node labels will now be present in the graph, and so you don’t need that index file you created. Perhaps a bug: In my experiments, node labels don’t seem to appear in the ‘graph overview’ pane, when working with the gdf file. Your experience might be different. However, they do appear when I export an image of the network, under ‘preview’ >> export.

Fin. Let me know if/how your experience differs, or if these steps require clarification.

Artefacts as Nodes in a Network

I’m working on a paper for a conference next month. In it, I consider artefact copresence at various sites as a means for generating networks, in an effort to get at some of the ideological or social frameworks underpinning the distribution of these networks. I’m looking at stamped brick from Constantinople. I create a list where each entry is a site and a single example of a stamped brick. These I can then visualize using Gephi, and using Sci2 I can convert the two mode network (brick – place) to two one mode networks (bricks – bricks, tied because they’re found at the same place; place – place, tied because they use the same bricks). When I have this data as 1-mode networks, I can then run network stats. Below is the one mode graph, showing places tied to other places, based on the distribution of over 2300 stamped bricks:

Topic Modeling With the JAVA GUI + Gephi

I’ve been having an interesting conversation with Ben Marwick, in the comments thread of my initial ‘Getting Started with Topic Modeling’ post. Ben pointed me to an interesting GUI for Mallet, which may be downloaded here. I’ve been trying it out this morning, and I like what I’m seeing. Topic modeling is becoming more and more popular amongst the Digital Humanities crowd. An interesting automated approach to generating networks of topics and ideas from texts is reported by Scott Weingart, using the writings of Newton.

While I have nothing near so polished available, the GUI for Mallet used with Gephi can do nearly the same thing. My body of data comes from Writing History in the Digital Age. An earlier experiment with the same data is recounted here. I re-ran the data using the GUI approach, and have to say, this is a much easier and accessible approach. Run the program; select the folder with your txt documents in it; select the target number of topics; select the appropriate language stopwords list if necessary; hit ‘train topics’. What is very neat about this program is how it presents its output in both html and csv.

So in the spirit of crowdsourcing, I’ve put the output files online, and haven’t tried to decide yet what the topics might mean. Instead, why don’t you view the files for yourself, and let’s identify the topics using the comments of this post?

I then took the CSV files, and got them ready for import into Gephi. Decide which two columns you’d like to represent as being connected, and prune away the extraneous data. I took the ‘topicsindocs.csv’ file, and pruned it so that each paragraph of each author is paired with its major topic. I stripped away the info about the paragraph itself, so that the resulting visualization is just authors to the topics they write about. In the screenshot below, you can see the open gephi file with my own ‘Wikiblitz’ article highlighted, and its connections.

What’s also interesting is when I ran the ‘modularity’ routine – identifying communities based on patterns of self-similarity of ties – only four communities emerged (albeit with a very low modularity measurement, 0.235, which suggests that these communities are all that strong). A natural grouping of the papers, perhaps? (by the way, here’s the pdf/svg file).