“How-to” to create and visually explore your first archaeological networks?
This very basic introduction to archaeological network creation will be predicated on the simple question “does entity A depend on entity B?” combined with the slightly more involved question of “how does entity A depend on entity B?”. Proving that things are absolutely and directly connected in the archaeological record is very difficult if not impossible. However, archaeologists have found many ways to discuss, attest and prove this in indirect ways. This “how-to” will therefore not only ask you to decide if dependencies are present or absent, but will challenge you to make clear how they are present or absent.
The idea of dependency is central to what a network is and in fact separates it from non-relational data, method and theories. If you would like to read more about how dependencies work in archaeology, I suggest you read Ian Hodder’s 2012 book, Entangled. If you do not have access to that book, you will be able to find the 2016 follow-up Studies in Human-Thing Entanglements that is available as an Open Access publication.
This “How-to” has two goals: (1) to allow more people to visualize and explore their archaeological data or theories as networks and (2) to structure these explorations and make the thinking behind the “deep structures” of archaeological network thinking more explicit.
We will use Microsoft Excel (or similar spreadsheet software) and visone. Visone is a java-based application, so your computer also needs to have an up-to-date installation of java (download here).
Please, note that this method will work best for relatively small networks as it is based on thinking the network through systematically as a matrix. This way of working is not recommended for larger networks. In fact, if you would like to model a large set of ideas or data as a network, database design and workflows will be very different from this “how-to”. If you would like to know more about this have a look at graph databases like neo4j.
Step 1
What network phenomenon would you like to explore? What is the network-based question you would like to visualize?
Many archaeological observable or relevant phenomena can be explored as a network and I have argued in my book that there is a strong “covert” tradition in archaeological network thinking, alongside more recent so-called “formal network approaches.” Please note, however that it does not necessarily always make sense to explore a phenomenon as a network. Before you start your work, always ask whether you could visualize or explore a phenomenon in a more effective way than through networks?
“What is network science?”, an editorial of the first issue of the “Network Science” journal is a non-technical and broad introduction to this issue. I recommend you read it before starting to create your first network. Collar and colleagues have also written a very informative editorial in 2015. It is a great intro and it comes with a glossary specifically relevant to archaeological networks. It is Open Access, so everybody interested in creating archaeological networks can and should read it.
Step 2
What are the relevant network entities (nodes) for exploring your phenomena.
Entities is a bit of a vague word, right? I have used it on purpose, since network entities or nodes can, in theory, be anything. Common ones in archaeology are sites, site loci, objects, iconographic motifs or (material) practices. It is important to realize that network methods and theories will not tell you how to define your entities, this has to arise from your own archaeological knowledge and theories.
The same applies to how you should choose which entities to include and which not. In other words, where you should draw the boundaries of your network. It is a common misconception that networks have no boundaries and “everything is connected.” Networks are not different than any other model: they are an abstraction of a phenomenon, rather than an accurate and “to scale” depiction of that phenomenon. When that is understood, let’s start the creation of the network.
- Open excel, you will see an empty spreadsheet. The first cell (A1) will remain unused. Start at A2 and list all your relevant entities in the first column. Save your file (as a .csv).
- Once you finished the list of all these entities, enter all of them in the same order in the top row, starting at B1.
- An easy way to do this is by copying the first column (A) and transpose this to the top row (right click, paste special, transpose).
- You have now created an empty matrix in which the first column and row show the names of what will become your nodes in the network. If we were to visualize this as a network at this moment, this would be a network in which there are as many nodes as entities on your list. Yet, since the matrix is empty, none of these entities would be connected. We will create ties between them in Step 3.
Step 3
How is entity A dependent on entity B?
This is where you will draw the dependencies or ties between your entities. Once again, network methods and theories will not and should not dictate how you consider your entities to be dependent on each other. This is why we will go through your list of entities, and consider for each entity in your list how it is dependent on every other entity in your list based on your data, your ideas, literature references, common logic. The important thing is to keep track of it in a separate spreadsheet, which I will refer to as the meta-structure. NB: in the process of creating your matrix, make sure to save both the matrix and the meta-structure file regularly.
- In the meta-structure spreadsheet write “Name”down in cell A1, in B1 “is dependent on” and C1 “Reason.” If you have multiple, separate lines of argumentation for the dependency or have additional info on the entity in cell A1, you can write this down in D1 and further.
- Move back to the matrix document. Start at the empty cell in B2, if the order is correct, your entity in the first column (A2) should be the same as your entity in the top row (B1). This is the matrix diagonal. Normally entities are not dependent on themselves, so this matrix diagonal will normally show 0’s. If you think there is a reason for an entity to be dependent on itself and that it would be important to visualize this, mark it “1” (this is called a loop).
- Move to the empty cell in C2. Ask yourself: “Is the entity in cell C1 dependent on the entity in cell A2?”. If it is, fill out this cell with a “1”, if it is not, mark it “0”.
- If you marked the cell “1” move to the meta-structure document. Fill out the name of the entity in the C1 cell in the matrix document under the “Name” column. Fill out the name of the entity in the A2 cell in the matrix document under the “is dependent on” column. In the “Reason” column write down your reasons and argumentations for connecting them.
- Go back to the matrix document, move to the empty cell in D2. Ask yourself: “Is the entity in cell C1 dependent on the entity in cell A2?”. If it is, fill out this cell with a “1” and provide the reasoning in the meta-structure document, if it is not, mark it “0”.
- Continue until you reach the end of the entities listed in the top row. Move to the empty cell in B3. Ask yourself: “Is the entity in cell B1 dependent on the entity in cell A3?.” If so, mark it “1” and provide reasoning for this in the meta-structure document. If not, mark it “0”.
- If you move to the empty cell C3 you will notice that the column and row entity are the same. You’ve reached the matrix diagonal again. If you do not want this node to loop to itself, mark it “0”.
- Move to the empty cell in B4. Ask yourself: “Is the entity in cell C1 dependent on the entity in cell A3?.” If so, mark it “1” and provide reasoning for this in the meta-structure document. If not, mark it “0”.
- Continue until you reach the end of the entities listed in the top row. Then move to the next row, repeat until you’ve reached the end of all the entities listed in column A
- You have now completed your matrix.
What’s next?
In the process of asking yourself about depedencies between your archaeological entities, you have created your first matrix, which for all intents and purposes is the same as creating your first network. Of course all these ones and zeroes do not look anywhere near how you envisioned your network would. Don’t worry, we will get there. In the next post of this “How-to”, I will walk you through how to visualize your network in visone, giving us the familiar view of dots connected by lines as well as many more and quite powerful ways of visualizing and descriptevely measuring your matrix.
Regardless of how your network looks at the moment, I hope that systematically thinking through your matrix, providing and making explicit your (internal) argumentations for dependencies between relevant entities, has already allowed you to make steps in the understanding of the phenomena you are interested in. I would love to hear about what you have learned so far in the comments.
I will expand and update this how-to in the next few weeks with figures and additional posts on how to start using some descriptive network statistics and more powerful visualizations in visone. If you run into any problems while following these steps, please let me know via the comments or the contact form. If you have any suggestions on how I could improve this “How-to”, that’d be great!