Field methods paper: lit review and comments

Moved to a new topic for improved readability. See reviewers comments here.

Reading the first paper now. It is clear that semantic network analysis as per Jiang et. el. is closely related to semantic networks as studied in Guy @melancon’s own work on, for example, analysis of news coverage in France. When @brenoust developed his notion of entanglement, he was working on such networks.

However, the framing and the language here look quite different. For example, if you consider the “Research procedure” section (p. 3717-3718) it slips from “most occurring words” to “co-occurring concepts”. I have to re-read several times to make sure I know what I’m looking at. Additionally, there is the puzzling move to throw away everything except the 150 highest occurring words, which appears to destroy plenty of information without any consideration of how this would influence results. This might be because these authors, unlike Guy and his colleagues, do not like to deploy math to deal with large networks. The semantic networks that Guy showed me are very large. This seems corroborated by sentences like:

This study used eigenvector centrality as the criterion measure because it indicates a word’s overall network centrality.

Which looks imprecise to me. There is no such thing as “overall” network centrality!

Another important difference is whether analysis considers the keywords associated to news pieces or their text. The reference to “stop words” in the paper seems to point to the latter method, with some simple NLP being deployed at step 1.

Am I wrong? Are there two separate traditions of building networks from word co-occurrence? If so, we should highlight it, writing a “related work” section of the paper that would be structured like this:

  • Semantic networks in anthro literature have been used in such and such a way [mostly @amelia]
  • Network scientists also use semantic networks, but with some differences [mostly by @melancon]

Invitation sent. I will def have read the second paper by then.

hi @alberto, I have read some of the articles (the ones I’ve been able to find), is it possible to join the meeting, just as a listener / observer :no_mouth:, I’m curious about the topic. thanks :slight_smile:

1 Like

Of course, @tah. We will meet on September 21st at 16.00 CEST. use this link: Launch Meeting - Zoom . You will need to download the app, so connect a couple of minutes early :grinning:

1 Like

From an anthropological perspective it’s almost totally unreadable. The sweeping assumptions about culture in general and national political culture make for what I see as very questionable conclusions. At best— sure, ok. I’m not sure I learned anything in the end— when you assume at the beginning that democracy is central to Americans and authority is central to the Chinese, of course that’s what you’ll find in the word associations. And how can you analyse civil society at that level? Anyway… more to say in the talk as far as it concerns our work

1 Like

I finished my second paper, Social and Semantic Coevolution in Knowledge Networks by Roth and Cointet. This is a fun paper! Bottom lines:

  • MAJOR difference: their data are built using NLP + a reduction of what they call “concepts” done by domain experts. So, it is not an ethnographic method.
  • MAJOR difference: the focus of the paper is structural, in the NetSci tradition. They don’t care what scientists are actually saying in the network of co-authorship on zebrafish embryology. They are interested in figuring out how individual behavior produces network structure. Our focus is on the effectiveness of SSNA to help us interpret large-scale ethno corpora, i.e. what they actually mean. It is semantic, not structural.
  • Less major difference: they keep the graph as bipartite instead of projecting. Interesting approach!
1 Like

Finished my second paper, Mediating Ethnography: Objectivity and the Making of Ethnographies of the Internet by Beaulieu. Decent overview, nothing too groundbreaking here, easy enough to cite in support of our method.

Notes below:

How do we do ethnography of the internet?

Brings up lurker problem (calls it “mechanical objectivity”) – that online, the ethnographer can lurk without being noticed. We discussed this in a different thread on ER.

Emphasis on value of interaction as part of ethnographic approach

Need to read: Phil Howard’s 2002 article in New Media and Society on Network Ethnography and the Hypermedia Organization (amusingly, he is my supervisor’s husband and is also at the OII); also read Hine 2000.

‘Reflexive Intersubjectivity’ – crucial to good ethnography. Useful phrase to capture how we engage online (e.g. we do not lurk!)

Hine — “dialogue between clearly interpelated individuals, and an ethnographer who attempts to make herself and her goals as clear as possible” (150)
Heath et al 1999 interact with users
- “knowledge comes from engagement and interaction, always both purposive and incidental” (150)

Blog – using blog as introduction to explain research and to let informants follow the development of the thesis (Mortensen and Walker 2002, 51)
- Just like our open research group on ER

To gamers, easier to fake an actual-world persona than it is to maintain a consistent long term online presence (ibid)

Blog and phenomena studied “live” in the same space (just like Edgeryders)
- Form of communication and use of technology are aligned to the cultural phenomena being investigated
○ Challenges practice of leaving field for writing up
○ Ethnography develops a ‘more ongoing’ character

Apparently, blogging has been a source of backlash by peers, considered too ‘journalistic’ (Glenn 2003)

Changing values in wake of new scientific communication

The Internet: An Ethnographic Approach by Miller and Slater (2000)

Arrival stories in ethnographies of the internet – Baym 1995; Correll 1995

Capture— internet allows for recovering mediated interactions as text
- Open and accessible (these are tropes)

Text is the only thing to be analyzed (also trope) — important for us to explain onboarding processes and our offline events, and the liveliness/live action elements of conversation. Beyond world-as-text (though no ethnographer actually sees online interaction as an example of true culture-as-text).

Digital allows for the scaling up ethnography, making it a sustainable and cost-effective way of observing everyday life for industry as part of R&D (she sees this as a hype/problematic kind of move, since it’s complicated, and also in the service of questionable aims) – Marsten and Plowman 2003

“The anxiety about a possible loss of the epistemic gain of the translating/trascribing ethnographer, through a realisation that textuality is always already present, may also be compensated by the creation of yet more complex transcriptions, such as network diagrams, large-scale semi-automated content analysis…this may lead to interesting new conversations between ethnography and other methodological approaches” (159)

First reading group under our belt! With @tah, @alberto, and @markomanka. And a productive session it was.

Main take-homes:

We (likely) need to be clearer about the fact that what we are doing is primarily an ethnographic method in service of better understanding communities and what they are collectively thinking. @markomanka, based upon insight from his readings which revolved around the ‘semantic’, theorises that it is possible that what we are doing at the end of the day is so different from ‘semantic network analysis’ historically that calling it something different may be useful. At any rate, we have clear points of difference from this historical work, and we can articulate them. Our task is to continue to read and see if this bears out, and if we need to make our poinnt of difference/main project more clear. (or call it something else, like ‘conceptual’ network analysis). Other papers are more closely related to SSNA, it seems, so we read on with these thoughts in mind.

We need to do the work of reading the literature and 1) talking about its relationship to our work and 2) organising them to give a review of prior literature (@alberto points out). So while we continue to move through the readings, let’s be sure to write what the relationship of that reading is to our work (what are our points of difference, departure, etc) and perhaps try to categorise it relative to other literature we have read so far. To put it another way: what different bodies of literature are we engaging?

None of the readings thus far (minus a review of ethnography of the internet) do ethnography or are primarily interested in ethnographic methods. In fact a lot are much closer to NLP work. Let’s see if this changes.

We will also add a section on ethnographic methodology and ethics.

Homework: keep reading. If we could do 2 more articles by next Friday then discuss, that would be great.


Keep reading, indeed. This time we involve @melancon, and no more cafés for me!

1 Like

Third paper: Carpenter-Li-Jiang, Social Network Research in Organizational Contexts: A Systematic Review of Methodological Issues and Choices

I struggle to understand how this is relevant! A community is not an organization, and it does not make sense to do Burt 1997-style analysis on it (or does it? At least, it should be discussed).

I conclude that this paper does not need to be read too closely, or indeed at all. More than lit review, this is meant to be used as an example of how we could use many network measures on our data, but don’t. I would assign the task of writing 1-2 paragraphs on this to @melancon.

Fourth paper: Shapiro, S. (1971). A net structure for semantic information storage, deduction and retrieval. Paper presented at the Second International Joint Conference on Artificial Intelligence.

This is a “vintage” CompSci paper, proposing a precursor of what today we call linked data (RDF, OWL etc). I only read the conceptual part, as the more technical stuff is probably irrelevant to us. I feel the comments made by @markomanka in our call apply completely: “semantic networks” are used in this literature in kind of the same way as “the semantic web”. By recommending it, I think reviever 4 wants us to say something like this:

SSNs as described in this paper differ from semantic networks described in computer science literature (later to evolve into the concept of the semantic web [Resource Description Framework (RDF) Model and Syntax Specification]) in several ways. First, SSNs are meant for human consumption, and not as a system for computers to efficiently store and retrieve information. Second, and as a consequence, they are underpinned by a simple ontology, that can easily fit in human short term memory. Our chosen representation has only two types of nodes, participants to the conversation and ethnographic codes, and only three types of edges, comments (participant-to-participant), references (participant-to-code), co-occurs-with (code-to-code). By contrast, graph databases (proposed as early as 1971 [Shapiro, 1971]) can and do encode many types of relationships

Does this work?

1 Like

@amelia @melancon @markomanka just flagging that I am about to disappear for a while, because several international trips in a row. I will be fully operational again on October 16th. I ask that Amelia keeps coordinating operations, without waiting for me.

On the up side, I have read and commented four of the six papers assigned to me. I am missing the two in Encyclopedia of Social Network Analysis and Mining. Springer, New York, NY, which I cannot access until I go to Alicante next month. Could you, Amelia, send me the PDFs?

Noted @alberto.

I’m having a hard time accessing that book as well. Neither of my institutions grant me access. I’ll see what I can dig up. Luckily they are both shorter encyclopedic entries.

@markomanka, @melancon can we try and schedule a call for next week? I’m flexible as it will be my late night/evening.

hi @amelia, I kept reading on the subject, and would be glad to join again, as a beginner / learner :slight_smile:

1 Like

Excellent :slight_smile:

Next week is a bit hectic, but we can try… It would be easier for me on the 1st, or the 2nd in the morning… The 4th is a no go.

Today will work for a quick check in if you can do it — anytime after 6:30 pm Singapore time should do for me. I’ve done two more readings and can upload the notes.

Another option I’d propose is: let’s finish all the reading assigned on this thread, then have a call. We can discuss the readings as a whole, then we can make plans for what to add on for this literature review to respond to the readings (e.g. assign more reading from our own knowledge bases to answer their questions) and divide up writing tasks. This might be a more efficient use of time since it’ll just be two co-authors on the call (and it’ll give @tah time to read with us too!)

I’d suggest (as we have been) continuously uploading our reading notes as we finish them, since some people are more visual than verbal/do best with a combination.

I’d then set a total reading deadline of 10/14 – if you finish early, start organising thoughts around how you’d respond/incorporate the readings into a lit review and respond to the reviewer comments.


Hello @amelia @markomanka @melancon and our “honorary co-author” @tah . I spoke to Guy: he has some time Tuesday afternoon for a call, only availability next week. @amelia, still in in Singapore? Would you mind terribly being on a call at 14.00 CEST? Would you, Marco?

In the call, we should move on from examining the literature to discussing the presentation of our arguments.

I have read and reviewed all papers assigned to me, except the encyclopedia ones (Encyclopedia of social network analysis and mining Semantic Social Networks, 2018). No one seems to have access to these, not even Guy. I am going to Alicante at the end of the month and will look there, but for now this will have to do.

We can do this!


Definitely can’t do Tuesday night I’m afraid… any other night works.