By what criteria is coding being done?

[ This is for @Amelia ]

Amelia, we are here with @melancon and @Jason_Vallet hacking away at the semantic graph. It looks really impressive!

We were wondering:

  1. Are you proceeding by any criterion in particular? For example, would it make sense to extract a part of the graph that you have already coded? If we see a comment with no tags, how can we know if you have not coded it yet or if you have, but found it uninteresting?
  2. Is there a hierarchy of tags? It would be computationally simple to color-code by keyword, but for it to make visual sense we need a limited number of top level codes?

Criteria and Hierarchies (+need re-access to taxonomy manager)

Hi all, glad to hear it is going well!

Currently I am moving through posts starting with those with the most comments and making my way downwards. Before the explorer was categorized in this way, I was going in chronological order (but sticking with challenge responses once they became codable, and coding posts with at least one comment).

I don’t partially code any posts, so if there is a comment in a coded post left untagged, it is because I did not find anything in the comment that I thought needed a tag assigned. With one notable exception of course ---- if the comment is more recent than when I tagged the post. (side note — if we could create a notification system to let me know when posts I’ve already coded get a new comment, that would make for more thoroughness, but I understand our resources are not unlimited :slight_smile: )

I am happy to change the criterion if it would make for a better graph sooner — I figured the more interactions I could code (e.g. more comments) the more robust of a graph we would have. I can do a double limit (coding posts with the most comments in one particular challenge response category, for example). The categories have overlapped a great deal however (lots of migration-related stories in categories other than people on the move!) and one category has far more responses than the others. Thoughts on how else to limt would be welcome.

As for hierarchies, I am keeping a list but have not expressed the hierarchy in the taxonomy manager. That was my project for today, but I seem to have lost access to the manager (see screenshot below). Can someone return access to me? If so, I can formalise the hierarchies this weekend so they are visible to all. As well as merging/forking some overlapping codes that I have kept a running list of. Can’t wait to see the graph!

1 Like

Try now

I am too tired to really think about it, but I gave you admin powers now. Be careful… :slight_smile:

Hierarchy a priority

Guy and Jason say: if you could implement hierarchies first, that would help with the prototype. Which is still only a prototype, but in the initial discussion we felt the need to go beyond mockup data to see if stull jumps at us. As I write this, Guy and Jason have pulled in all your annotations from the APIs and are building the graph of co-occurrences of codes. :slight_smile:

Quick chat

oh boy, let’s have a quick chat to make sure I don’t do any serious damage here  :) I am flexible all weekend and on Monday if anyone has time for a 5-10 minute conversation about the hierarchies. I can find the list of annotations, but I don’t see an immediately obvious way to organize them/build hierarchies.

My ideal: a way to see all the codes on one page, and a system of dragging and dropping them into hierarchies. But any way of visualizing the codes so that I can organize them into categories would be really useful (or if someone could just give me a workaround approximating this kind of organization). Even just a list where I can select a bunch at once and put them underneath a different code would be great.

@melancon @Alberto @Jason_Vallet

We’ll be together hacking all weekend

So when you get up ping us and we’ll Skype, or something.

Hierarchies (so you have something in text form)

Here’s a list of higher level tags. Methodologically it makes sense not to create too much hierarchy before the first full pass is done (so as not to bias the preliminary coding), but once I do some analysis there will definitely be more hierarchies at the level of content. (e.g. alternative medicine being a higher level tag with acupuncture nested under it)

I nested some tags but there’s more work to be done there, just wanted to give you an idea.

alternative approaches

alternative infrastructure

alternatives to the State

case study

case study : access space

case study : Activist Net

case study : Airbnb

case study : Buoy

case study : Celly

case study : Internationales Congress Centrum (ICC) Berlin

case study : Metropolitan Community Clinic at Helleniko

case study : orange house

case study : sensorica

case study : the Jungle

case study : Uber

case study : Woodbine Health Autonomy Resource Center

case study: Little Device

case study: multibed mode

case study: open value network model

case study: somatic experiencing with hurricane survivors

case study: trauma tour

case study: unmonastery

challenge

challege: legal barriers

challenge : mobility and time

challenge : rigidity of NGOs

challenge: funding

challenge: inclusion

challenge: motivation

challenge: NGOs

challenge: re-education

challenge: replicating

country comparison

design intervention

event

example

implementation

methodology

observation : connection made on site

research question

research question : criteria for technical functionalities

research question : emotional stress

research question : explaining causality

research question : openness

research question : quackery

1 Like

Question!

Got it @Amelia . Do you want me to attempt to incorporate in the database, editing the taxonomy term hierarchy? Or do you prefer to keep the hierarchy flat, and we introduce it only in a local copy that we download?

With a finished product, the taxonomy should come down via the OpenEthnographer APIs and be rebuilt automatically by the web dashboard.

Go ahead and incorporate

I don’t see why not— these ones for sure should be higher-level. I’ll keep the rest flat for now.

Other question

Ok. Are you sure you want to keep this format?

metaTagName : tagName

example

case study : The Jungle 

With a hierarchy, you do not really need it.

Happy to ditch it

:slight_smile:

Done!

Ok, I am done.

I noticed that there are many codes with the forms case study : codeName and challenge : codeName that you have NOT mentioned. Those I have ignored.

On the other hand, I created top-level codes for “challenge”, “observation” and “research question”. Need sleep now. Tomorrow we will re-import the data including the meta-codes.