OK - useful, challenging!

This is complex. It would be good to get some links to specific examples, but the description is pretty clear.

Links added.

👍

The thing I didn't realise until late on is that they dynamism is in the data (content or specification) rather than the interface, representation and interaction. So this is time-varying data, with unknown characteristics? It might be good to try to be a little more explicit about this.

Having read through, I do wonder whether the idiom isn't: "software agents for automatic updating of dynamic data streams" What do you think?

Having read the reflection, it feels as though this idiom would be better framed as: "using agents to simulate streaming data to test automated visualization systems" or something. User story is : As a visualization system designer I want to test the robustness of my visualization system in preparation for unknown data streams So that I can assess the reliability of my system. Sorry if I have this wrong, but it feels as though this is the problem you are trying to solve with the agents. This focus on a specific problem and a documented solution (presuming that you used ideas from Forbes et al, or came up with better solutions based upon this thinking) should help you reflect on what worked, what didn't what you had to change and what you would recommend for the future.

I changed the idiom to "Updating dynamic data for visualization" to highlight the data part more.

Dynamic in the context of this idiom is that datasets change dynamically, e.g., the number of cases in an area is a time series. This time series changes every day. So the VIS needs to change dynamically. If you want for a more specific title, yes. However, this could be said for almost all idioms. Somehow, it sounds more like a paper title than an idiom. Perhaps like OD map, we can create a new term such as "DDS Agent". On testing your system, yes, in theory you can. You can create testing data streams. The agent can download it automatically. However, "unexpected" is not easily definable. It is the designer's job to guess the unexpected. For example, if one design a COVID time series VIS needs to anticipate the change of y-values dramatically.

That's an important problem to address and helps clarify. I hope you can use it in the user story or description to help clarify the problem. The same is true of the first para - it helps make it clear what the problem is.

The revisions have tidied things up nicely and it’s all better framed now, and it’s clear to see what you have done. But it’s much harder to really understand what the piece of knowledge is that you are using, and there is very little reflection that addresses a particular known idea / approach / method in this context. So, I think it would help to continue framing this around some idiom that you select (I Like the idea of Dynamic Data Scheduling Agents - find the paper that describes them, explain why they might be useful in VIS, how you applied them, what was achieved and that was learned) and changing the title from an intention / aim to an idiom / idea / technology may help with this refocussing. Have a look at my comments and see whether or not they help you shape this up as a description, discussion and application of a named idea/technique/ approach.

Added Scheduling Agents to the title

This is interesting. So you implemented something that was described in Forbes, or Rheingans, but were limited by an incomplete implementation and description? Interesting challenge! Recommendation: implementations, full descriptions and Open code are important Action: that is what we have done!

Yes. We can add a github reference.

Added a ref to the review paper of software agents.

Isn't there a whole body of work on progressive analytics and data science for big and time varying data? Is this relevant? JDF and even Cagatay have worked on this. Apols if this is not relevant. Fekete, J. D., & Primet, R. (2016). Progressive analytics: A computation paradigm for exploratory data analysis. arXiv preprint arXiv:1607.05162. Turkay, C., Kaya, E., Balcisoy, S., & Hauser, H. (2016). Designing progressive and interactive analytics processes for high-dimensional data analysis. IEEE transactions on visualization and computer graphics, 23(1), 131-140. Stolper, C. D., Perer, A., & Gotz, D. (2014). Progressive visual analytics: User-driven visual exploration of in-progress analytics. IEEE Transactions on Visualization and Computer Graphics, 20(12), 1653-1662. Chandramouli, B., Goldstein, J., & Quamar, A. (2013). Scalable progressive analytics on big data in the cloud. Proceedings of the VLDB Endowment, 6(14), 1726-1737.

It also contradicts MC's comment below: "Dynamic visualization was an old challenge. Many industrial applications require a VIS system to work with dynamic data streams"

Thanks Jason. I added references to JDF and Cagatay's papers. The data we're dealing with in the generic system is relatively small and doesn't require any sophisticated progressive loading or analysis.

OK - sorry if I have sent you in the wrong direction here (I keep asking people to focus!) I'd love to try to focus in on the specific known thing that you tried to use to achieve dynamic data vis. Maybe the Turkay and Fekete papers are not what you used! If you had to pick one informative paper that offers methods (that you used) for "Updating dynamic data for visualization" and to which you might point people (like the ones you describe in the user story) who might want to do this - which would it be? Probably not Rheingans as that's about interaction? So Cakmak et al? The description and reflection then tell them how you used it, whether it worked and how you innovated to try to make it work well!

We're a bit limited without some links to, or screen dumps of, the thing you are describing. Hope you can add them!

I added description in text.

OK - it would be great to get some more comprehensive documentation of what the solution was - this allows us to make judgments and learn about the idea that you were transferring.

Added a short intro taken from the RAMPVIS paper. It could include more though.

I'm sure Phong did really well - he's fantastically talented, but we're really looking for comments on how the concept worked in this context. What did we learn about the concept of Dynamic Data Visualization (as documented in 3 and 4) from what we did in SCRC (as described in 6). Sections 3 and 4 are a little vague and section 6 is empty - so this is a challenge.

See comments at top of document!

cleaned up the 'Phong' parts

OK - great. The reflection is a little limited here. It would be great if you folks could get together to think about what was learned. Oh, and remove the notes!

We need names please, including modelling collaborators. BibTex format if possible! Dykes, Jason and Chen, Min and Abdul-Rahman, Alfie and Turkay, Cagatay and Vardy, Jamie Like that!

Typo! OK - useful, and do we know whether they use the approaches listed above, or how they do so?

Delete the notes please!