Algebraic Semiotics (2020) | Hacker News Reader

Algebraic Semiotics (2020)

gjoncas.github.io

57 pointsauggierose10 comments

No mention of Algebraic Semiotics on the ncatlab wiki, which is sort of the standard venue for these things. There is an article on Goguen but it is short and does not really discuss this work.

The categorical structure seems simple enough that it has probably been described already under some other name, though.

If you look at the couple of semiotics related pages that ncatlab had, they all moved offsite, so it's just that ncatlab doesn't address semiotics, not that they don't address the algebraic interpretation of it.

https://oeis.org/wiki/Sign_relation#Semiotic_equivalence_rel...

https://ncatlab.org/nlab/search?query=SEMIOTIC

But why?? ncatlab contains formalizations of many branches of philosophy, such as this on Hegel

https://ncatlab.org/nlab/show/Science+of+Logic

I feel like I didn't get the joke. Can someone explain WTF is going on? This feels like the subject was using precise mathematical terms by metaphor to do ... procedural generation?

Looks like its advocating for a modeling language like UML, except it's based in category theory. The advantage is that you can formalize certain heuristics to help you quantify if your architecture is consistent and simple. You can also use it to model a jazz performance, supposedly.

My dream is to do this with narrative elements, but it would take at least a lifetime to realize. (Double)pushout rules can be viewed as a Turing-complete system so the possibility/reward space is vast.

Uhm maybe to generate narratives procedurally, for a video game for example? Like what is done for level design in roguelikes

Exactly, there's application videogames and interactive fiction[1]. I'm rather experienced in procgen for roguelike development so those techniques and their value and pitfalls have left an impression on me, but also illuminated how narrow the application of current techniques is.

Though there's some work that's been done in the area in terms of storylets[2], and story graphs[3], from a practitioner's point of view, there's a lot left to be desired. In particular accessible and demonstrable implementations, declarative syntax/language/config, ontologies of narration, and libraries of story elements.

In a lot of ways, it feels like the space is fresh enough that even basic questions like data-driven bottom-up approaches vs symbolic top-down approaches haven't even been thoroughly explored. Currently, the research space is small on the academic side, and small on the industry side, but for different reasons. I guess I feel like there's a lot of untapped potential, but the systems which would drive such innovation aren't yet poised to recognize the opportunity or incentivized to pursue it.

1. although IF is arguably a less popular medium, the approach is still applicable.

2. https://emshort.blog/2019/11/29/storylets-you-want-them/

3. https://www.aaai.org/Papers/AIIDE/2005/AIIDE05-019.pdf

I kind of wish I lived in a timeline where C.S. Peirce's work wasn't suppressed and the science of semiotic had as many brilliant people developing it as followed, say, Frege.

Deep... and interesting. Will need to think hard about this. Possibly with a glass of cyrogenic brain coolant (Chardonnay)

Reply on news.ycombinator.com