You're reading the public-facing archive of the Category Theory Zulip server.
To join the server you need an invite. Anybody can get an invite by contacting Matteo Capucci at name dot surname at gmail dot com.
For all things related to this archive refer to the same person.
Fabrizio Genovese (Mar 25 2020 at 09:02):What would be insanely cool is to have a graphs of different arguments in CT connected (vertexes are research topics, arrows are dependencies). Then we can tag each vertex with the name of the people working on that thing and have a very visual and useul map of "who-does-what" in the space. :D
Matteo Capucci (he/him) (Mar 25 2020 at 09:11):Fabrizio Genovese said:
What would be insanely cool is to have a graphs of different arguments in CT connected (vertexes are research topics, arrows are dependencies). Then we can tag each vertex with the name of the people working on that thing and have a very visual and useul map of "who-does-what" in the space. :D
Maybe some data scientist could mine this topic to do that!
Lee Mondshein (Mar 25 2020 at 09:37):I’m delighted to see the adjunction of Zulip to the seminar : ) I love the suggestion of a graphical/ compositional visualization of research topics and community. At its foundation, I would suggest exploring the industry-strength, well-documented Stanford Protege system for the (graphical) construction of ontologies — it might provide a semantically deep framework upon which to build the visual interface you envision. (The support team is terrific.) To be clear: Protege beautifully supports the creation of sophisticated ontologies that are then accessed logically.
Fabrizio Genovese (Mar 25 2020 at 09:46):I am not an expert, but I was thinking more something like a data visualization program. That is, we slap all the relevant stuff in a database
(e.g. person, research topics, location, mail) and then we use some data visualization thing to project the entries in different ways
(e.g. putting people on a map, creatng the graph with the dependencies etc). Does Protege do this kind of stuff?
Lee Mondshein (Mar 25 2020 at 10:09):Hi Fabrizio ! Protege basically would serve as a database. The idea is that the database is not built on a set of relational tables, but on a labelled graph/ network. The result is that queries can trigger really neat, sophisticated inferences via path following (how Categorial!). I hope this brief explanation makes sense/ is somewhat clear !
My motivation is not sophistication for its own sake. On the “practical” side, I like to say that mathematics is ( among many things) metaphor — on steroids. Really great collaboration can arise when two people discover unsuspected metaphors: “look at that diagram — it shows that an A is really just a kind-of B.”
An ontology-based database can help to support wondering and wandering around that can trigger new metaphors — discovering what you didn’t know you looking for : )
Matteo Capucci (he/him) (Mar 25 2020 at 10:13):Lee Mondshein said:
Hi Fabrizio ! Protege basically would serve as a database. The idea is that the database is not built on a set of relational tables, but on a labelled graph/ network. The result is that queries can trigger really neat, sophisticated inferences via path following (how Categorial!). I hope this brief explanation makes sense/ is somewhat clear !
My motivation is not sophistication for its own sake. On the “practical” side, I like to say that mathematics is ( among many things) metaphor — on steroids. Really great collaboration can arise when two people discover unsuspected metaphors: “look at that diagram — it shows that an A is really just a kind-of B.”
An ontology-based database can help to support wondering and wandering around that can trigger new metaphors — discovering what you didn’t know you looking for : )
That sounds interesting! I created this topic to avoid polluting the other one. My question is: what should nodes and egdes of this graph be? Like, what's the philosphy in choosing that when setting up such a database?
Matteo Capucci (he/him) (Mar 25 2020 at 10:14):Here it is
Fabrizio Genovese (Mar 25 2020 at 10:19):Lee Mondshein said:
Hi Fabrizio ! Protege basically would serve as a database. The idea is that the database is not built on a set of relational tables, but on a labelled graph/ network. The result is that queries can trigger really neat, sophisticated inferences via path following (how Categorial!). I hope this brief explanation makes sense/ is somewhat clear !
My motivation is not sophistication for its own sake. On the “practical” side, I like to say that mathematics is ( among many things) metaphor — on steroids. Really great collaboration can arise when two people discover unsuspected metaphors: “look at that diagram — it shows that an A is really just a kind-of B.”
An ontology-based database can help to support wondering and wandering around that can trigger new metaphors — discovering what you didn’t know you looking for : )
Isn't this similar to what Conexus' CQL is doing?
Fabrizio Genovese (Mar 25 2020 at 10:23):Matteo Capucci said:
Lee Mondshein said:
Hi Fabrizio ! Protege basically would serve as a database. The idea is that the database is not built on a set of relational tables, but on a labelled graph/ network. The result is that queries can trigger really neat, sophisticated inferences via path following (how Categorial!). I hope this brief explanation makes sense/ is somewhat clear !
My motivation is not sophistication for its own sake. On the “practical” side, I like to say that mathematics is ( among many things) metaphor — on steroids. Really great collaboration can arise when two people discover unsuspected metaphors: “look at that diagram — it shows that an A is really just a kind-of B.”
An ontology-based database can help to support wondering and wandering around that can trigger new metaphors — discovering what you didn’t know you looking for : )
That sounds interesting! I created this topic to avoid polluting the other one. My question is: what should nodes and egdes of this graph be? Like, what's the philosphy in choosing that when setting up such a database?
So, the graph is just an idea. What I really want is a database with all the information, that we can then project in some different ways. The three things I think would be most useful to have are:
Matteo Capucci (he/him) (Mar 25 2020 at 10:24):Fabrizio Genovese said:
- Sorting by name. Just the usual big table in alphabetical order if you are looking for someone
- Map view. Having everyone's name as a pin on a map, together with their research topic. Useful if you want to find someone nearby doing research in some field
- Graph view. Here nodes are research topics, edges denote some sort of relatedness/dependnecy. E.g. You may have edges from/to "algebraic geometry" and "higher category theory". On each vertex, you have a set of people working in that topic. BTW, I don't think this will end up being a (pre)sheaf, but if would be very cool if it did!
Can we call this "Cerebro for CT"
Matteo Capucci (he/him) (Mar 25 2020 at 10:27):Fabrizio Genovese said:
- Graph view. Here nodes are research topics, edges denote some sort of relatedness/dependnecy. E.g. You may have edges from/to "algebraic geometry" and "higher category theory". On each vertex, you have a set of people working in that topic. BTW, I don't think this will end up being a (pre)sheaf, but if would be very cool if it did!
Cool, I don't see how the presheaf would act on arrows btw :P not that it is relevant, but since you brought it up
Lee Mondshein (Mar 25 2020 at 10:28):Good question re. Protege!
Just to start a (quick) response: The philosophy is to begin with the kinds of questions you would like to ask — starting with basic ones, and then getting as deep as you want. Then you look at the terms/ notions and the relationships that would underlie the (construction/ inference of) a response/ answer. You do not need to figure out potential reasoning.
You boil the terms/ relationships down to a labelled network, and Protege helps you construct/ edit it graphically. Searches are done logically, as in a conventional database — but the Protege system does a really deep inferential search. The responses come out, along with the underlying reasoning ( which can be metaphorically based).
Hope this helps.
Lee Mondshein (Mar 25 2020 at 10:29):I hope this gets into the “visualization” thread !
Matteo Capucci (he/him) (Mar 25 2020 at 10:30):Maybe there's a graph homomorphism (lifting to a functor between the free cats) from the graph of topics you propose and the graph of papers and citations.
Lee Mondshein (Mar 25 2020 at 10:31):Sounds like a very good suggestion. I hope my post of a minute ago got sent here rather than to “Introductions” !
Matteo Capucci (he/him) (Mar 25 2020 at 10:33):Lee Mondshein said:
Sounds like a very good suggestion. I hope my post of a minute ago got sent here rather than to “Introductions” !
Unfortunately, I had to move them here manually. You need to either reply here (on the topic chat) or edit the topic from the editor
Fabrizio Genovese (Mar 25 2020 at 10:45):Matteo Capucci said:
Maybe there's a graph homomorphism (lifting to a functor between the free cats) from the graph of topics you propose and the graph of papers and citations.
Probably, yes. But this creates a new problem: You'd need to categorize all the papers using keywords and such. This is doable, but it has to be obviously automated, and when you do so data will be dirty (mismatches, etc) and in need to be cleaned. I'd forget about the paper stuff altogether for now, because this is the kind of task that could assume monumental dimensions very quickly xD
Matteo Capucci (he/him) (Mar 25 2020 at 10:54):Oh yeah, that was absolutely a non-practical thought
Matteo Capucci (he/him) (Mar 25 2020 at 10:55):BTW I asked a friend of mine, who is a data scientist, what kind of tools he would use to mine such data
Lee Mondshein (Mar 25 2020 at 11:42):rRE: CQL. Good question, CQL is similar in a sense --- at a much higher level of abstraction/ manipulation/ power.
Georgios Bakirtzis (Mar 25 2020 at 12:53):I've worked with graph visualization extensively and would be willing to help. Have never used CQL.
Morgan Rogers (he/him) (Mar 26 2020 at 09:49):I've often dreamed of building a visualisation of this kind. However, for me its greatest potential is in enabling collaborative efforts into research areas which require a great deal of systematic work. If one could just look at a big graph of their research area and visually see not only what work has been done and by whom, but also where the holes and open questions are that they might contribute to filling, this could be invaluable.