Academic knowledge if often pretty narrow. Recently I was searching for a way of describing human movement. It turns out that while biologists do have large controlled vocabularies for describing spiders, I couldn”t find a controlled vocabulary for the task that’s done by good biologists.
It seems that the frameworks that are out there were done by dance theorists and aren”t well updated to the 21st century. And by that I mean it”s not trival to express everything in an XML file.
If you want to know which genes express which proteins than biologists are the right people to ask. On the other hand there a lot in biology, where biologists don”t know very much.
Recently I was searching for a way of describing human movement.
Can you tell me some more about that? (PM is fine if it’s not relevant to LessWrong.) I have worked on procedural animation for sign languages, based on the HamNoSys notation, and I am interested in extending it to more general movement and more general applications.
In studying bioinformatics, so I have a general interest in movment that goes beyond any specific project.
At the moment one of my projects is to build a new constructed language. A logical language like Lojban but better than Lojban.
Lojban takes a lot of semantic distinctions that exist in natural language for granted and doesn’t reorganise them.
One example would be months. You don’t need a seperate word for every month. It’s much better to call them with compound words like month-4 and month-6.
At the moment my design provides for 5 letter compound words for months and week days that use numbers. That reorganisation cuts down the number of words that you have to learn. It puts you never in the situation to ask yourself whether July or June comes first.
As part of that semantic reorganisation I think talking about human movement is in need of reorganisation. It’s hard to describe some Salsa steps in English. Example sentence:
“Keep your weight of the heel of your right foot while your move your left foot so that’s in an angle of twenty degrees in front of your right foot and that the foot touches the ground with the toes.”
That sentence is quite long. I could imagine that a semantic reorganisation would allow your to cut the sentence to a third or a fourth.
Because it’s hard to use English language to describe the movements I don’t have Anki cards to learn Salsa steps.
If I would have a language that would allow me to describe this precisely I could make Anki cards for it. That means I get a benefit for using the language. This gives people incentives to learn the language.
Of course I don’t want the language to be specialized on Salsa steps. I want the language to capture as much human movement as possible. Having an academic ontology would help because I could simply copy the structure of that ontology and make the words shorter and integrate them with general concepts I created elsewhere in the language.
An area where I already did the reorganisation are relationships. If you look at mathematical graph theory you see that it borrows concepts such as parent and child. If you design a language from the ground up that doesn’t make sense. You start with the graph theory. That allows you to have a word for node A is connected to node B that doesn’t define whether node A is the child or the parent.
In a graph parent gets a three letter word. A biological parent (as defined by inheriting something) gets an additional letter and is a 4 letter word. And you can add the same letter to turn graph child into biological child.
Father is then simply a male parent with means that it adds a 2 letter syllable at the end. The design allow to also easily have a 6 letter word for my older parent.
A word like grand-grand-grantparent gets replaced by a 6 letter word with parent plus the syllable for the word three.
HamNoSys notation seems very interesting. I will try to make the language able to express those distinctions.
Physical Therapists would also fit in there as well. But yeah, biology is huge considering it would theoretically include medicine, which itself has a multitude of sub-disciplines which don’t necessarily know much about each other.
If I type “notation for systems of physical therapy” into Google the top results I get are Eshkol-Wachman Movement Notation, Benesh Movement Notation and Laban Movement Analysis.
They are system made by dancers. They are general enough that they can also be used for describing non dancing movement.
In case I have missed something, do you know of a general system used by physical therapists which an controlled vocabularies that available online?
There a lot of money that goes in diagnosis equipment like fMRI”s, but not much for recording human movement in a analyzable way. Computer game designers who worked at Microsoft did work on the issue for kinect, but kinect isn’t really designed to be used for research purposes.
You can”t throw 10,000 dollar worth of high resolution camera’s at it to get clearer data. It would be nice if there would be open source bioinformatics software available for tasks like that.
I think we could learn a lot about how humans work by analyzing data like that instead of investing so much money into fMRI scanners and gene sequencing.
If I understand it right most of the software keeps the motion data in a pretty raw form. At 0.01 s the hand is that location A while at 0.02 s the hand is at location B. It doesn”t describe the timeframe of 0.01 till 0.60 as moving the arm to the right.
Especially it doesn”t have a well defined vocabulay for what moving a arm means. If you do moleculary biology you can download an ontology file in owl or obo which gives you a controlled vocabulary for describes moleculary biology.
Even controlled vocabulary to describe emotional states makes it into OBO Foundary.
If you want to make progress with science you need to have detailed language with specific meaning. But you are right, the movie industry and the game industry did produce some useful tools in that area. Much more than the medicine/biology/bioinformatics folks.
Academic knowledge if often pretty narrow. Recently I was searching for a way of describing human movement. It turns out that while biologists do have large controlled vocabularies for describing spiders, I couldn”t find a controlled vocabulary for the task that’s done by good biologists.
It seems that the frameworks that are out there were done by dance theorists and aren”t well updated to the 21st century. And by that I mean it”s not trival to express everything in an XML file.
If you want to know which genes express which proteins than biologists are the right people to ask. On the other hand there a lot in biology, where biologists don”t know very much.
Can you tell me some more about that? (PM is fine if it’s not relevant to LessWrong.) I have worked on procedural animation for sign languages, based on the HamNoSys notation, and I am interested in extending it to more general movement and more general applications.
In studying bioinformatics, so I have a general interest in movment that goes beyond any specific project.
At the moment one of my projects is to build a new constructed language. A logical language like Lojban but better than Lojban.
Lojban takes a lot of semantic distinctions that exist in natural language for granted and doesn’t reorganise them.
One example would be months. You don’t need a seperate word for every month. It’s much better to call them with compound words like month-4 and month-6. At the moment my design provides for 5 letter compound words for months and week days that use numbers. That reorganisation cuts down the number of words that you have to learn. It puts you never in the situation to ask yourself whether July or June comes first.
As part of that semantic reorganisation I think talking about human movement is in need of reorganisation. It’s hard to describe some Salsa steps in English. Example sentence: “Keep your weight of the heel of your right foot while your move your left foot so that’s in an angle of twenty degrees in front of your right foot and that the foot touches the ground with the toes.”
That sentence is quite long. I could imagine that a semantic reorganisation would allow your to cut the sentence to a third or a fourth. Because it’s hard to use English language to describe the movements I don’t have Anki cards to learn Salsa steps.
If I would have a language that would allow me to describe this precisely I could make Anki cards for it. That means I get a benefit for using the language. This gives people incentives to learn the language.
Of course I don’t want the language to be specialized on Salsa steps. I want the language to capture as much human movement as possible. Having an academic ontology would help because I could simply copy the structure of that ontology and make the words shorter and integrate them with general concepts I created elsewhere in the language.
An area where I already did the reorganisation are relationships. If you look at mathematical graph theory you see that it borrows concepts such as parent and child. If you design a language from the ground up that doesn’t make sense. You start with the graph theory. That allows you to have a word for node A is connected to node B that doesn’t define whether node A is the child or the parent.
In a graph parent gets a three letter word. A biological parent (as defined by inheriting something) gets an additional letter and is a 4 letter word. And you can add the same letter to turn graph child into biological child. Father is then simply a male parent with means that it adds a 2 letter syllable at the end. The design allow to also easily have a 6 letter word for my older parent.
A word like grand-grand-grantparent gets replaced by a 6 letter word with parent plus the syllable for the word three.
HamNoSys notation seems very interesting. I will try to make the language able to express those distinctions.
That is exactly how months work in Mandarin Chinese.
Physical Therapists would also fit in there as well. But yeah, biology is huge considering it would theoretically include medicine, which itself has a multitude of sub-disciplines which don’t necessarily know much about each other.
If I type “notation for systems of physical therapy” into Google the top results I get are Eshkol-Wachman Movement Notation, Benesh Movement Notation and Laban Movement Analysis.
They are system made by dancers. They are general enough that they can also be used for describing non dancing movement.
In case I have missed something, do you know of a general system used by physical therapists which an controlled vocabularies that available online?
There a lot of money that goes in diagnosis equipment like fMRI”s, but not much for recording human movement in a analyzable way. Computer game designers who worked at Microsoft did work on the issue for kinect, but kinect isn’t really designed to be used for research purposes.
You can”t throw 10,000 dollar worth of high resolution camera’s at it to get clearer data. It would be nice if there would be open source bioinformatics software available for tasks like that. I think we could learn a lot about how humans work by analyzing data like that instead of investing so much money into fMRI scanners and gene sequencing.
Look at motion capture for computer animation, both for games and for movies?
If I understand it right most of the software keeps the motion data in a pretty raw form. At 0.01 s the hand is that location A while at 0.02 s the hand is at location B. It doesn”t describe the timeframe of 0.01 till 0.60 as moving the arm to the right.
Especially it doesn”t have a well defined vocabulay for what moving a arm means. If you do moleculary biology you can download an ontology file in owl or obo which gives you a controlled vocabulary for describes moleculary biology. Even controlled vocabulary to describe emotional states makes it into OBO Foundary.
If you want to make progress with science you need to have detailed language with specific meaning.
But you are right, the movie industry and the game industry did produce some useful tools in that area. Much more than the medicine/biology/bioinformatics folks.
Motion capture by itself probably doesn’t.
However things like skeletal animation systems probably do.
Sorry, I guess I was making an assumption and merely adding to the noise.