How to represent an abstract concept by digits?
-
This is generally done by "weighting" factors. Or producing a new scale based on a multiple of each factor based on their importance to the relative subject. "power" is a relative term, so you need to figure out a factor multiplier of how important each factor is to the total. We'll switch to computation power as an example, specifically graphics computation in my line of work. So in my "power" assessment of a computer for work, I have four factors, memory speed, CPU speed, Graphics speed, and disk speed. In my work disk speed is important because I render earth geometry on a massive scale, it must be "paged" from the disk. I create a relative "importance" scale for each which also puts them in comparitable units. You start from what you know is "not powerful" (a laptop with a 4200rpm hard drive and 66mhz memory) and what is "REAL powerful" (a workstation with RAID0 15krpm drives, DDR400 and dual 3.8ghz processors). To cross brands I would have to come up with a new scale to match units to my relative "worth" of each component.... CPU: 0.25 MEM: 0.20 Disk: 0.15 Graphics: 0.40 that all adds up 1.0 (I hope) and is my relative importance. Then you adjust each items "measure" because they are all in different units to come up with your own "new" unit that matches your relative scale of importance. You have just created a benchmark number. You can do the same with humans, adjust various parameters into a scale of health, or in your case "power." Since everyone's idea of power is different, you focus on what YOU think is powerful and adjust the scale to match. Just as my scale of computers would not work for everyone. In a program where disk access was much lower, it's relative worth would be lowered and other significant factors raised to compensate. _________________________ Asu no koto o ieba, tenjo de nezumi ga warau. Talk about things of tomorrow and the mice in the ceiling laugh. (Japanese Proverb) -- modified at 22:17 Tuesday 14th February, 2006
im very grateful for your reply, that is so help. but i still confuse with some points, how to define the "relative scale or importance scale" ? subjective assessment? i think it maynot scientific if define the scale by the feel， i believe there should have some more scientific and feasible approaches.
-
im very grateful for your reply, that is so help. but i still confuse with some points, how to define the "relative scale or importance scale" ? subjective assessment? i think it maynot scientific if define the scale by the feel， i believe there should have some more scientific and feasible approaches.
Bluebamboo wrote:
i think it maynot scientific if define the scale by the feel， i believe there should have some more scientific and feasible approaches.
Although my bench importance is scientific, it is still subjective. It is important to my software and represents the relative importance to one criteria of comparison. The relative importance to Quake game play, or for super-computing computation, or for Lucas digital rendering all would have different ratios or they are all subjective to the specifics of the purpose of the benchmark. Thus the controversy over "futuremark" benchmarks on computers, trying to creat any one scale to represent all function is impossible. All possible uses are different. What is important for Quake is not as important for Microsoft flight simulator, or Doom, or Fear, etc. So you find a specific "purpose" of your power scale, and rate each factor as a ratio of importance to that power scale, then adjust the ratios to units of measure to come out with a power scale. If you want one "power" scale for all humans, it is impossible. Power for what? Charisma is very powerful, age can be powerful in strength, but not always in wisdom. So you still have to come up with a relative purpose, which is subjective. If your intent of "power" is physical strength, then weight may help if the body mass is muscle. If the idea of power is intelectual, higher age will be more important and weight less important. Still subjective. You can search forever for one relative scale of "power" that encompasses all humans for all purposes, and never get there. No matter what you choose as a "purpose" it is still subjective. _________________________ Asu no koto o ieba, tenjo de nezumi ga warau. Talk about things of tomorrow and the mice in the ceiling laugh. (Japanese Proverb)
-
in my opinion, u should first know the relation between the output and the inputs. or if you just want to record the concept by digit which can be reversely decomposed back into its composed elements, then try (e.g.height=192cm,weight=200p, age=25year old) power = 1000,000xheight + 1000*weight + age is that what u want ?
Thanks for your reply i think you misunderstand what i mean, the first reply from Mr. Jeffry J. Brickley basically reflect my thought. to compute power of human-being is just a example, what i mean is that some abstract objects such as power, speed, etc should have more or less factors which affect the value of objects, so what i want is to represent these objects by digits(not record them).
-
Bluebamboo wrote:
i think it maynot scientific if define the scale by the feel， i believe there should have some more scientific and feasible approaches.
Although my bench importance is scientific, it is still subjective. It is important to my software and represents the relative importance to one criteria of comparison. The relative importance to Quake game play, or for super-computing computation, or for Lucas digital rendering all would have different ratios or they are all subjective to the specifics of the purpose of the benchmark. Thus the controversy over "futuremark" benchmarks on computers, trying to creat any one scale to represent all function is impossible. All possible uses are different. What is important for Quake is not as important for Microsoft flight simulator, or Doom, or Fear, etc. So you find a specific "purpose" of your power scale, and rate each factor as a ratio of importance to that power scale, then adjust the ratios to units of measure to come out with a power scale. If you want one "power" scale for all humans, it is impossible. Power for what? Charisma is very powerful, age can be powerful in strength, but not always in wisdom. So you still have to come up with a relative purpose, which is subjective. If your intent of "power" is physical strength, then weight may help if the body mass is muscle. If the idea of power is intelectual, higher age will be more important and weight less important. Still subjective. You can search forever for one relative scale of "power" that encompasses all humans for all purposes, and never get there. No matter what you choose as a "purpose" it is still subjective. _________________________ Asu no koto o ieba, tenjo de nezumi ga warau. Talk about things of tomorrow and the mice in the ceiling laugh. (Japanese Proverb)
i don't want to generalize power using one scale, maybe you misunderstand what i said due to my poor english writting. it is necessary to clarify what i said : one abstract object with single meanings(physical strength) are only impacted by several identifiable objects, and then i think if we can digitize the abstract object by analyzing those identifiable objects with database analysis technology?
-
i don't want to generalize power using one scale, maybe you misunderstand what i said due to my poor english writting. it is necessary to clarify what i said : one abstract object with single meanings(physical strength) are only impacted by several identifiable objects, and then i think if we can digitize the abstract object by analyzing those identifiable objects with database analysis technology?
Bluebamboo wrote:
one abstract object with single meanings(physical strength) are only impacted by several identifiable objects, and then i think if we can digitize the abstract object by analyzing those identifiable objects with database analysis technology?
You have already made a subjective judgement, that physical strength is impacted only by several identifiable objects. I gave you a general plan for evaluating non-uniform, non-equal-unit factors into a common scale. You will find such rating scales used regularly from sports performance, to betting strategies, to photography equipment, to computers. But as you said they are all subjective, as any attempt to quantify an abstract principle using concrete factors... thus it is called abstract. For instance, physical strength, would not other factors play a part? body density? adaptive health (as well as general health) agility/dexterity accuracy bone-density (it holds the muscles up) Any any abstract rating of physical prowess, although you can measure any one factor easily, a full "rating" of physical strength will always be an abstract concept. Every scale you come up with will be at least partially subjective. Absolutely put every thought into comparing factors relative to a goal of physical strength, but also be aware at the same time that you will be ignoring subtlties that will always make your results subjective. It is still scientific, but ultimately all such scales are subjective. Thus, back to computers, the argument over a futuremark or any other generalized benchmark of computer performance being basically useless. If you want to use it for a specific purpose, use a scale of performance for that one and only that one purpose. That scale will be completely useless for comparing to another purpose, even though ultimately they are both scales of performance/power, they are always subjective. In the end, you must make a subjective decision. You would think something so easily quantifiable as computer performance would be 100% easy to evaluate and compare, and it is certainly easier than making an assessment of human performance/strength, and still it is subjective. If you want a long drawn out example of such performance evaluations, take a look at horse-racing. You have a physical creature being evaluated on a physical criteria through historic performance, genetic blood-lines (and their past performance), as well as physical characteristics such as height, weight, length of l
-
im badgerred with an question how to represent an abstract concept by digits. for example, how much power he or she has should relate with some factors such as height, weight, and age( i just show a example). and all of these factor can be depicted by digits: height(190cm), weight(200p), age(25year old). then the question is coming up: how to express the abstract power which he or she has using those factors???? finally, just like this: Mr. John(height=192cm,weight=200p, age=25year old) and then his power is 300. plz, everybody give me some advices or clues~
-
One bloody long word! ;P
-
im badgerred with an question how to represent an abstract concept by digits. for example, how much power he or she has should relate with some factors such as height, weight, and age( i just show a example). and all of these factor can be depicted by digits: height(190cm), weight(200p), age(25year old). then the question is coming up: how to express the abstract power which he or she has using those factors???? finally, just like this: Mr. John(height=192cm,weight=200p, age=25year old) and then his power is 300. plz, everybody give me some advices or clues~
Not sure I understand the question. Let me ask you: Do you mean you have a "training" set of data? In other words you know person 1 has attributes x1,y1,z1 and his power output = p1, and person 2 has x2,y2,z2 = p2 etc. etc. Now do you want to predict what power (pu) a unknown person with attribute xu, yu, zu has? In that case you need to use a Neural Net, or some other form of AI. For "fuzzy" data like what you seem to be referring to, that is the simplest way IMO.
-
Not sure I understand the question. Let me ask you: Do you mean you have a "training" set of data? In other words you know person 1 has attributes x1,y1,z1 and his power output = p1, and person 2 has x2,y2,z2 = p2 etc. etc. Now do you want to predict what power (pu) a unknown person with attribute xu, yu, zu has? In that case you need to use a Neural Net, or some other form of AI. For "fuzzy" data like what you seem to be referring to, that is the simplest way IMO.
exactly that is what i mean
Pandion wrote:
In that case you need to use a Neural Net, or some other form of AI.
i think it is not correlate with AI, because the example of human-being i shew , only in order to express what i think, the point is not at human or its power. actually, it is a problem in my project, which i want to find the correlation between eye and mouse cursor, but this correlation is very abstract, so i want to quantitate this kind of correlation according to some factors which can impact the correlation.
-
exactly that is what i mean
Pandion wrote:
In that case you need to use a Neural Net, or some other form of AI.
i think it is not correlate with AI, because the example of human-being i shew , only in order to express what i think, the point is not at human or its power. actually, it is a problem in my project, which i want to find the correlation between eye and mouse cursor, but this correlation is very abstract, so i want to quantitate this kind of correlation according to some factors which can impact the correlation.