Breaking the Qual/Quant Barrier

Usually information is either qualitative like profession and gender or quantitative like age and weight. These characteristics can generally not be compared or have operations on them that combine them. In addition qualitative data can not have quantitative operations on it, and quantitative data is resistant to qualitative operations. I call this the qual/quant barrier.

The Endeme system begins to break down the qual/quant barrier.

Endemes:

An endeme is a list of characteristics which are placed in a particular order indicating the value of what is endemized. The list of characteristics is drawn from an 'endeme set' and all of the characteristics in the endeme set are used up in each endeme. So an endeme contains the complete list just in a different order for each one.

So, a person's height could be stored with other physical measurements like their girth, shoe size, hat size, etc with fudge factors for each measurement in the endeme set. Colors could be stored together as an endeme showing the strength for each. Weights of items in a pack could be stored as an endeme.

other examples:
- height,length,width
- weight,mass
- color,shade,saturation
- face dimensions
- distance
- ages,dates
- power,energy
- speed

Quantity is still fudged because the data are stored relative to each other so exact quantities will not be obtainable, just approximations. If you keep 'units' plus 'averages' of each characteristic in the endeme set this will allow these data to be mixed in the same endeme and reconstituted as needed from and endeme of that set. This means that the characteristics will have to be selected so that their values tend to be orthogonal and even opposing to each other. Percentages of a resource are perfect example.

An example that will not work would be physical characteristics related to size. It doesn't do to have all characteristics that have a positive relationship to a person's weight for example weight, height, girth, stride etc. because their values would all be high or low together and the endeme only shows relationships, not actual quantities directly, so some of the items if they are related would be actually forced to be too low in the list, if say they all need to be high in the list.

A good example that would work would be resource distributions in an army pack. There is a finite resource (weight), and the various kinds of items would each use up a certain percentage, so when one went higher another would tend to go lower. And the endeme would properly account for this. Colors work, but only if darkness and desaturation and things like that are included, so the colors have counterweights.

OK, now back to bringing quality and quantity together.

Endemes combine qualitative and quantitative data. So there might be a breakpoint in each endeme or endeme set that says where the qualitative (combination) stuff ends. We already should have one saying where zero is. The fields qual/quant ized have to be comparable to each other and you have to accept that their quantitative values will be heavily fudged.

On the qualitative side, simply set a breakpoint below which characteristics do not apply in a particular instance. For example if you say only the first six colors in the 22 characteristic list are 'used', then qualitatively the color indicated is a combination of those six.

On the quantitative side, store units for each characteristic and make sure the characteristics are not 'encumbered'. Weight and gender might not be combinable because gender has no measurable nature, but profession does Lets say profession was divided into rough measures of how technical, how people oriented, and how task oriented. These could be included along with weight and hat size in the same endeme (given units for a characteristics), and since they tend not to have relationships with each other, most (not all) endemes would contain fairly accurate quantitative data for each and could be matched with other endemes for other people using the same characteristics. True, someone who was high on most characteristics - worked intensely with people, technical things, tasks, and had a large weight and hat size would be stored inaccurately but this is the price you pay for the benefit of having quantitative and qualitative information mergeable, matchable, comparable, and relatable.

Number of Characteristics in an Endeme Set

This is why you have a large number of characteristics in one endeme. The more characteristics, the lower chance of inaccuracy. 22 (17-26) seems to be a good number for characteristics. Beyond that I suspect that characteristics are hard to chose that don't relate to each other excessively and below that I suspect that the inaccuracy effect I just mentioned will surface too often for the information to be usable. In other words the lower the number of characteristics the more likely it is that the data in particular instances is encumbered; the higher the number of characteristics the more likely it is that the characteristics themselves are encumbered. Because some characteristics will generally be encumbered (you are going to want to have height and weight in your list for other reasons), you generally need to get out near 22 characteristics to have enough to resolve this. The sweet spot seems to be 22 or 23 items.

[4] comments (810 views) |  [0] Trackbacks   [0] Pingbacks

Machine Understandable Definitions for 5000 Common English Words

After three year of work I have finally entered a semantic defining endeme for each of my 5000 common English words. The endeme set is A)area, B)biology, C)change, D)directive, E)express, F)feel, G)good, H)human, I)incident, J)movement, K)kind, L)love, M)mental, N)new, O)object, P)physical, Q)quantity, R)resource, S)system, T)time, U)ugly, and V)value. These characteristics have been combined/ordered (as an endeme) for each word to provide a machine understandable definition assuming the machine can understand the 22 characteristics. And I completed this phase two weeks early!

[4] comments (469 views) |  [0] Trackbacks   [0] Pingbacks

Partial Endeme Sets

First a definition: An endeme set is a list of items that can be combined in any way. Often a permutation of the entire list is used to define the combination. An endeme is a particular combination. Endemes is an old English word for 'together' and is probably the origin of the word endemic, which carries the connotation of melded entirely together with a particular location.

I've been tinkering with the idea of partial endeme sets. My work with endeme sets for superhero powers has shown me that some sort of structure could be larger than the endeme set. I divided super powers into four groups: Materially based (control of file for example), medium based (for example flying through air), other based (for example talking to space aliens) and personal (for example slowing villain's movements). Each group consisted of two endeme sets, a target or focus to work on (for example fire) and a verb (control, fly through, talk to etc.). The target lists had almost entirely separate items, h0owever the verb lists had lots of overlap (for example one could control fire, air, space aliens, or villain movements). They were also not entirely the same (you can't exactly talk to villain movements or fly through aliens). The overlap between the verb lists was about 60%. I was able to come up with a generalized verb list which each of the types of verb (endeme sets) overlap about 75%. In addition I've discovered I will probably have to use something like this for Character types from my sheet from a couple weeks ago.

So what do we have here? Something like a larger structure organized around a 'central' endeme set from which the other endeme sets differ. Endeme sets are like types for an endeme. One could use strong typing for that. We also have here a means of blending types into each other. I guess I would call that blended typing? The structure allows for 'strong' typing but does not unnecessarily restrict typing. Also partial endeme sets would be able to evolve by slowly dropping in or pulling out individual items. Pretty cool huh?

Here is my superhero powers sheet:

[4] comments (437 views) |  [0] Trackbacks   [0] Pingbacks