KAES: Kinship Algebra Expert System
From InterSciWiki
Contents
|
[edit] KAES program
Download and read pages at Java program KAES: Kinship Algebra Expert System.
The key article to read is Dwight W. Read, 2006. Kinship Algebra Expert System (KAES): A Software Implementation of a Cultural Theory. Social Science Computer Review 24(1): 43-67.
More about KAES in other articles
Also download the kinterm maps
Install java (its free from Sun)
Once you download the program, right click the Kaes.jar file and Open with Java.
File/Open one of the .xml files:
- Omaha.xml
- AKT_KinMap.xml - American - Structure will be the same as Spanish
- Punjabi_KinMap.xml - Let's see if Sinhalese maps onto this
- SHIPIBOmap.xml
- Trobriand(MaleTerms)_KinMap.xml
(DRW: 0_KAES directory is on root)
[edit] The exercise
THere are two sets of instructions from Dwight Read under 2.4 and 2.5 for different methods. Practice one of the other on AKT and see if you get the idea, documenting the steps you use. For extra credit (3 points) and try one of the other cases, while documenting the steps used. Then save a copy of your reduced map in pdf and another when you reconstruct the map through the algebra. Turn in a 2 pp paper with what you tried to do, what you managed to do or where you have problems, what you showed, and what is the relevance of the exercise.
[edit] What is kinship
"The world of individuals with a shared set of moral and social values is one’s kin. Kin, by virtue of being kin, share a common framework of behavior that is appropriate even if the persons in question have not previously met.With larger scale societies, a domain of kin-like relations that extends many of the properties of shared, expected patterns of behavior among kin to a larger domain of individuals has been added. Tribes, for example, include individuals who may not knowtheir kin relationships but do knowtheir connection to one another through the social groups to which they belong such as lineages and how these social groups are linked via connections derived from kin connections. In state societies, patterns of expected behavior have been extended outside of the domain of kin and kin-like relations through institutions such as citizenship that define for a person the set of rules and laws that are to be obeyed and where infraction of those rules and laws may lead to punishment. But even citizenship has a quasikin basis through the notion that citizenship is passed on from parent to offspring. We may reasonably assert, then, that kinship is the fundamental concept on which human societies have been formed and constructed.Without understanding what constitutes kinship and how kinship articulates with, and provides the basis for, interaction in all other domains, we only have an incomplete picture of the nature of human societies and social systems. Trying to understand human societies without understanding the underlying kinship system is analogous to trying to understand the chemical properties of matter without understanding the underlying atomic and molecular properties from which those chemical properties derive. The kin relations that connect individuals provide the underlying framework for the interaction of individuals, and without understanding the properties and structural form that kinship relations may take on in a given society, we cannot fully understand how the individuals making up that society become social and not just individual beings and how the society is structured as a social system." -- Read 2006:6
[edit] Modeling objectives
"The algebraic modeling begins by structurally simplifying the kin term map through userselected options for simplification of a kin term map. The simplification is run in reverse during the algebraic modeling as one elaborates on an initial algebraic model constructed in accordance with the simplified kin term map. A modified kin term map is defined to be simplified when the modified kin term map has a single, ancestral generating kin term and only consists of ancestral kin terms. The KAES program allows for the user to model a kin term map using user-selected options that activate an ensemble of individual steps composing a stage in the algebraic modeling, such as constructing the descendant structure from the initial, ascendant structure (see Figure 5). Alternatively, the user may sequentially activate the individual steps making a stage in the algebraic modeling. As the modeling proceeds, the KAES program automatically introduces structural equations that are part of the distinctions made regarding kin terms in the kinship terminology. Each stage in the algebraic modeling of a kinship terminology structure is successful if the algebraic modeling ends up with an algebraic structure isomorphic to the kin term map from the initial simplification of the kin term map corresponding to the current stage in the algebraic modeling. The algebraic construction stops when an algebraic structure that is isomorphic with the kin term map for the complete kinship terminology has been generated." Read 2006:4
[edit] Generating terms and Operations
"The kin term map is constructed by first deciding on the kin terms that will serve as the generating terms through which kin terms are linked to one another. The generating terms must account for the closest kin term linkages between a person and another person, and so they will likely be some variant on the kin terms that can be transliterated as mother, father (and possibly parent), and their reciprocals, daughter, son (and possibly child). In addition, terms that can be transliterated as brother or sister (or possibly older brother, younger brother, etc.) will be generating terms in some terminologies but not in other terminologies. In the AKT, the analysis of the logic underlying the structural form of the kin term map displayed in Figure 3 establishes that the kin term brother is the product of the kin terms parent and son: brother = son o parent. In other terminologies, a kin term that can be transliterated as brother (or sister) or possibly as older brother (or older sister) is a generating term and not expressible as a kin term product. The distinction between a sibling term taken as a product versus a sibling term taken as a generating term provides the conceptual basis for the distinction between so-called descriptive terminologies versus classificatory terminologies (see below)." -- Read 2006:14
[edit] Incomplete Example: American terms (AKT)
Referring to Figure 6 (Read 2006:20): "The solid, single-headed arrows showthe result of taking a product with the generator,P. The dashed, singleheaded arrows show the result of taking a product with the reciprocal generator, C. The gray, double-headed arrows showthe result of taking a product with the spouse generator, S.
[edit] Map and Algebra Operations
I'll start with American terms, composed by generators P (Parents), C (Children), and S (Spouse), plus G (Gender). We simplify in reverse: S - C - P - G and then compose by P - C - S. Click this main menu item:
- Map Operations - where the second and last sets three items are simplifications. Ok, lets do the first of the second set
- Simplify structurally similar operations
- Simplify Structurally Similar - and we get
- Algebra Generators
- Transliteration of Generators
- Kin Term Generators
- [Mother, Father]
- [Daughter, Son]
- Self
Dialog box says:
- Generators were merged.
- Kin terms with multiple products ('one-to-many products') were split into structually equivalent kin terms.
- The kin term map is still complex.
Click this main menu item:
- Map Operations
- Remove Descendants
Dialog box says:
- Descendant kin terms have been removed from the kin term map.\
- The kin term map has been simplified.
NOW WE SHOULD BE ABLE TO Enter Generators to RECOMPOSE THE KIN TERMS but I don't see where to do that
- (If you reduce too far you cannot reconstruct)
[edit] Step-by Step Approach - Dwight Read 08:09, 12 October 2008 (PDT)
to me, Michael Fischer
Doug,
Intriguing! You went down a pathway that is logically possible but would end up with something like the Polish kinship terminology, not the AKT, due to a difference in the generating elements for these two terminologies. To clarify, let me indicate the steps one would follow to arrive at the AKT.
AKT simplification:
(1) Structural simplification Note that there are no sex marked terms in the resulting structure. Instead the reduced structure is based, implicitly, on Parent and Child kin terms (Parent = [Father, Mother] is the covering term for the structurally equivalent pair, Father/Mothter; Child = [Son, Daughter], and so on) -- the structural simplification identifies that Parent and Child, Not Father and Son (or Mother and Daugher) are the generating kin terms. We could, if we wanted, replace [Father, Mother] by Parent, [Son, Daughter] by Child, and so on, since "[Father, Mother]" is just a label and we can relabel the position with a different label without changing the structure.
(2) Remove Descendant terms.
At this point the reduced kin term map is simplified.
The algebra construction can now begin and in the algebra construction the algebra element P will be mapped to [Father, Mother]. (SEE BELOW XXX)
Your simplification:
(A) Structural Simplification -- the same as (1) above.
(B) Simplify Same Sex (Male) -- Since the label [Father, Mother] is constructed from the kin terms Father and Mother, one can remove the kin term Mother form [Father, Mother] and you are left with the kin term Father, and similar comments for the other [K, L] expressions. Note that doing Structural Simplification and then Simplify Same Sex (Male) is logically the same as just doing Simplify Same Sex (Male) on the original kin term map. (C) Remove Descendant terms. At this point you have a simplified kin term map, but one in which the generating term is Father.
The algebra construction can now begin and in the algebra construction the algebra element P will be mapped to Father. (SEE BELOW XXX)
NOTE: There is a glitch in the code that prevents the Construct Sex Structure from operating -- Mike needs to fix the glitch.
XXX
You write:
- Doug: Well, not quite. What I get under the GENERATORS window is
- Algebra Generators: P, I
- Transliteration of Generators: P: "Father", I: "Male Self"
- Kin Term Generators: Father Self
What I get under the EQUATIONS window is
- [x] I_=_
- [x] _I=_
At this point you have only entered the generators, I and P. When you click on Construct Graph, you get the algebra constructed from I and P. There are no equations at this point other than the equations that make I an identity element in the algebra. The graph will just be I --> P --> P^2 --> ....
So what you stated above is exactly correct.
When you tried "Find Recursive Equations." nothing happened because there is no recursive equation in the kin term map; that is, products of the generating kin term extend indefinitely (which makes me think that the wording should be "Find Reflexive Equations"). This is relevant in terminologies in which kin term products do not generate new kin terms indefinitely.
When you tried "Find Parent of Sib Equations." nothing happened because this does not occur in the kin term map since there is not sib generator.
So the next thing one can do (and which you did next) is to construct the "descendant" structure; i.e., the reciprocal structure.
The option Find Recursive Equations does nothing for the reason given above (its the same option as discussed above, but now searches the kin term map that is displayed in the window, whereas it only searched the simplified kin term map when selected earlier). The option "Add Reciprocal Equations" searches across all equations now in the algebra to determine whether the reciprocal form of an equation is in the algebra and if not, it adds the reciprocal equation to the algebra. Since the equation at this point is PC = I, the reciprocal equation is Cr Pr = I and Cr = P, Pr = C and so the reciprocal equation for PC = I is PC = I. Hence nothing happens at this point.
Until Mike gets the glitch fixed, it won't construct the sex marked terms.
This is the point where one either goes in the direction of the AKT or in the direction of the Polish Terminology.
There are 2 ways to add sex marked terms: (1) add two new elements labelled M and F (they are literally sex markeds for an algebra product) and structural equation they will satisfy are added automatically (this is how we get sex marking for the AKT) or (2) make an isomorphic copy of the algebra, with the original algebra the (say) male marked terms and the isomorphic copy the female marked terms. If the isomorphic copy has a distinction identity element, at this point we will have two, disjoint structures. The Polish terminology arises in this manner, with one identity element interpreted as a spouse element from the perspective of the othe identity element. That is, if the identity element in the male structure is I and the identitiy element in the female structure is i, then iI will be mapped to Wife (ms) and Ii will be mapped to Husband (ws).
Dwight
[edit] Structural Approach - also Dwight Read
Doug, It has been a while since I've run the program and I completely forgot that there are two pathways for doing the algebra operations. One is stage by stage and the other is by the steps involved in the structural stage.
The way I've discussed the algebra operations in my several emails has been the step-by-step approach. The structural stage approach is as follows.
I'll assume the AKT map has been simplified via structural equivalence and then via remove descendant structure.
Click on Construct Algebra, then click on Construct Base Algebra. This will automatically add an identity element, add a generating element, make the algebra, and show the graph of the algebra.
Next click on Construct Reciprocal Structure. This will make the make the reciprocal algebra, add the reciprocal equation, make the algebra, and graph it, automatically.
Next click on Construct Sex Structure. It will add the M and F sex generators, the appropriate equations, make the algebra and graph the algebra automatically. However, it does not show that the M and F generators have been added and it does not show the equations that have been added. This is a glitch. (In the step-by-step method a bigger glitch arises since the steps in Construct Sex Structure are not active.)
Next click on Construct Affinal Structure. It will add a Spouse generator, the appropriate equations will be added, the algebra will be constructed automatically. It will show that an affinal generator ("A") has been added and it will show the structural equations. It will say that the algebra and the kin term map are not isomorphic.
Next click on Apply Rules. Two rules will be added: One for sex marking of kin terms for the AKT (that is, kin term K is not sex marked if neither Spouse of K nor Spouse of Kr is a kin term; e.g. Cousin is not sex marked since Spouse of Cousin is not a kin term and the reciprocal of Cousin is Cousin, so Spouse of Reciprocal of Cousin is not a kin term) and the other is the rule for ith cousin j-times removed.
It will say that the algebra and the kin term map are not isomorphic. This is correct because the kin term map that is being used is not complete and has many missing kin terms (which will be listed in the dialogue box).
We did this deliberately to show that we can generate the missing kin terms so long as we have the basic structural information in the part of the kin term map that is being used.
Dwight
[edit] From Dwight Read (Earlier: In response to a first try, mistaken)
Doug, Glad to see that you are trying it out!
Try the following;
Click on Algebra Operations; Go to Steps: Base Algebra; click on Enter Identity; then go to Steps: Base Algebra; Click on Enter Generator;
- Doug: Right: But also, when Dwight says
- "Try the following;
- Click on Algebra Operations; Go to Steps: Base Algebra; click on Add Identity"
- 1 When I load AKT and Click on Algebra Operations; Go to Steps: Base Algebra; I get the message "The kin term map must be simplified before constructing the base algebra." Therefore I must figure out something else.
- 2 So the idea is to follow "The kin term map must be simplified before constructing the base algebra." So now I Click on Map Operations; click on Simplify Structurally Similar; and get under Diagnastics: "Affinal terms removed. Kin term map is complex. Cannot proceed without simplification of the kin terms map." (before constructing the base algebra).
- 3 So now I Click on Map Operations; click on Simplify Same Sex (Male); and get under Diagnastics: "simplification of kin term map. Kin term map is complex. Cannot proceed without simplification of the kin terms map." (before constructing the base algebra). Still, when I try (1) again, the kin term map is not simplified.
- 4 So now I Click on Map Operations; click on Remove Descendants. Under Dialogue I get "Kin term map has been simplified. NOW I Click on Algebra Operations; Go to Steps: Base Algebra; click on Add Identity. And success: Dialog now says: "Added the Identity element I"
- --Doug. Now:
When you select Enter Generator, the dialogue box will say "Added the generator, P". Then when you click on Construct Graph, the algebra will be generated and the various boxes will show the results of constructing the algebra. -- Dwight 09:32, 9 October 2008 (PDT)
- Doug: Well, not quite. What I get under the GENERATORS window is
- Algebra Generators: P, I
- Transliteration of Generators: P: "Father", I: "Male Self"
- Kin Term Generators: Father Self
What I get under the EQUATIONS window is
- [x] I_=_
- [x] _I=_
And the Map has not been expanded. So I am missing a step or two. Remember: we are given NO instructions pertinent to the menus other than a Help Window that explains ANALYZE - CONSTRUCT ALGEBRA - WRITE ALGEBRA = MODIFY ALGEBRA - DRAW GRAPH. My bet is to go back to Algebra Operations; Go to Steps: Base Algebra; Find Recursive Equations. (The alternative is Find Parent to Sib Equations.) I did both in sequence with no result.
- So I went to a logical possiblility: go to Algebra Operations; click Steps: Reciprocal Structure; and click the only current choice: Make Reciprocal Algebra.
- AH! This is cool, I have rebuilt a good deal of the structure, the reciprocals.
- Am getting the hang of this, but there are lots of choices. How about: go to Algebra Operations; click Steps: Reciprocal Structure; click add Reciprocal Element Equations (This did nothing to the map that I could see).
- (I might have preferred this: click Steps: Add Equivalences. But it was not an option)
- So I went to Steps: Sibling Structure; Enter Sib Generators. Still did nothing.
- So I tried Steps: Add Affinal Structure; Test for Spouse Generator. (The Dialogue has said twice not "The Focal element of the algebra is I. The algebra does not have a spouse element)
- There is nothing left to do now but Steps: Reciprocal Structure: and either
- Find Recursive Equations or
- Add Reciprocal Equations. Nothing happens in either case.
I have now exhausted all possibilities but not gotten a complete reconstruction. Seem to have missed the females somehow. Am asking Dwight to edit so we can follow the full reconstruction.
IGNORE ALL THE REST HERE Small graph will show in Green window. The list of generators, equations, etc. will show in the small windows labeled Generators, Equations. To see a large graph, go to Window, click on Graph. In the graph view, click on Kin term symbols to see the mapping from algebra symbols to kin terms deduced by comparing the structure of the algebra to the structure of the kin term map. You can move the graph by clicking on it and moving the mouse.
Click on genealogical to get a mapping of the algebra elements (or the kin term symbols) onto a genealogical grid.
Click on Algebra Operations; go to Steps: Reciprocal Structure; click on Make Reciprocal Algebra;
Go to Algebra Operations; click on Construct Graph (it is active even if it is currently checked). You will get a fractal structure since the reciprocal relation between a generator for the ascending structure and its isomorphic symbol for the descending structure has not yet been entered.
Go to Algebra Operations; click on Steps: Reciprocal Structure; click on Add Reciprocal Element Equations
Go to Algebra Operations; click on Construct Graph ; you will now get the algebra structure isomorphic to the structurally reduced terminology.
All algebra operations that are logically possible (whether relevant to the terminology in question) are active since the algebra construction is independent of the kin term map (except for things such as number of generators, and certain structural equations such as equations that limit the extent of the kin term structure)
Dwight
[edit] References
see Dwight Read
Robbins Burling. 1970. American Kinship Terms Once More. Southwestern Journal of Anthropology 26(1): 15-24. One of the early proponents of relative products of kinship terms, e.g., Cousin = Child of Uncle or Aunt (English terms).
Read, Dwight W. 2006. Kinship Algebra Expert System (KAES): A Software Implementation of a Cultural Theory. eScholarship Repository, University of California, University of California. Originally published in Social Science Computer Review 24(1):43-67, © 2006 Sage Publications.
