Section 1-4 FGRA CORPUS ANALYSIS



Individuals on the family group sheet.   A family group sheet of the family group record archives (FGRA) contains spaces for recording a good deal of identifying information on the members of one sibship and some on members of others. Figure 1 below diagrams the genealogical relationship of the various individuals on a compiled family group sheet of the FGRA.

Figure 1  Symbols for Individuals on a Family Group Record

Central to the sheet are the husband (H), a wife (W), and their children. Children are distinguished by sex so that they are usually either male (S) or female (D). However, there is also space for certain additional dates and names for people related to these family members by marriage, as well as spaces for names of the parents of the husband and wife (K and Q). These additional spaces are to identify other individuals as relatives. How many roles or positions there are depends on how you decide to count them; they are: parents of husband and wife (K, Q), other spouses of husband and wife (O), primary spouse of child (G), and other spouses of children (GO). Each space allows for a differing amount of information. Figure 2 lays out some of the conventional symbols in use as convenient ways to referring to the various identifiers recorded on a family group sheet.

Figure 2  Identifying Individuals in a Nuclear Family

Families on the family group sheet.   Since there is no difference between the amount of information between G and GO, it should be possible to let the family group record engine handle them in the model as the same class of object. Thus there are only four classes of families represented as laid out on figure 3: 1) two individuals “maximally” identified (name, birth, christening, marriage, death, burial events), 2) names only, 3) individual, marriage date, spouse (sequence, name), 4) individual, marriage date, spouse (sequence, name).

Figure 3  Four Classes of Nuclear Families on the Family Group Sheet

The symbolization of the above figures 1 and 2 groups the attributes of sex, names, events (birth, death, etc.), having dates, places, etc., into the various individuals represented. Figure 1 symbolizes nine families (each resulting from the union of two individuals) using the equal signs (=). Each family groups together the attributes of certain events (marriage, divorce, etc.) and the husband and wife involved. The classes in figure 3 are indicated by one possible symbolization of these families of figure 1: K=Q, H=W, S=G, D=G, etc.

Linking family group sheets.   The fundamental perspective of record linkage is the establishment of two files of records for the entities to be linked. The simplest procedure in linking family records is a "2 2" strategy. We first set up two sets of two files each containing records of individuals. In the first set we have one file for all the male children (S) in an O-family, and one file for all the female children (D). We want to link these two files to corresponding files of individuals in a second set: one file for all the husbands (H) in each P-family, and one file for all the wives (W). This algorithm then requires a transformation of S to H and D to W, or vice versa. There are three additional kinds of linkage involving the individuals on the family group sheet. We discuss these in more detail in the following section.
123456789
1KHKH/QHQHHH/OOW
2KHKH/QHQHHH/WW
3HH/WWSS/GGF
4GMG/DDWH/WH
5HH/WWQWKW/QWKW
6OHO/WWQWKW/QWKW
Table 1 — Transformation Matrix; FGRA Model

It will probably help to look at all the symbols of figure 3 and account for them in record linkage, as we did above for children and parents in two families. It may be clearer to place them in a matrix arranged so that elements along a row will transform to the elements in any other row, according to the column it is in. Thus in table 1 the S in row 3 to H in row 2 is symmetric with the D in row 4 to the W in row 5. Each row is a family of transformations. Row 3 to row 2 means that when we discover that an S in one group is the same as an H in another, we also discover that his corresponding GF in the one group is the W on the other, and that the W and H of the first group are entered as the QW and KW on the other.

Individual linkage entities.   The transformations allow the family group model and engine to build searches based on the various individuals in the family group. The record linkage system should only weight comparisons between records that make sense to compare — that could possibly represent the same entity. In other words, the individual entities to be compared must be in the same family of orientation and to some extent have non-overlapping, etc., families of propagation.