More Like This Technical Calculations
The workbench More Like This module calculations are based on signifier values being close to the signification of the chosen fragment.
The basic approach is outlined as follows:
- The fragment selected to find more like it, is taken from the dataset and its triad values noted. This is the set-aside fragment
- For each of the remaining Workbench fragments, for each triad, the distance between it and the set-aside triad is calculated.
- If the distance between this fragments triads and the set-aside fragments triads fits within a set threshold, then that fragment is included in the more like this, otherwise it is excluded.
Triad Distance Calculation
(note - diagrams will be added once image uploads are available)
Triad data is compositional data. This means that the value for each anchor sums to 1 (or 100% if represented in percentages) and thus reflect proportions not absolute values.
The anchors are the left (L), top (T) and right (R) corners. In any instance, their meanings are defined by the text label (or symbol) defined at each anchor.
The position of a specific triad value instance in a ternary diagram (shown as a dot in the triad page of the Workbench) indicates the proportion of L, T and R for particular triad placement.
it is possible to calculate the distance between two or more placements on a triad - that is the distance between 2 or more dots placed on the triad as seen in the Workbench triad page.
But this distance calculation within a triad