Category Confusion Subtab
This topic describes part of the functionality of Genesys Content Analyzer.
This subtab, shown in "Models Tab: Category Confusion Subtab," lists up to 10 pairs of categories that the model is likely to confuse.
The Confusion column gives the probability (between 0 and 1) of confusion between the categories in the Category 1 and Category 2 columns. A rating of 0.5 would mean total confusion: the model cannot distinguish A from B. A rating of 1.0 would mean that the model always calls A B and always calls B A —a complete reversal.
If a pair of categories has a rating of over 0.2 and both categories have more than three or four members, you should consider modifying them. You can modify them in either of two directions:
- Merge them (decide that they are so similar they amount to a single category).
- Further differentiate them by adding more highly contrasting e-mails to them in the training object.