Persona Similarity Analysis
Pairwise comparison of synthetic personas in 6-dimensional feature space. Similarity computed via cosine distance.
Most Similar Pair
Least Similar Pair
All Pairwise Comparisons
| Rank | Persona A | Persona B | Similarity | Distance |
|---|---|---|---|---|
| 1 | Emoji Enthusiast | Technical Researcher | 0.9682 | 0.0318 |
| 2 | Expressive English | Emoji Enthusiast | 0.9752 | 0.0248 |
| 3 | Balanced Bilingual | Emoji Enthusiast | 0.9767 | 0.0233 |
| 4 | Concise Mandarin | Emoji Enthusiast | 0.9844 | 0.0156 |
| 5 | Bilingual Early Adopter | Emoji Enthusiast | 0.9855 | 0.0145 |
| 6 | Concise Mandarin | Technical Researcher | 0.9940 | 0.0060 |
| 7 | Concise Mandarin | Expressive English | 0.9957 | 0.0043 |
| 8 | Bilingual Early Adopter | Technical Researcher | 0.9963 | 0.0037 |
| 9 | Concise Mandarin | Balanced Bilingual | 0.9973 | 0.0027 |
| 10 | Bilingual Early Adopter | Expressive English | 0.9983 | 0.0017 |
| 11 | Bilingual Early Adopter | Concise Mandarin | 0.9984 | 0.0016 |
| 12 | Bilingual Early Adopter | Balanced Bilingual | 0.9988 | 0.0012 |
| 13 | Balanced Bilingual | Technical Researcher | 0.9992 | 0.0008 |
| 14 | Expressive English | Technical Researcher | 0.9995 | 0.0005 |
| 15 | Expressive English | Balanced Bilingual | 0.9998 | 0.0002 |
Distance = 1 - Similarity. Lower distance = more similar personas. n=15 pairs from 6 personas.
Feature Space
Each persona is represented as a vector in 6-dimensional space:
Average Length
Mean message length in characters
Emoji Rate
Emoji per turn
Latin Share
Proportion of Latin words (code-switching)
Bubbles Per Turn
Messages sent in each turn
Laughter Rate
Laughter tokens per turn
No Punctuation Rate
Proportion of turns without end punctuation
Cosine Similarity Formula
sim(A, B) = (A · B) / (||A|| × ||B||)
Cosine of the angle between two vectors. Range: -1 (opposite) to +1 (identical). All personas in this dataset have positive similarity (same general direction in feature space).