pairs <- expand.grid(x = seq_len(nrow(dta)), y = seq_len(nrow(dta))) %>% 
  # Only need to compare i to  all records j, with j > i 
  filter(y > x) %>%
  left_join(dta, by = c(x = 'i')) %>% rename(cityx = city, nx = n) %>%
  left_join(dta, by = c(y = 'i')) %>% rename(cityy = city, ny = n) %>%
  mutate(similarity = stringsim(cityx, cityy, method = "jw")) %>% 
  arrange(desc(similarity))