Currently in any given page, the area of the speech bubbles and the area of the shortest sentences in the given terms are sorted by size and matched together. This is obviously a very inexact and rough way of trying to get the right size text content in the applicable speech bubble.
Options:
-
Take two arrays (with terms, an array incorporating all sentences in specific set of terms) and apply some form of k-means matching (but not sure how to k-means two separate arrays as opposed to values in one array. Only problem is that this matching may still be rough.
-
Scrape terms from multiple sites and sources to create database of sentences and the length/area (assuming average font size) and then when given a speech bubble area, search that database to return appropriate size of text.
-
Give up and do something productive
Currently in any given page, the area of the speech bubbles and the area of the shortest sentences in the given terms are sorted by size and matched together. This is obviously a very inexact and rough way of trying to get the right size text content in the applicable speech bubble.
Options:
Take two arrays (with terms, an array incorporating all sentences in specific set of terms) and apply some form of k-means matching (but not sure how to k-means two separate arrays as opposed to values in one array. Only problem is that this matching may still be rough.
Scrape terms from multiple sites and sources to create database of sentences and the length/area (assuming average font size) and then when given a speech bubble area, search that database to return appropriate size of text.
Give up and do something productive