Implementation of the LDA topic model in Squeak. Useful for natural language processing and code mining tasks.
'LDAllocator' consumes a set of "documents" (arrays of numbers between 1 to n), computes a set of "topics" (distributions over those numbers) and assigns to each document a distribution of topics. Topics and topic-document-assignments are chosen to maximize the probability of all documents being sampled, which results in orthogonal "concerns" being isolated.