R/topic_flow.R
Accepts two topic term matrices generated by different LDA runs.
Topics may have different terms in their respective vocabulary.
Non-existent terms will be added with probability 0 to each other.
See CalculateTopicFlow for an example of it's usage.
CalculateHighestTopicCosineSimilarity(ttm1, ttm2)
| ttm1 | A topic term matrix |
|---|---|
| ttm2 | A topic term matrix (e.g. of the following month) |
Returns a cartesian product data.table of the ttm pair containing the topics with maximum likehood.