I imagine this is a variant of a Latent Dirichlet Allocation[0]. Informally, this will identify words "most associated" with 2013, which is not the same thing a

I imagine this is a variant of a Latent Dirichlet Allocation[0]. Informally, this will identify words "most associated" with 2013, which is not the same thing as the most searched words in 2013. Put another way, unless there's a huge spike of interest in the weather in 2013 compared to other years, "weather" won't be included. [0] https://en.wikipedia.org/wiki/Latent_dirichlet_allocation https://en.wikipedia.org/wiki/Latent_dirichlet_allocation