align_graph.RdThis provides a more generic interface to alignment between arbitrary pairs of topics, compared to `align_topics`. Rather than requiring sequential or all-vs-all comparisons, this function supports comparisons between any pairs of models, as specified by the `edges` parameter. Any graph linking pairs of models can be the starting point for an alignment.
align_graph(edges, gamma_hats, beta_hats, weight_fun, ...)A data frame with two columns, $from and $to, giving the names of the models to be aligned. These names must be the names of the lists in `gamma_hats` and `beta_hats`.
A named list of matrices, giving estimated mixed-membership parameters across a collection of topic models. The names of this list must correspond to the names of models to compare in `edges`.
A named list of matrices, giving estimated topic parameters across a collection of topic models. The names of this list must correspond to the names of models to compare in `edges`.
A function that returns a data.frame giving weights between all pairs of topics between two models. The first argument must accept a list of two gamma_hat matrices, the second argument must accept a list of two beta_hat matrices. See `product_weights` or `transport_weights` for examples.
(optional) Further keyword arguments passed to the weight
function. For example, passing reg = 10 when using the
transport method will use a regularization level fo 10 in the Sinkhorn
optimal transport algorithm.