mixle.represent.posterior module

Posterior retrieval by fitted-model affinity rather than raw-feature cosine.

Fit a mixture to heterogeneous records and retrieval similarity becomes posterior affinity: two records are close when the model’s field-restricted latent posteriors agree. The implementation uses the per-field Bhattacharyya-style balanced affinity from mixle.utils.hvis, with an evidence cap so one inconsistent field can contribute negative evidence without dominating every other field. Raw-feature cosine has neither property: it weights fields by numeric scale, and one high-variance field can dominate the dot product:

m = mixle.propose(records, fit=True)
r = PosteriorRetriever(m.fitted, records)          # any mixture over the records works
r.retrieve(query, k=5)                             # [(corpus index, log-affinity), ...]

Cost note: affinities are computed jointly over corpus + queries through the model’s per-field likelihoods. Model passes are linear in rows, while the affinity block is quadratic, so this is intended for moderate corpora. For large-corpus first-stage recall, use mixle.represent.fit_embedder() and rerank the shortlist here.

class PosteriorRetriever(model, corpus, *, evidence_cap=1.0, field_weights=None)

Bases: object

Retrieve over raw heterogeneous records by the fitted mixture’s posterior affinity.

Parameters:

• model (Any)

• corpus (Any)

• evidence_cap (float | None)

• field_weights (Any)

affinity_matrix()

The corpus’s dense (n, n) log-affinity matrix (diagonal -inf).

Return type:

ndarray

retrieve(query, k=5)

Top-k corpus records for one query: [(corpus_index, log_affinity), ...] best first.

Parameters:

• query (Any)

• k (int)

Return type:

list[tuple[int, float]]

retrieve_batch(queries, k=5)

Top-k per query, computed in one joint pass over corpus + queries.

Parameters:

• queries (Any)

• k (int)

Return type:

list[list[tuple[int, float]]]