Advanced Reranker with Learned Weights

Combines multiple signals for better reranking:

• Semantic similarity (word vectors)

• BM25/keyword overlap

• Query coverage

• Position bias

• Length normalization

Public fields

weights

Feature weights

Methods

Public methods

• AdvancedReranker$new()

• AdvancedReranker$set_embedder()

• AdvancedReranker$rerank()

• AdvancedReranker$learn_weights()

• AdvancedReranker$clone()

---

Method new()

Create a new AdvancedReranker

Usage

AdvancedReranker$new(
  semantic_weight = 0.4,
  bm25_weight = 0.3,
  coverage_weight = 0.2,
  position_weight = 0.1,
  sentence_embedder = NULL
)

Arguments

semantic_weight

Weight for semantic similarity (0-1)

bm25_weight

Weight for BM25 score (0-1)

coverage_weight

Weight for query term coverage (0-1)

position_weight

Weight for position bias (0-1)

sentence_embedder

Optional SentenceEmbedder for semantic scoring

---

Method set_embedder()

Set sentence embedder

Usage

AdvancedReranker$set_embedder(embedder)

Arguments

embedder

SentenceEmbedder object

---

Method rerank()

Rerank results

Usage

AdvancedReranker$rerank(
  query,
  query_vector = NULL,
  results,
  doc_vectors = NULL,
  limit = 10
)

Arguments

query

Query text

query_vector

Query embedding vector

results

List of result objects with id, text, score

doc_vectors

Matrix of document vectors (optional)

limit

Number of results to return

Returns

Reranked list of results

---

Method learn_weights()

Learn optimal weights from relevance judgments

Usage

AdvancedReranker$learn_weights(
  queries,
  results_list,
  relevance_list,
  iterations = 100
)

Arguments

queries

Character vector of queries

results_list

List of result lists (one per query)

relevance_list

List of relevance scores (1=relevant, 0=not)

iterations

Number of optimization iterations

---

Method clone()

The objects of this class are cloneable with this method.

Usage

AdvancedReranker$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.