Package com.tnsai.server.rag.retrieval

Class BM25Stream

java.lang.Object
com.tnsai.server.rag.retrieval.BM25Stream
All Implemented Interfaces:
RetrievalStream
public final class BM25Stream extends Object implements RetrievalStream
BM25-based keyword retrieval, bounded by a Caffeine LRU cache (issue #36). See the issue for rationale; the index is now a bounded LRU window rather than a complete in-memory index.

  • Field Details

    • DEFAULT_MAXIMUM_SIZE

      public static final int DEFAULT_MAXIMUM_SIZE
      See Also:

    • DEFAULT_EXPIRE_AFTER_ACCESS

      public static final Duration DEFAULT_EXPIRE_AFTER_ACCESS

  • Constructor Details

    • BM25Stream

      public BM25Stream()

    • BM25Stream

      public BM25Stream(int maximumSize, Duration expireAfterAccess)

  • Method Details

    • name

      public String name()
      Specified by:
      name in interface RetrievalStream

    • getMaximumSize

      public int getMaximumSize()

    • size

      public long size()

    • addDocument

      public void addDocument(com.tnsai.knowledge.Document doc)

    • removeDocument

      public void removeDocument(String docId)

    • replaceDocuments

      public void replaceDocuments(Collection<String> removeIds, List<com.tnsai.knowledge.Document> add)
      Removes then adds documents as one generation under the index lock.
      Parameters:
      removeIds - document identifiers to drop
      add - documents to index after the removals

    • clear

      public void clear()

    • retrieve

      public List<com.tnsai.knowledge.SearchResult> retrieve(String query, int topK)
      Specified by:
      retrieve in interface RetrievalStream