Package 'VectrixDB'

Title: Lightweight Vector Database with Embedded Machine Learning Models
Description: A lightweight vector database for text retrieval in R with embedded machine learning models and no external API (Application Programming Interface) keys. Supports dense and hybrid search, optional HNSW (Hierarchical Navigable Small World) approximate nearest-neighbor indexing, faceted filters with ACL (Access Control List) metadata, command-line tools, and a local dashboard built with 'shiny'. The HNSW method is described by Malkov and Yashunin (2018) <doi:10.1109/TPAMI.2018.2889473>.
Authors: Kwadwo Daddy Nyame Owusu Boakye [aut, cre]
Maintainer: Kwadwo Daddy Nyame Owusu Boakye <[email protected]>
License: Apache License (>= 2)
Version: 1.1.2
Built: 2026-05-22 18:51:35 UTC
Source: https://github.com/knowusuboaky/vectrixdb-r

Help Index

Create ACL Config from List

Description

Create ACLConfig from list of ACL strings

Usage

acl_config_from_list(acl_list)

Arguments

acl_list

Character vector of ACL strings

Value

ACLConfig object

ACL Configuration

Description

ACL configuration for a document or collection

Public fields

read_principals

Who can read

deny_principals

Who cannot read (takes precedence)

is_public

Is public access allowed

Methods

Public methods

Method new()

Create a new ACLConfig

Usage
ACLConfig$new(
  read_principals = list(),
  deny_principals = list(),
  is_public = FALSE
)
Arguments
read_principals

List of ACLPrincipal objects

deny_principals

List of ACLPrincipal objects

is_public

Logical

Method clone()

The objects of this class are cloneable with this method.

Usage
ACLConfig$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

ACL Filter

Description

Access Control List filter for security-aware search

Public fields

acl_field

Metadata field containing ACLs

Methods

Public methods

Method new()

Create a new ACLFilter

Usage
ACLFilter$new(acl_field = "_acl")
Arguments
acl_field

Field name for ACLs (default: "_acl")

Method filter()

Filter documents based on user's ACL principals

Usage
ACLFilter$filter(documents, user_principals, default_allow = FALSE)
Arguments
documents

List of documents with metadata

user_principals

Character vector or list of ACLPrincipal

default_allow

Allow if no ACL defined (default: FALSE)

Returns

Filtered documents

Method add_acl()

Add ACL to document metadata

Usage
ACLFilter$add_acl(metadata, principals)
Arguments
metadata

Document metadata

principals

Character vector of principal strings

Returns

Updated metadata

Method create_filter_condition()

Create ACL filter condition for query

Usage
ACLFilter$create_filter_condition(user_principals)
Arguments
user_principals

Character vector of principals

Returns

Filter condition list

Method clone()

The objects of this class are cloneable with this method.

Usage
ACLFilter$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Examples

## Not run: 
acl_filter <- ACLFilter$new()
filtered <- acl_filter$filter(
  documents = results,
  user_principals = c("user:alice", "group:engineering")
)

## End(Not run)

ACL Operator Types

Description

ACL matching operators

Usage

ACLOperator

Format

An object of class list of length 5.

ACL Principal

Description

An ACL principal (user, group, or role)

Public fields

type

Principal type

value

Principal value

Methods

Public methods

Method new()

Create a new ACLPrincipal

Usage
ACLPrincipal$new(type, value)
Arguments
type

Principal type (user, group, role)

value

Principal value

Method matches()

Check if this principal matches another

Usage
ACLPrincipal$matches(other)
Arguments
other

Another ACLPrincipal

Returns

Logical

Method to_string()

Convert to string

Usage
ACLPrincipal$to_string()

Method clone()

The objects of this class are cloneable with this method.

Usage
ACLPrincipal$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

VectrixDB Advanced Search Features

Description

Enterprise-grade search capabilities:

  • Faceted search with aggregations

  • ACL/Security filtering

  • Text analyzers (stemming, synonyms, stopwords)

Advanced Reranker with Learned Weights

Description

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

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.

Analyzer Chain

Description

Chain multiple analyzers together

Public fields

analyzers

List of TextAnalyzer objects

Methods

Public methods

Method new()

Create a new AnalyzerChain

Usage
AnalyzerChain$new(analyzers)
Arguments
analyzers

List of TextAnalyzer objects

Method analyze()

Run text through all analyzers

Usage
AnalyzerChain$analyze(text)
Arguments
text

Input text

Returns

Character vector of tokens

Method clone()

The objects of this class are cloneable with this method.

Usage
AnalyzerChain$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Base Cache

Description

Abstract base class for cache backends

Public fields

config

Cache configuration

stats

Cache statistics

Methods

Public methods

Method new()

Create a new cache

Usage
BaseCache$new(config = NULL)
Arguments
config

CacheConfig object

Method get()

Get a value from cache

Usage
BaseCache$get(key)
Arguments
key

Cache key

Returns

Cached value or NULL

Method set()

Set a value in cache

Usage
BaseCache$set(key, value, ttl = NULL)
Arguments
key

Cache key

value

Value to cache

ttl

Time to live (optional)

Method delete()

Delete a key from cache

Usage
BaseCache$delete(key)
Arguments
key

Cache key

Returns

Logical success

Method exists()

Check if key exists

Usage
BaseCache$exists(key)
Arguments
key

Cache key

Returns

Logical

Method clear()

Clear all cache entries

Usage
BaseCache$clear()

Method size()

Get cache size

Usage
BaseCache$size()
Returns

Integer count

Method get_many()

Get multiple values

Usage
BaseCache$get_many(keys)
Arguments
keys

Character vector of keys

Returns

Named list of values

Method set_many()

Set multiple values

Usage
BaseCache$set_many(items, ttl = NULL)
Arguments
items

Named list of values

ttl

Time to live

Method delete_many()

Delete multiple keys

Usage
BaseCache$delete_many(keys)
Arguments
keys

Character vector of keys

Returns

Integer count of deleted keys

Method make_key()

Make a prefixed key

Usage
BaseCache$make_key(key)
Arguments
key

Raw key

Returns

Prefixed key

Method clone()

The objects of this class are cloneable with this method.

Usage
BaseCache$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

VectrixDB Cache Layer

Description

High-performance caching for low latency

Supports multiple cache backends:

  • InMemory LRU: Ultra-fast, limited by RAM

  • File-based: Persistent cache using RDS files

Create Cache Config from Environment

Description

Create config from environment variables

Usage

cache_config_from_env()

Value

CacheConfig object

Cache Backend Types

Description

Available cache backends

Usage

CacheBackend

Format

An object of class list of length 3.

Cache Configuration

Description

Configuration for cache layer

Public fields

backend

Cache backend type

memory_max_size

Max items in memory

memory_ttl_seconds

Default TTL in seconds

file_cache_dir

Directory for file cache

file_ttl_seconds

File cache TTL

prefix

Cache key prefix

compression

Use compression

Methods

Public methods

Method new()

Create a new CacheConfig

Usage
CacheConfig$new(
  backend = "memory",
  memory_max_size = 10000,
  memory_ttl_seconds = 3600,
  file_cache_dir = NULL,
  file_ttl_seconds = 86400,
  prefix = "vectrix:",
  compression = TRUE
)
Arguments
backend

Backend type

memory_max_size

Max memory items

memory_ttl_seconds

Memory TTL

file_cache_dir

File cache directory

file_ttl_seconds

File TTL

prefix

Key prefix

compression

Use compression

Method clone()

The objects of this class are cloneable with this method.

Usage
CacheConfig$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Cache Entry

Description

A cached entry with metadata

Public fields

value

Cached value

created_at

Creation timestamp

ttl

Time to live in seconds

hits

Number of hits

Methods

Public methods

Method new()

Create a new CacheEntry

Usage
CacheEntry$new(value, ttl)
Arguments
value

The value to cache

ttl

Time to live in seconds

Method is_expired()

Check if entry is expired

Usage
CacheEntry$is_expired()
Returns

Logical

Method clone()

The objects of this class are cloneable with this method.

Usage
CacheEntry$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Cache Statistics

Description

Cache statistics for monitoring

Public fields

hits

Cache hits

misses

Cache misses

sets

Cache sets

deletes

Cache deletes

evictions

Cache evictions

Methods

Public methods

Method record_hit()

Record a cache hit

Usage
CacheStats$record_hit()

Method record_miss()

Record a cache miss

Usage
CacheStats$record_miss()

Method record_set()

Record a cache set

Usage
CacheStats$record_set()

Method record_delete()

Record a cache delete

Usage
CacheStats$record_delete()

Method record_eviction()

Record a cache eviction

Usage
CacheStats$record_eviction()

Method hit_rate()

Get hit rate

Usage
CacheStats$hit_rate()
Returns

Numeric hit rate

Method to_list()

Convert to list

Usage
CacheStats$to_list()
Returns

List representation

Method reset()

Reset statistics

Usage
CacheStats$reset()

Method clone()

The objects of this class are cloneable with this method.

Usage
CacheStats$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

VectrixDB Command Line Interface

Description

CLI tools for VectrixDB operations

Provides command-line style functions for:

  • Creating and managing collections

  • Adding and searching documents

  • Exporting and importing data

  • Database statistics and info

CLI Configuration

Description

Configuration for CLI behavior

Public fields

verbose

Print verbose output

color

Use colored output

data_dir

Default data directory

Methods

Public methods

Method new()

Create CLI config

Usage
CLIConfig$new(verbose = TRUE, color = TRUE, data_dir = NULL)
Arguments
verbose

Verbose output

color

Colored output

data_dir

Data directory

Method clone()

The objects of this class are cloneable with this method.

Usage
CLIConfig$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Collection Class

Description

Vector collection with indexing and search

Public fields

name

Collection name

dimension

Vector dimension

metric

Distance metric

language

Language setting ("en" or "ml")

Methods

Public methods

Method new()

Create a new Collection

Usage
Collection$new(
  name,
  dimension,
  metric = "cosine",
  storage = NULL,
  language = "en"
)
Arguments
name

Collection name

dimension

Vector dimension

metric

Distance metric

storage

Storage backend

language

Language behavior ("en" = ASCII-focused, "ml" = Unicode-aware)

Method add()

Add documents to collection

Usage
Collection$add(ids, vectors, metadata = NULL, texts = NULL)
Arguments
ids

Document IDs

vectors

Matrix of vectors

metadata

List of metadata

texts

Character vector of texts

Method search()

Search collection

Usage
Collection$search(query, limit = 10, filter = NULL, include_vectors = FALSE)
Arguments
query

Query vector

limit

Number of results

filter

Metadata filter

include_vectors

Include vectors in results

Returns

Results object

Method keyword_search()

Keyword search

Usage
Collection$keyword_search(query_text, limit = 10, filter = NULL)
Arguments
query_text

Query text

limit

Number of results

filter

Metadata filter

Returns

Results object

Method hybrid_search()

Hybrid search (dense + sparse)

Usage
Collection$hybrid_search(
  query,
  query_text,
  limit = 10,
  vector_weight = 0.5,
  text_weight = 0.5,
  filter = NULL,
  include_vectors = FALSE,
  rrf_k = 60,
  prefetch_multiplier = 10
)
Arguments
query

Query vector

query_text

Query text

limit

Number of results

vector_weight

Weight for vector search

text_weight

Weight for text search

filter

Metadata filter

include_vectors

Include vectors in results

rrf_k

RRF constant

prefetch_multiplier

Prefetch multiplier

Returns

Results object

Method get()

Get documents by ID

Usage
Collection$get(ids)
Arguments
ids

Document IDs

Returns

List of results

Method delete()

Delete documents by ID

Usage
Collection$delete(ids)
Arguments
ids

Document IDs to delete

Method count()

Get document count

Usage
Collection$count()
Returns

Integer count

Method clear()

Clear collection

Usage
Collection$clear()

Method clone()

The objects of this class are cloneable with this method.

Usage
Collection$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Community

Description

A community of entities

Public fields

id

Community ID

level

Hierarchy level

entity_ids

Member entity IDs

summary

Community summary

parent_id

Parent community ID

child_ids

Child community IDs

Methods

Public methods

Method new()

Create a new Community

Usage
Community$new(
  id,
  level = 0,
  entity_ids = character(0),
  summary = NULL,
  parent_id = NULL
)
Arguments
id

ID

level

Level

entity_ids

Members

summary

Summary

parent_id

Parent

Method size()

Get size

Usage
Community$size()
Returns

Integer

Method clone()

The objects of this class are cloneable with this method.

Usage
Community$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Simple Community Detector

Description

Detects communities using connected components

Public fields

min_size

Minimum community size

max_levels

Maximum hierarchy levels

Methods

Public methods

Method new()

Create a new CommunityDetector

Usage
CommunityDetector$new(min_size = 5, max_levels = 3)
Arguments
min_size

Min size

max_levels

Max levels

Method detect()

Detect communities in graph

Usage
CommunityDetector$detect(graph)
Arguments
graph

KnowledgeGraph object

Returns

List of Community objects

Method clone()

The objects of this class are cloneable with this method.

Usage
CommunityDetector$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Create Cache

Description

Factory function to create cache backend

Usage

create_cache(config = NULL)

Arguments

config

CacheConfig object or NULL for defaults

Value

Cache object

Create Default GraphRAG Config

Description

Create default config with regex extractor

Usage

create_default_graphrag_config(...)

Arguments

...

Additional options

Value

GraphRAGConfig object

Create HNSW Index

Description

Factory function to create HNSW index

Usage

create_hnsw_index(dimension, metric = "angular", n_trees = 50)

Arguments

dimension

Vector dimension
metric

Distance metric
n_trees

Number of trees

Value

HNSWIndex object

Create GraphRAG Pipeline

Description

Factory function for GraphRAGPipeline

Usage

create_pipeline(config = NULL)

Arguments

config

GraphRAGConfig (optional)

Value

GraphRAGPipeline object

Create a sentence embedder with automatic download

Description

Convenience function to create a SentenceEmbedder with GloVe vectors

Usage

create_sentence_embedder(model = "glove-100", use_idf = TRUE)

Arguments

model

Model name (default: "glove-100")
use_idf

Use IDF weighting

Value

SentenceEmbedder object

Examples

## Not run: 
# Downloads GloVe if not present
embedder <- create_sentence_embedder("glove-100")

# Embed texts
vectors <- embedder$embed(c("Hello world", "Machine learning is cool"))

## End(Not run)

Create Vector Cache

Description

Create a VectorCache with specified backend

Usage

create_vector_cache(backend = "memory", ...)

Arguments

backend

Backend type: "memory", "file", or "none"
...

Additional config options

Value

VectorCache object

Dense Embedder using word2vec or GloVe

Description

Generates dense vector embeddings using pre-trained word vectors

Public fields

dimension

Embedding dimension

model_type

Type of model being used

language

Language setting ("en" or "ml")

Methods

Public methods

Method new()

Create a new DenseEmbedder

Usage
DenseEmbedder$new(
  dimension = 100,
  model_path = NULL,
  model_type = "tfidf",
  sentence_embedder = NULL,
  auto_download = FALSE,
  language = "en"
)
Arguments
dimension

Vector dimension (default: 100 for word2vec, 50/100/200/300 for GloVe)

model_path

Optional path to pre-trained model file

model_type

Type: "word2vec", "glove", "glove-pretrained", or "tfidf"

sentence_embedder

Optional SentenceEmbedder object to use

auto_download

Auto-download GloVe vectors if model_type is glove-pretrained

language

Language behavior ("en" = ASCII-focused, "ml" = Unicode-aware)

Method set_sentence_embedder()

Set a SentenceEmbedder to use for embeddings

Usage
DenseEmbedder$set_sentence_embedder(embedder)
Arguments
embedder

SentenceEmbedder object

Method embed()

Embed texts to vectors

Usage
DenseEmbedder$embed(texts)
Arguments
texts

Character vector of texts

Returns

Matrix of embeddings (rows are documents)

Method fit()

Train embedder on corpus (for TF-IDF)

Usage
DenseEmbedder$fit(texts)
Arguments
texts

Character vector of training texts

Method clone()

The objects of this class are cloneable with this method.

Usage
DenseEmbedder$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Distance Metric Enumeration

Description

Available distance metrics for vector comparison

Usage

DistanceMetric

Format

An object of class list of length 4.

Document Chunker

Description

Splits documents into text units

Public fields

chunk_size

Target chunk size

chunk_overlap

Overlap size

by_sentence

Preserve sentences

Methods

Public methods

Method new()

Create a new DocumentChunker

Usage
DocumentChunker$new(chunk_size = 1200, chunk_overlap = 100, by_sentence = TRUE)
Arguments
chunk_size

Target size

chunk_overlap

Overlap

by_sentence

Preserve sentences

Method chunk()

Chunk a document

Usage
DocumentChunker$chunk(text, document_id = NULL)
Arguments
text

Document text

document_id

Document ID

Returns

List of TextUnit objects

Method clone()

The objects of this class are cloneable with this method.

Usage
DocumentChunker$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Download pre-trained word vectors

Description

Download GloVe or other pre-trained word vectors

Usage

download_vectors(model = "glove-50", dest_dir = NULL)

Arguments

model

Model name: "glove-50", "glove-100", "glove-200", "glove-300"
dest_dir

Destination directory

Value

Path to downloaded model

Download pre-trained word vectors

Description

Downloads GloVe or fastText word vectors

Usage

download_word_vectors(model = "glove-100", dest_dir = NULL, overwrite = FALSE)

Arguments

model

Model to download: "glove-50", "glove-100", "glove-200", "glove-300", "glove-twitter-25", "glove-twitter-50", "glove-twitter-100", "glove-twitter-200"
dest_dir

Destination directory (default: user cache)
overwrite

Overwrite existing files

Value

Path to the downloaded vectors file

Examples

## Not run: 
# Download 100-dimensional GloVe vectors (~130MB)
path <- download_word_vectors("glove-100")

# Use with Vectrix
db <- Vectrix$new("docs", model = "glove", model_path = path)

## End(Not run)

VectrixDB Embedders (Pure R Implementation)

Description

Embedding models for text vectorization using R-native packages

English Stopwords

Description

Common English stopwords

Usage

ENGLISH_STOPWORDS

Format

An object of class character of length 45.

Enhanced Search Results

Description

Search results with enterprise features

Public fields

results

List of result items

facets

Named list of FacetResult objects

total_count

Total results before filtering

filtered_count

Results after ACL filtering

query_time_ms

Query time in milliseconds

rerank_time_ms

Rerank time in milliseconds

facet_time_ms

Facet time in milliseconds

Methods

Public methods

Method new()

Create new EnhancedSearchResults

Usage
EnhancedSearchResults$new(
  results,
  facets = list(),
  total_count = 0,
  filtered_count = 0,
  query_time_ms = 0,
  rerank_time_ms = 0,
  facet_time_ms = 0
)
Arguments
results

List of results

facets

Named list of FacetResult

total_count

Total count

filtered_count

Filtered count

query_time_ms

Query time

rerank_time_ms

Rerank time

facet_time_ms

Facet time

Method to_list()

Convert to list

Usage
EnhancedSearchResults$to_list()
Returns

List representation

Method clone()

The objects of this class are cloneable with this method.

Usage
EnhancedSearchResults$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Entity

Description

An extracted entity

Public fields

id

Unique identifier

name

Entity name

type

Entity type

description

Description

source_chunks

Source chunk IDs

embedding

Vector embedding

metadata

Additional metadata

Methods

Public methods

Method new()

Create a new Entity

Usage
Entity$new(
  id = NULL,
  name,
  type,
  description = NULL,
  source_chunks = NULL,
  embedding = NULL,
  metadata = NULL
)
Arguments
id

Unique ID

name

Name

type

Type

description

Description

source_chunks

Sources

embedding

Vector

metadata

Metadata

Method to_list()

Convert to list

Usage
Entity$to_list()

Method clone()

The objects of this class are cloneable with this method.

Usage
Entity$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Extraction Result

Description

Result of entity extraction

Public fields

entities

List of Entity objects

relationships

List of Relationship objects

source_chunk

Source chunk ID

Methods

Public methods

Method new()

Create new ExtractionResult

Usage
ExtractionResult$new(
  entities = list(),
  relationships = list(),
  source_chunk = NULL
)
Arguments
entities

Entities

relationships

Relationships

source_chunk

Source

Method clone()

The objects of this class are cloneable with this method.

Usage
ExtractionResult$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Extractor Types

Description

Extractor Types

Usage

ExtractorType

Format

An object of class list of length 4.

Facet Aggregator

Description

Faceted search aggregator for computing aggregations/counts

Methods

Public methods

Method aggregate()

Aggregate facet values from documents

Usage
FacetAggregator$aggregate(documents, facet_configs)
Arguments
documents

List of documents with metadata

facet_configs

List of field names or FacetConfig objects

Returns

Named list mapping field names to FacetResult

Method to_list()

Convert facet results to list format

Usage
FacetAggregator$to_list(facet_results)
Arguments
facet_results

Named list of FacetResult objects

Returns

List format suitable for JSON

Method clone()

The objects of this class are cloneable with this method.

Usage
FacetAggregator$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Examples

## Not run: 
aggregator <- FacetAggregator$new()
facets <- aggregator$aggregate(
  documents = list(
    list(category = "tech", author = "Alice"),
    list(category = "science", author = "Bob")
  ),
  facet_fields = c("category", "author")
)

## End(Not run)

Facet Configuration

Description

Configuration for a facet field

Public fields

field

Field name to facet on

limit

Max values to return

min_count

Minimum count to include

sort_by

Sort by "count" or "value"

include_zero

Include zero-count values

Methods

Public methods

Method new()

Create a new FacetConfig

Usage
FacetConfig$new(
  field,
  limit = 10,
  min_count = 1,
  sort_by = "count",
  include_zero = FALSE
)
Arguments
field

Field name

limit

Max values (default: 10)

min_count

Min count (default: 1)

sort_by

Sort method (default: "count")

include_zero

Include zeros (default: FALSE)

Method clone()

The objects of this class are cloneable with this method.

Usage
FacetConfig$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Facet Result

Description

Result of facet aggregation

Public fields

field

Field name

values

List of FacetValue objects

total_count

Total count

other_count

Count of values not in top-N

Methods

Public methods

Method new()

Create a new FacetResult

Usage
FacetResult$new(field, values, total_count, other_count = 0)
Arguments
field

Field name

values

List of FacetValue objects

total_count

Total count

other_count

Other count

Method to_list()

Convert to list

Usage
FacetResult$to_list()

Method clone()

The objects of this class are cloneable with this method.

Usage
FacetResult$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Facet Value

Description

A single facet value with count

Public fields

value

The facet value

count

Number of occurrences

Methods

Public methods

Method new()

Create a new FacetValue

Usage
FacetValue$new(value, count)
Arguments
value

The value

count

The count

Method clone()

The objects of this class are cloneable with this method.

Usage
FacetValue$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

File Cache

Description

File-based persistent cache using RDS files

Super class

VectrixDB::BaseCache -> FileCache

Methods

Public methods

Inherited methods

Method new()

Create a new FileCache

Usage
FileCache$new(config = NULL)
Arguments
config

CacheConfig object

Method get()

Get value from cache

Usage
FileCache$get(key)
Arguments
key

Cache key

Returns

Value or NULL

Method set()

Set value in cache

Usage
FileCache$set(key, value, ttl = NULL)
Arguments
key

Cache key

value

Value to cache

ttl

Time to live

Method delete()

Delete key from cache

Usage
FileCache$delete(key)
Arguments
key

Cache key

Returns

Logical success

Method exists()

Check if key exists

Usage
FileCache$exists(key)
Arguments
key

Cache key

Returns

Logical

Method clear()

Clear cache

Usage
FileCache$clear()

Method size()

Get cache size

Usage
FileCache$size()
Returns

Integer

Method cleanup_expired()

Cleanup expired entries

Usage
FileCache$cleanup_expired()
Returns

Integer count removed

Method clone()

The objects of this class are cloneable with this method.

Usage
FileCache$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Filter Class for Metadata Filtering

Description

Build metadata filters for search queries

Public fields

conditions

List of filter conditions

Methods

Public methods

Method new()

Create a new Filter

Usage
Filter$new(...)
Arguments
...

Named filter conditions

Method eq()

Add equality condition

Usage
Filter$eq(field, value)
Arguments
field

Field name

value

Value to match

Returns

Self for chaining

Method ne()

Add not-equal condition

Usage
Filter$ne(field, value)
Arguments
field

Field name

value

Value to exclude

Returns

Self for chaining

Method gt()

Add greater-than condition

Usage
Filter$gt(field, value)
Arguments
field

Field name

value

Threshold value

Returns

Self for chaining

Method lt()

Add less-than condition

Usage
Filter$lt(field, value)
Arguments
field

Field name

value

Threshold value

Returns

Self for chaining

Method in_list()

Add in-list condition

Usage
Filter$in_list(field, values)
Arguments
field

Field name

values

Vector of values

Returns

Self for chaining

Method to_list()

Convert to list for API

Usage
Filter$to_list()
Returns

List representation

Method clone()

The objects of this class are cloneable with this method.

Usage
Filter$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Global Searcher

Description

Community-based graph search

Public fields

communities

List of communities

k

Number of communities

Methods

Public methods

Method new()

Create a new GlobalSearcher

Usage
GlobalSearcher$new(communities, k = 5)
Arguments
communities

List of Community objects

k

Number of communities

Method search()

Search communities

Usage
GlobalSearcher$search(query)
Arguments
query

Query string

Returns

GlobalSearchResult

Method clone()

The objects of this class are cloneable with this method.

Usage
GlobalSearcher$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Global Search Result

Description

Result from global community search

Public fields

communities

Matching communities

summaries

Community summaries

context

Combined context

score

Relevance score

Methods

Public methods

Method new()

Create new GlobalSearchResult

Usage
GlobalSearchResult$new(
  communities = list(),
  summaries = character(0),
  context = NULL,
  score = 0
)
Arguments
communities

Communities

summaries

Summaries

context

Context

score

Score

Method clone()

The objects of this class are cloneable with this method.

Usage
GlobalSearchResult$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

VectrixDB GraphRAG Module

Description

Native GraphRAG implementation for VectrixDB

Features:

  • Entity and relationship extraction

  • Hierarchical community detection

  • Local, global, and hybrid search strategies

  • Incremental graph updates

GraphRAG Configuration

Description

Configuration for VectrixDB's GraphRAG implementation

Public fields

enabled

Whether GraphRAG is enabled

chunk_size

Target tokens per chunk

chunk_overlap

Overlapping tokens

chunk_by_sentence

Preserve sentence boundaries

extractor

Extraction method

nlp_model

NLP model name

llm_provider

LLM provider

llm_model

Model name

llm_api_key

API key

llm_endpoint

Custom endpoint

llm_temperature

Temperature

llm_max_tokens

Max tokens

max_community_levels

Max hierarchy depth

min_community_size

Min entities per community

relationship_threshold

Min relationship strength

deduplicate_entities

Merge similar entities

entity_similarity_threshold

Similarity for dedup

search_type

Default search strategy

local_search_k

Seed entities for local search

global_search_k

Communities for global search

traversal_depth

Max hops

include_relationships

Include relationship context

include_community_context

Include community summaries

enable_incremental

Incremental updates

batch_size

Chunks per batch

use_cache

Cache embeddings

cache_ttl

Cache TTL seconds

entity_types

Types to extract

relationship_types

Types to extract

Methods

Public methods

Method new()

Create a new GraphRAGConfig

Usage
GraphRAGConfig$new(enabled = FALSE, ...)
Arguments
enabled

Enable GraphRAG

...

Additional configuration options

Method with_openai()

Configure for OpenAI

Usage
GraphRAGConfig$with_openai(model = "gpt-4o-mini", api_key = NULL)
Arguments
model

Model name

api_key

API key

Returns

Self

Method with_ollama()

Configure for Ollama

Usage
GraphRAGConfig$with_ollama(
  model = "llama3.2",
  endpoint = "http://localhost:11434"
)
Arguments
model

Model name

endpoint

Endpoint URL

Returns

Self

Method clone()

The objects of this class are cloneable with this method.

Usage
GraphRAGConfig$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Examples

## Not run: 
config <- GraphRAGConfig$new(enabled = TRUE)
db <- Vectrix$new("knowledge_base", graphrag_config = config)

## End(Not run)

GraphRAG Pipeline

Description

Complete GraphRAG processing pipeline

Public fields

config

GraphRAGConfig

graph

KnowledgeGraph

communities

Detected communities

Methods

Public methods

Method new()

Create a new GraphRAGPipeline

Usage
GraphRAGPipeline$new(config = NULL)
Arguments
config

GraphRAGConfig

Method process()

Process documents

Usage
GraphRAGPipeline$process(texts, document_ids = NULL)
Arguments
texts

Character vector of documents

document_ids

Document IDs

Returns

Self

Method search()

Search the graph

Usage
GraphRAGPipeline$search(query, search_type = NULL)
Arguments
query

Query string

search_type

"local", "global", or "hybrid"

Returns

Search result

Method stats()

Get statistics

Usage
GraphRAGPipeline$stats()
Returns

Named list

Method clone()

The objects of this class are cloneable with this method.

Usage
GraphRAGPipeline$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Graph Search Types

Description

Graph Search Types

Usage

GraphSearchType

Format

An object of class list of length 3.

VectrixDB HNSW Index

Description

Hierarchical Navigable Small World graph for fast approximate nearest neighbor search

Uses RcppAnnoy for high-performance ANN search. Falls back to brute-force search if RcppAnnoy is not available.

HNSW Index

Description

High-performance approximate nearest neighbor index

Public fields

dimension

Vector dimension

metric

Distance metric

n_trees

Number of trees (for Annoy)

search_k

Search parameter

Methods

Public methods

Method new()

Create a new HNSWIndex

Usage
HNSWIndex$new(dimension, metric = "angular", n_trees = 50, search_k = -1)
Arguments
dimension

Vector dimension

metric

Distance metric: "angular", "euclidean", "manhattan", "dot"

n_trees

Number of trees for index (higher = more accuracy)

search_k

Search parameter (higher = more accuracy, -1 = auto)

Method add_items()

Add items to the index

Usage
HNSWIndex$add_items(ids, vectors)
Arguments
ids

Character vector of IDs

vectors

Matrix of vectors (rows = items)

Returns

Self

Method build()

Build the index (required before searching)

Usage
HNSWIndex$build()
Returns

Self

Method search()

Search for nearest neighbors

Usage
HNSWIndex$search(query, k = 10, include_distances = TRUE)
Arguments
query

Query vector

k

Number of neighbors

include_distances

Return distances

Returns

Data frame with id, distance columns

Method get_vector()

Get vector by ID

Usage
HNSWIndex$get_vector(id)
Arguments
id

Item ID

Returns

Vector or NULL

Method get_ids()

Get all IDs

Usage
HNSWIndex$get_ids()
Returns

Character vector

Method size()

Get item count

Usage
HNSWIndex$size()
Returns

Integer

Method remove_items()

Remove items from index

Usage
HNSWIndex$remove_items(ids)
Arguments
ids

IDs to remove

Returns

Self

Method clear()

Clear the index

Usage
HNSWIndex$clear()
Returns

Self

Method save()

Save index to file

Usage
HNSWIndex$save(path)
Arguments
path

File path

Method load()

Load index from file

Usage
HNSWIndex$load(path)
Arguments
path

File path

Returns

Self

Method clone()

The objects of this class are cloneable with this method.

Usage
HNSWIndex$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Examples

## Not run: 
# Create index
index <- HNSWIndex$new(dimension = 128, metric = "angular")

# Add vectors
index$add_items(ids = c("a", "b", "c"),
                vectors = matrix(rnorm(384), nrow = 3))

# Search
results <- index$search(query = rnorm(128), k = 5)

## End(Not run)

Keyword Analyzer

Description

Treats entire input as single token

Super class

VectrixDB::TextAnalyzer -> KeywordAnalyzer

Methods

Public methods

Inherited methods

Method analyze()

Analyze text as single keyword

Usage
KeywordAnalyzer$analyze(text)
Arguments
text

Input text

Returns

Single-element character vector

Method clone()

The objects of this class are cloneable with this method.

Usage
KeywordAnalyzer$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Knowledge Graph

Description

Graph storage for entities and relationships

Public fields

name

Graph name

Methods

Public methods

Method new()

Create a new KnowledgeGraph

Usage
KnowledgeGraph$new(name = "default")
Arguments
name

Graph name

Method add_entity()

Add an entity

Usage
KnowledgeGraph$add_entity(entity)
Arguments
entity

Entity object

Method add_relationship()

Add a relationship

Usage
KnowledgeGraph$add_relationship(relationship)
Arguments
relationship

Relationship object

Method get_entity()

Get entity by ID

Usage
KnowledgeGraph$get_entity(entity_id)
Arguments
entity_id

Entity ID

Returns

Entity or NULL

Method get_all_entities()

Get all entities

Usage
KnowledgeGraph$get_all_entities()
Returns

List of Entity objects

Method get_all_relationships()

Get all relationships

Usage
KnowledgeGraph$get_all_relationships()
Returns

List of Relationship objects

Method get_neighbors()

Get neighbors of an entity

Usage
KnowledgeGraph$get_neighbors(entity_id, direction = "both")
Arguments
entity_id

Entity ID

direction

"out", "in", or "both"

Returns

List of Entity objects

Method traverse()

Traverse graph from seed entities

Usage
KnowledgeGraph$traverse(seed_ids, max_depth = 2)
Arguments
seed_ids

Starting entity IDs

max_depth

Maximum depth

Returns

SubGraph object

Method entity_count()

Get entity count

Usage
KnowledgeGraph$entity_count()
Returns

Integer

Method relationship_count()

Get relationship count

Usage
KnowledgeGraph$relationship_count()
Returns

Integer

Method search_entities()

Search entities by name

Usage
KnowledgeGraph$search_entities(query, limit = 10)
Arguments
query

Query string

limit

Max results

Returns

List of Entity objects

Method clone()

The objects of this class are cloneable with this method.

Usage
KnowledgeGraph$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Late Interaction Embedder (Simplified ColBERT-style)

Description

Token-level embeddings for late interaction scoring

Public fields

dimension

Token embedding dimension

language

Language setting ("en" or "ml")

Methods

Public methods

Method new()

Create a new LateInteractionEmbedder

Usage
LateInteractionEmbedder$new(dimension = 64, language = "en")
Arguments
dimension

Embedding dimension per token

language

Language behavior ("en" = ASCII-focused, "ml" = Unicode-aware)

Method embed()

Embed texts to token-level embeddings

Usage
LateInteractionEmbedder$embed(texts)
Arguments
texts

Character vector of texts

Returns

List of matrices (each matrix is token embeddings for a document)

Method score()

Compute late interaction (MaxSim) score

Usage
LateInteractionEmbedder$score(query_embeddings, doc_embeddings)
Arguments
query_embeddings

Query token embeddings matrix

doc_embeddings

Document token embeddings matrix

Returns

Numeric score

Method clone()

The objects of this class are cloneable with this method.

Usage
LateInteractionEmbedder$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

LLM Provider Types

Description

LLM Provider Types

Usage

LLMProvider

Format

An object of class list of length 4.

Load HNSW Index

Description

Load saved index from file

Usage

load_hnsw_index(path)

Arguments

path

File path

Value

HNSWIndex object

Load word vectors into memory

Description

Loads pre-trained word vectors from a file

Usage

load_word_vectors(path, max_words = NULL, normalize = TRUE)

Arguments

path

Path to word vectors file (GloVe .txt or word2vec .bin)
max_words

Maximum number of words to load (NULL for all)
normalize

Normalize vectors to unit length

Value

WordVectors object

Local Searcher

Description

Entity-based graph search

Public fields

graph

Knowledge graph

k

Number of seed entities

traversal_depth

Max hops

Methods

Public methods

Method new()

Create a new LocalSearcher

Usage
LocalSearcher$new(graph, k = 10, traversal_depth = 2)
Arguments
graph

KnowledgeGraph

k

Seed entities

traversal_depth

Max depth

Method search()

Search the graph

Usage
LocalSearcher$search(query)
Arguments
query

Query string

Returns

LocalSearchResult

Method clone()

The objects of this class are cloneable with this method.

Usage
LocalSearcher$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Local Search Result

Description

Result from local graph search

Public fields

entities

Matching entities

relationships

Related relationships

subgraph

Traversed subgraph

context

Combined context text

score

Relevance score

Methods

Public methods

Method new()

Create new LocalSearchResult

Usage
LocalSearchResult$new(
  entities = list(),
  relationships = list(),
  subgraph = NULL,
  context = NULL,
  score = 0
)
Arguments
entities

Entities

relationships

Relationships

subgraph

SubGraph

context

Context

score

Score

Method clone()

The objects of this class are cloneable with this method.

Usage
LocalSearchResult$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Memory Cache

Description

In-memory LRU cache with TTL support

Ultra-low latency, limited by available RAM. Best for hot data, session data, frequently accessed vectors.

Super class

VectrixDB::BaseCache -> MemoryCache

Methods

Public methods

Inherited methods

Method new()

Create a new MemoryCache

Usage
MemoryCache$new(config = NULL)
Arguments
config

CacheConfig object

Method get()

Get value from cache

Usage
MemoryCache$get(key)
Arguments
key

Cache key

Returns

Value or NULL

Method set()

Set value in cache

Usage
MemoryCache$set(key, value, ttl = NULL)
Arguments
key

Cache key

value

Value to cache

ttl

Time to live

Method delete()

Delete key from cache

Usage
MemoryCache$delete(key)
Arguments
key

Cache key

Returns

Logical success

Method exists()

Check if key exists

Usage
MemoryCache$exists(key)
Arguments
key

Cache key

Returns

Logical

Method clear()

Clear cache

Usage
MemoryCache$clear()

Method size()

Get cache size

Usage
MemoryCache$size()
Returns

Integer

Method cleanup_expired()

Cleanup expired entries

Usage
MemoryCache$cleanup_expired()
Returns

Integer count removed

Method clone()

The objects of this class are cloneable with this method.

Usage
MemoryCache$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Maximal Marginal Relevance (MMR) Reranker

Description

Reranks for diversity using MMR algorithm

Public fields

lambda

Balance between relevance and diversity (0-1)

Methods

Public methods

Method new()

Create a new MMRReranker

Usage
MMRReranker$new(lambda = 0.7)
Arguments
lambda

Relevance vs diversity tradeoff (higher = more relevance)

Method rerank()

Rerank for diversity

Usage
MMRReranker$rerank(query_vector, doc_vectors, doc_ids, scores, limit = 10)
Arguments
query_vector

Query embedding

doc_vectors

Matrix of document embeddings

doc_ids

Vector of document IDs

scores

Original relevance scores

limit

Number of results

Returns

Data frame with reranked results

Method clone()

The objects of this class are cloneable with this method.

Usage
MMRReranker$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

No-Op Cache

Description

Disabled cache (no caching)

Super class

VectrixDB::BaseCache -> NoCache

Methods

Public methods

Inherited methods

Method get()

Get value from cache (always returns NULL)

Usage
NoCache$get(key)
Arguments
key

Cache key

Returns

NULL

Method set()

Set cache value (no-op)

Usage
NoCache$set(key, value, ttl = NULL)
Arguments
key

Cache key

value

Value to cache

ttl

Time-to-live in seconds (ignored)

Returns

Invisibly returns NULL

Method delete()

Delete key from cache (always FALSE)

Usage
NoCache$delete(key)
Arguments
key

Cache key

Returns

FALSE

Method exists()

Check if key exists (always FALSE)

Usage
NoCache$exists(key)
Arguments
key

Cache key

Returns

FALSE

Method clear()

Clear cache (no-op)

Usage
NoCache$clear()
Returns

Invisibly returns NULL

Method size()

Get cache size (always 0)

Usage
NoCache$size()
Returns

Integer zero

Method clone()

The objects of this class are cloneable with this method.

Usage
NoCache$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Parse ACL String

Description

Parse ACL string like 'user:alice' or 'group:engineering'

Usage

parse_acl(acl_string)

Arguments

acl_string

ACL string

Value

ACLPrincipal object

Quick search - Index texts and search immediately

Description

Quick search - Index texts and search immediately

Usage

quick_search(texts, query, limit = 5)

Arguments

texts

Character vector of texts to index
query

Search query
limit

Number of results

Value

Results object

Examples

## Not run: 
results <- quick_search(
  texts = c("Python is great", "Java is verbose", "Rust is fast"),
  query = "programming language"
)
print(results$top()$text)

## End(Not run)

Regex Entity Extractor

Description

Simple regex-based entity extractor (no external dependencies)

Public fields

entity_types

Entity types to extract

Methods

Public methods

Method new()

Create a new RegexExtractor

Usage
RegexExtractor$new(entity_types = NULL)
Arguments
entity_types

Types to extract

Method extract()

Extract entities from text

Usage
RegexExtractor$extract(text, chunk_id = NULL)
Arguments
text

Text to extract from

chunk_id

Chunk ID

Returns

ExtractionResult

Method clone()

The objects of this class are cloneable with this method.

Usage
RegexExtractor$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Relationship

Description

A relationship between entities

Public fields

id

Unique identifier

source_id

Source entity ID

target_id

Target entity ID

type

Relationship type

description

Description

weight

Relationship weight

source_chunks

Source chunk IDs

metadata

Additional metadata

Methods

Public methods

Method new()

Create a new Relationship

Usage
Relationship$new(
  source_id,
  target_id,
  type,
  description = NULL,
  weight = 1,
  source_chunks = NULL,
  metadata = NULL
)
Arguments
source_id

Source entity

target_id

Target entity

type

Relationship type

description

Description

weight

Weight

source_chunks

Sources

metadata

Metadata

Method to_list()

Convert to list

Usage
Relationship$to_list()

Method clone()

The objects of this class are cloneable with this method.

Usage
Relationship$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Advanced Reranking Module

Description

Learned weight reranking with BM25 + semantic fusion

Reranker (Cross-Encoder Style Scoring)

Description

Reranks results using term overlap and semantic similarity

Public fields

language

Language setting ("en" or "ml")

Methods

Public methods

Method new()

Create a new RerankerEmbedder

Usage
RerankerEmbedder$new(language = "en")
Arguments
language

Language behavior ("en" = English stopwords, "ml" = Unicode tokens)

Method score()

Score query-document pairs

Usage
RerankerEmbedder$score(query, documents)
Arguments
query

Query text

documents

Character vector of document texts

Returns

Numeric vector of scores (0-1)

Method clone()

The objects of this class are cloneable with this method.

Usage
RerankerEmbedder$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Single Search Result

Description

Represents a single search result with id, text, score, and metadata

Public fields

id

Document ID

text

Document text

score

Relevance score

metadata

Document metadata

Methods

Public methods

Method new()

Create a new Result object

Usage
Result$new(id, text, score, metadata = list())
Arguments
id

Document ID

text

Document text

score

Relevance score

metadata

Optional metadata list

Method print()

Print result summary

Usage
Result$print()

Method clone()

The objects of this class are cloneable with this method.

Usage
Result$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Search Results Collection

Description

Collection of search results with convenient accessors

Public fields

items

List of Result objects

query

Search query

mode

Search mode

time_ms

Execution time in ms

Methods

Public methods

Method new()

Create a new Results object

Usage
Results$new(items = list(), query = "", mode = "hybrid", time_ms = 0)
Arguments
items

List of Result objects

query

Search query string

mode

Search mode used

time_ms

Execution time in milliseconds

Method length()

Get number of results

Usage
Results$length()

Method texts()

Get all result texts

Usage
Results$texts()
Returns

Character vector of texts

Method ids()

Get all result IDs

Usage
Results$ids()
Returns

Character vector of IDs

Method scores()

Get all scores

Usage
Results$scores()
Returns

Numeric vector of scores

Method top()

Get top result

Usage
Results$top()
Returns

Result object or NULL if empty

Method get()

Get result by index

Usage
Results$get(i)
Arguments
i

Index

Returns

Result object

Method foreach()

Iterate over results

Usage
Results$foreach(fn)
Arguments
fn

Function to apply to each result

Method print()

Print results summary

Usage
Results$print()

Method clone()

The objects of this class are cloneable with this method.

Usage
Results$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Search Mode Enumeration

Description

Available search modes for VectrixDB

Usage

SearchMode

Format

An object of class list of length 5.

Sentence Embedder using Word Vectors

Description

Creates sentence embeddings by averaging word vectors with IDF weighting

Public fields

dim

Embedding dimension

vocab_size

Vocabulary size

Methods

Public methods

Method new()

Create a new SentenceEmbedder

Usage
SentenceEmbedder$new(word_vectors, use_idf = TRUE, smooth_idf = 1)
Arguments
word_vectors

WordVectors object from load_word_vectors()

use_idf

Use IDF weighting (recommended)

smooth_idf

Smoothing for IDF

Method fit()

Fit IDF weights on a corpus

Usage
SentenceEmbedder$fit(texts)
Arguments
texts

Character vector of texts

Method embed()

Embed texts to sentence vectors

Usage
SentenceEmbedder$embed(texts)
Arguments
texts

Character vector of texts

Returns

Matrix of embeddings (rows are sentences)

Method get_word_vector()

Get word vector for a single word

Usage
SentenceEmbedder$get_word_vector(word)
Arguments
word

Word to look up

Returns

Numeric vector or NULL if not found

Method has_word()

Check if word is in vocabulary

Usage
SentenceEmbedder$has_word(word)
Arguments
word

Word to check

Returns

Logical

Method most_similar()

Find most similar words

Usage
SentenceEmbedder$most_similar(word, n = 10)
Arguments
word

Query word

n

Number of results

Returns

Data frame with word and similarity

Method clone()

The objects of this class are cloneable with this method.

Usage
SentenceEmbedder$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

VectrixDB Server Functions

Description

REST API and dashboard server for VectrixDB

Set CLI Config

Description

Set CLI Config

Usage

set_cli_config(config)

Arguments

config

CLIConfig object

Simple Stemmer

Description

Simple suffix-stripping stemmer (no external dependencies)

Public fields

suffixes

List of suffixes to remove

Methods

Public methods

Method stem()

Stem a word

Usage
SimpleStemmer$stem(word)
Arguments
word

Word to stem

Returns

Stemmed word

Method stem_words()

Stem multiple words

Usage
SimpleStemmer$stem_words(words)
Arguments
words

Character vector

Returns

Stemmed words

Method clone()

The objects of this class are cloneable with this method.

Usage
SimpleStemmer$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Sparse Embedder (BM25/TF-IDF)

Description

Generates sparse BM25 embeddings for keyword search

Public fields

vocab

Vocabulary

language

Language setting ("en" or "ml")

Methods

Public methods

Method new()

Create a new SparseEmbedder

Usage
SparseEmbedder$new(language = "en")
Arguments
language

Language behavior ("en" = ASCII-focused, "ml" = Unicode-aware)

Method fit()

Fit the embedder on a corpus

Usage
SparseEmbedder$fit(texts)
Arguments
texts

Character vector of texts

Method embed()

Embed texts to sparse vectors

Usage
SparseEmbedder$embed(texts)
Arguments
texts

Character vector of texts

Returns

Sparse matrix of BM25 scores

Method query_terms()

Get term scores for a query

Usage
SparseEmbedder$query_terms(query)
Arguments
query

Query text

Returns

Named vector of term scores

Method clone()

The objects of this class are cloneable with this method.

Usage
SparseEmbedder$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

VectrixDB Storage Classes

Description

Storage backends for VectrixDB

SubGraph

Description

A subset of a knowledge graph

Public fields

entities

Entities in subgraph

relationships

Relationships in subgraph

Methods

Public methods

Method new()

Create a new SubGraph

Usage
SubGraph$new(entities = list(), relationships = list())
Arguments
entities

Entities

relationships

Relationships

Method to_list()

Convert to list

Usage
SubGraph$to_list()

Method clone()

The objects of this class are cloneable with this method.

Usage
SubGraph$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Create English Text Analyzer

Description

English analyzer with stemming and stopwords

Usage

text_analyzer_english()

Value

TextAnalyzer object

Create Keyword Text Analyzer

Description

No tokenization - treat input as single token

Usage

text_analyzer_keyword()

Value

TextAnalyzer object

Create Simple Text Analyzer

Description

Lowercase + letter-only tokenization

Usage

text_analyzer_simple()

Value

TextAnalyzer object

Create Standard Text Analyzer

Description

Lowercase + basic tokenization

Usage

text_analyzer_standard()

Value

TextAnalyzer object

Text Analyzer

Description

Text analyzer for search indexing

Provides text processing pipelines:

  • Tokenization

  • Lowercasing

  • Stopword removal

  • Stemming

  • Synonym expansion

Public fields

lowercase

Convert to lowercase

remove_stopwords

Remove stopwords

stopwords

Set of stopwords

stemmer

Stemmer object

synonyms

Synonym dictionary

min_token_length

Minimum token length

max_token_length

Maximum token length

token_pattern

Regex pattern for tokens

Methods

Public methods

Method new()

Create a new TextAnalyzer

Usage
TextAnalyzer$new(
  lowercase = TRUE,
  remove_stopwords = FALSE,
  stopwords = NULL,
  use_stemmer = FALSE,
  synonyms = NULL,
  min_token_length = 1,
  max_token_length = 100,
  token_pattern = "[a-zA-Z0-9]+"
)
Arguments
lowercase

Lowercase text (default: TRUE)

remove_stopwords

Remove stopwords (default: FALSE)

stopwords

Custom stopwords (default: ENGLISH_STOPWORDS)

use_stemmer

Use stemming (default: FALSE)

synonyms

Named list of synonyms

min_token_length

Min length (default: 1)

max_token_length

Max length (default: 100)

token_pattern

Regex pattern

Method analyze()

Analyze text and return tokens

Usage
TextAnalyzer$analyze(text)
Arguments
text

Input text

Returns

Character vector of tokens

Method analyze_query()

Analyze a query string

Usage
TextAnalyzer$analyze_query(query)
Arguments
query

Query text

Returns

Character vector of tokens

Method clone()

The objects of this class are cloneable with this method.

Usage
TextAnalyzer$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Examples

## Not run: 
analyzer <- TextAnalyzer$english()
tokens <- analyzer$analyze("The quick brown foxes are jumping")
# c("quick", "brown", "fox", "jump")

## End(Not run)

Text Unit

Description

A chunk of text from a document

Public fields

id

Unique identifier

text

Text content

document_id

Source document

chunk_index

Index in document

start_char

Start position

end_char

End position

metadata

Additional metadata

Methods

Public methods

Method new()

Create a new TextUnit

Usage
TextUnit$new(
  id,
  text,
  document_id = NULL,
  chunk_index = 0,
  start_char = 0,
  end_char = 0,
  metadata = NULL
)
Arguments
id

Unique ID

text

Content

document_id

Source doc

chunk_index

Index

start_char

Start

end_char

End

metadata

Metadata

Method clone()

The objects of this class are cloneable with this method.

Usage
TextUnit$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Add Documents

Description

Add documents to a collection

Usage

vdb_add(db, texts, metadata = NULL, ids = NULL)

Arguments

db

Vectrix object or collection name
texts

Character vector of texts
metadata

Optional metadata
ids

Optional IDs

Value

Vectrix object

Examples

## Not run: 
vdb_add(db, c("Document 1", "Document 2"))
vdb_add("my_docs", c("Another doc"))

## End(Not run)

Batch Add from Directory

Description

Add all text files from a directory

Usage

vdb_add_dir(db, dir_path, pattern = "\\.txt$", recursive = TRUE)

Arguments

db

Vectrix object or collection name
dir_path

Directory path
pattern

File pattern (default: "*.txt")
recursive

Search subdirectories

Value

Vectrix object

Examples

## Not run: 
vdb_add_dir(db, "./documents/")

## End(Not run)

Create Collection

Description

Create a new VectrixDB collection

Usage

vdb_create(name, model = "tfidf", dimension = NULL, data_dir = NULL)

Arguments

name

Collection name
model

Embedding model
dimension

Vector dimension
data_dir

Data directory

Value

Vectrix object

Examples

## Not run: 
db <- vdb_create("my_docs")

## End(Not run)

Launch VectrixDB Dashboard

Description

Start the VectrixDB API server and mount the HTML dashboard at ⁠/dashboard⁠.

Usage

vdb_dashboard(
  db = NULL,
  data_path = NULL,
  port = 7377,
  host = "127.0.0.1",
  launch.browser = TRUE,
  api_key = NULL
)

Arguments

db

Optional Vectrix object.
data_path

Path to vector database directory.
port

Port number (default: 7377).
host

Host address (default: "127.0.0.1").
launch.browser

Whether to open browser on start.
api_key

Optional API key for authenticated write operations.

Value

Invisibly returns server object from vectrix_serve().

Launch Simple Dashboard

Description

Convenience wrapper around vdb_dashboard() using db$path.

Usage

vdb_dashboard_simple(db)

Arguments

db

Vectrix object.

Value

Invisibly returns server object from vdb_dashboard().

Delete Collection

Description

Delete a collection

Usage

vdb_delete(name, data_dir = NULL, confirm = TRUE)

Arguments

name

Collection name
data_dir

Data directory
confirm

Require confirmation

Value

Logical success

Examples

## Not run: 
vdb_delete("my_docs")

## End(Not run)

Delete Documents

Description

Delete documents by ID

Usage

vdb_delete_docs(db, ids)

Arguments

db

Vectrix object or collection name
ids

Document ID(s)

Value

Vectrix object

Export Collection

Description

Export collection to JSON file

Usage

vdb_export(db, path)

Arguments

db

Vectrix object or collection name
path

Output file path

Value

Logical success

Examples

## Not run: 
vdb_export(db, "backup.json")

## End(Not run)

Get Document

Description

Get document by ID

Usage

vdb_get(db, ids)

Arguments

db

Vectrix object or collection name
ids

Document ID(s)

Value

List of Result objects

Import from File

Description

Import documents from text file

Usage

vdb_import(db, path, separator = "\n")

Arguments

db

Vectrix object or collection name
path

Input file path
separator

Line separator for documents

Value

Vectrix object

Examples

## Not run: 
vdb_import(db, "documents.txt")

## End(Not run)

Collection Info

Description

Get collection information

Usage

vdb_info(db)

Arguments

db

Vectrix object or collection name

Value

Named list of info

Examples

## Not run: 
vdb_info(db)
vdb_info("my_docs")

## End(Not run)

Start Interactive CLI

Description

Start an interactive VectrixDB session

Usage

vdb_interactive(collection = NULL)

Arguments

collection

Default collection name

Examples

## Not run: 
vdb_interactive()

## End(Not run)

List Collections

Description

List all VectrixDB collections in the data directory

Usage

vdb_list(data_dir = NULL)

Arguments

data_dir

Data directory path

Value

Character vector of collection names

Examples

## Not run: 
vdb_list()

## End(Not run)

Open Collection

Description

Open an existing collection

Usage

vdb_open(name, data_dir = NULL)

Arguments

name

Collection name
data_dir

Data directory

Value

Vectrix object

Examples

## Not run: 
db <- vdb_open("my_docs")

## End(Not run)

Search Collection

Description

Search a collection

Usage

vdb_search(db, query, limit = 10, mode = "hybrid", show = TRUE)

Arguments

db

Vectrix object or collection name
query

Search query
limit

Number of results
mode

Search mode: "dense", "sparse", "hybrid", "ultimate"
show

Print results

Value

Results object

Examples

## Not run: 
results <- vdb_search(db, "machine learning")
results <- vdb_search("my_docs", "AI", limit = 5)

## End(Not run)

Collection Statistics

Description

Get detailed statistics

Usage

vdb_stats(db)

Arguments

db

Vectrix object or collection name

Value

Named list of stats

Vector Cache

Description

Specialized cache for vector search results

Features:

  • Query result caching

  • Vector embedding caching

  • Automatic cache invalidation

Public fields

prefix

Cache key prefix

Methods

Public methods

Method new()

Create a new VectorCache

Usage
VectorCache$new(cache, prefix = "vec:")
Arguments
cache

Base cache backend

prefix

Key prefix (default: "vec:")

Method get_search_results()

Get cached search results

Usage
VectorCache$get_search_results(collection, query, filter = NULL, limit = 10)
Arguments
collection

Collection name

query

Query vector

filter

Filter conditions

limit

Result limit

Returns

Cached results or NULL

Method set_search_results()

Cache search results

Usage
VectorCache$set_search_results(
  collection,
  query,
  results,
  filter = NULL,
  limit = 10,
  ttl = 300
)
Arguments
collection

Collection name

query

Query vector

results

Search results

filter

Filter conditions

limit

Result limit

ttl

Time to live (default: 300)

Method get_vector()

Get cached vector

Usage
VectorCache$get_vector(collection, vector_id)
Arguments
collection

Collection name

vector_id

Vector ID

Returns

Cached vector data or NULL

Method set_vector()

Cache vector data

Usage
VectorCache$set_vector(collection, vector_id, data, ttl = 3600)
Arguments
collection

Collection name

vector_id

Vector ID

data

Vector data

ttl

Time to live (default: 3600)

Method invalidate_vector()

Invalidate cached vector

Usage
VectorCache$invalidate_vector(collection, vector_id)
Arguments
collection

Collection name

vector_id

Vector ID

Method stats()

Get cache statistics

Usage
VectorCache$stats()
Returns

CacheStats object

Method clone()

The objects of this class are cloneable with this method.

Usage
VectorCache$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

VectrixDB Easy API - The Simplest Vector Database

Description

Zero config. Text in, results out. One line for everything.

Public fields

name

Collection name

path

Storage path

dimension

Vector dimension

model_name

Model identifier

model_type

Model type

language

Language setting

tier

Storage tier

Methods

Public methods

Method new()

Create or open a VectrixDB collection

Usage
Vectrix$new(
  name = "default",
  path = NULL,
  model = NULL,
  dimension = NULL,
  embed_fn = NULL,
  model_path = NULL,
  language = NULL,
  tier = "dense",
  auto_download = TRUE
)
Arguments
name

Collection name

path

Storage path. Defaults to a session temp directory.

model

Embedding model: "tfidf" (default), "glove-50", "glove-100", "glove-200", "glove-300", or "word2vec"

dimension

Vector dimension (auto-detected for GloVe)

embed_fn

Custom embedding function: fn(texts) -> matrix

model_path

Path to pre-trained word vectors (GloVe .txt or word2vec .bin)

language

Language behavior: "en" (English-focused) or "ml" (multilingual/Unicode)

tier

Storage tier: "dense", "hybrid", "ultimate", or "graph"

auto_download

Automatically download GloVe vectors if needed (default: TRUE)

Examples
\dontrun{
# Default TF-IDF embeddings (no external files needed)
db <- Vectrix$new("docs")

# With GloVe 100d word vectors (auto-downloads ~130MB)
db <- Vectrix$new("docs", model = "glove-100")

# With pre-downloaded GloVe
db <- Vectrix$new("docs", model_path = "path/to/glove.6B.100d.txt")

# Custom embedding function
db <- Vectrix$new("docs", embed_fn = my_embed_function, dimension = 768)
}

Method add()

Add texts to the collection

Usage
Vectrix$add(texts, metadata = NULL, ids = NULL)
Arguments
texts

Single text or character vector of texts

metadata

Optional metadata list or list of lists

ids

Optional custom IDs

Returns

Self for chaining

Examples
\dontrun{
db$add(c("text 1", "text 2"))
db$add("another text", metadata = list(source = "web"))
}

Method set_language()

Update collection language behavior

Usage
Vectrix$set_language(language = "en")
Arguments
language

Language behavior: "en" or "ml"

Returns

Self for chaining

Method search()

Search the collection

Usage
Vectrix$search(
  query,
  limit = 10,
  mode = "hybrid",
  rerank = NULL,
  filter = NULL,
  diversity = 0.7
)
Arguments
query

Search query text

limit

Number of results (default: 10)

mode

Search mode: "dense", "sparse", "hybrid", "ultimate"

rerank

Reranking method: NULL, "mmr", "exact", "cross-encoder"

filter

Metadata filter

diversity

Diversity parameter for MMR (0-1)

Returns

Results object with search results

Examples
\dontrun{
results <- db$search("python programming")
results <- db$search("AI", mode = "ultimate", rerank = "mmr")
print(results$top()$text)
}

Method delete()

Delete documents by ID

Usage
Vectrix$delete(ids)
Arguments
ids

Document ID(s) to delete

Returns

Self for chaining

Method clear()

Clear all documents from collection

Usage
Vectrix$clear()
Returns

Self for chaining

Method count()

Get number of documents

Usage
Vectrix$count()
Returns

Integer count

Method get()

Get documents by ID

Usage
Vectrix$get(ids)
Arguments
ids

Document ID(s)

Returns

List of Result objects

Method similar()

Find similar documents to a given document

Usage
Vectrix$similar(id, limit = 10)
Arguments
id

Document ID

limit

Number of results

Returns

Results object

Method close()

Close the database connection

Usage
Vectrix$close()

Method print()

Print Vectrix summary

Usage
Vectrix$print()

Method clone()

The objects of this class are cloneable with this method.

Usage
Vectrix$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Examples

## Not run: 
# Create and add - ONE LINE
db <- Vectrix$new("my_docs")$add(c("Python is great", "Machine learning is fun"))

# Search - ONE LINE
results <- db$search("programming")

# Full power - STILL ONE LINE
results <- db$search("AI", mode = "ultimate")  # dense + sparse + rerank

## End(Not run)


## ------------------------------------------------
## Method `Vectrix$new`
## ------------------------------------------------

## Not run: 
# Default TF-IDF embeddings (no external files needed)
db <- Vectrix$new("docs")

# With GloVe 100d word vectors (auto-downloads ~130MB)
db <- Vectrix$new("docs", model = "glove-100")

# With pre-downloaded GloVe
db <- Vectrix$new("docs", model_path = "path/to/glove.6B.100d.txt")

# Custom embedding function
db <- Vectrix$new("docs", embed_fn = my_embed_function, dimension = 768)

## End(Not run)

## ------------------------------------------------
## Method `Vectrix$add`
## ------------------------------------------------

## Not run: 
db$add(c("text 1", "text 2"))
db$add("another text", metadata = list(source = "web"))

## End(Not run)

## ------------------------------------------------
## Method `Vectrix$search`
## ------------------------------------------------

## Not run: 
results <- db$search("python programming")
results <- db$search("AI", mode = "ultimate", rerank = "mmr")
print(results$top()$text)

## End(Not run)

Create a new Vectrix collection

Description

Create a new Vectrix collection

Usage

vectrix_create(name = "default", ...)

Arguments

name

Collection name
...

Additional arguments passed to Vectrix$new()

Value

Vectrix object

Display VectrixDB information

Description

Show database statistics and info

Usage

vectrix_info(path = NULL)

Arguments

path

Database path

Examples

## Not run: 
vectrix_info(file.path(tempdir(), "my_data"))

## End(Not run)

Open an existing Vectrix collection

Description

Open an existing Vectrix collection

Usage

vectrix_open(name = "default", path = NULL)

Arguments

name

Collection name
path

Storage path

Value

Vectrix object

Start VectrixDB server

Description

Launch a REST API server with optional dashboard

Usage

vectrix_serve(
  path = NULL,
  host = "127.0.0.1",
  port = 7377,
  api_key = NULL,
  dashboard = TRUE,
  launch.browser = FALSE
)

Arguments

path

Database path
host

Host address (default: "127.0.0.1")
port

Port number (default: 7377)
api_key

Optional API key for authentication
dashboard

Enable dashboard (default: TRUE)
launch.browser

Open dashboard/docs URL in browser (default: FALSE)

Value

Invisible NULL (server runs until stopped)

Examples

## Not run: 
vectrix_serve(path = file.path(tempdir(), "my_data"), port = 7377)

## End(Not run)

VectrixDB Database Class

Description

Main database interface managing collections

Usage

vectrixdb(path = NULL, storage_type = "memory")

Arguments

path

Storage path
storage_type

Storage type

Value

VectrixDB object

Public fields

path

Database storage path

Methods

Public methods

Method new()

Create or open a VectrixDB database

Usage
VectrixDB$new(path = NULL, storage_type = "memory")
Arguments
path

Storage path

storage_type

Storage type ("memory" or "sqlite")

Method create_collection()

Create a new collection

Usage
VectrixDB$create_collection(
  name,
  dimension,
  metric = "cosine",
  enable_text_index = TRUE
)
Arguments
name

Collection name

dimension

Vector dimension

metric

Distance metric

enable_text_index

Enable text indexing

Returns

Collection object

Method get_collection()

Get an existing collection

Usage
VectrixDB$get_collection(name)
Arguments
name

Collection name

Returns

Collection object

Method list_collections()

List all collections

Usage
VectrixDB$list_collections()
Returns

Character vector of collection names

Method delete_collection()

Delete a collection

Usage
VectrixDB$delete_collection(name)
Arguments
name

Collection name

Method has_collection()

Check if collection exists

Usage
VectrixDB$has_collection(name)
Arguments
name

Collection name

Returns

Logical

Method stats()

Get database statistics

Usage
VectrixDB$stats()
Returns

List with stats

Method close()

Close the database

Usage
VectrixDB$close()

Method print()

Print database summary

Usage
VectrixDB$print()

Method clone()

The objects of this class are cloneable with this method.

Usage
VectrixDB$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

Word Vector Management

Description

Download, load, and use pre-trained word vectors (GloVe, fastText)