Package: alethiaR 0.1.0

Saket Choudhary

alethiaR: Match and Deduplicate Messy Entity Names Against a Reference List

Aligns messy entity strings against a reference list using either string similarity or language-model embeddings.

Authors:Saket Choudhary [aut, cre]

alethiaR_0.1.0.tar.gz
alethiaR_0.1.0.zip(r-4.7-any)alethiaR_0.1.0.zip(r-4.6-any)alethiaR_0.1.0.zip(r-4.5-any)
alethiaR_0.1.0.tgz(r-4.6-any)alethiaR_0.1.0.tgz(r-4.5-any)
alethiaR_0.1.0.tar.gz(r-4.7-any)alethiaR_0.1.0.tar.gz(r-4.6-any)
alethiaR_0.1.0.tgz(r-4.6-emscripten)
manual.pdf |manual.html✨
DESCRIPTION
card.svg |card.png
alethiaR/json (API)

# Install 'alethiaR' in R:

install.packages('alethiaR', repos = c('https://saketlab.r-universe.dev', 'https://cloud.r-project.org'))

Bug tracker:https://github.com/saketlab/alethia/issues

Pkgdown/docs site:https://alethia.saketlab.org

On CRAN:

entity-matching fuzzy-matching llms messy-data

3.65 score 3 stars 3 scripts 33 exports 1 dependencies

Last updated from:7d3f46a4b4 (on main). Checks:9 OK. Indexed: yes.

Target	Result	Time
linux-devel	OK	137
source / vignettes	OK	165
linux-release	OK	107
macos-release	OK	94
macos-oldrel	OK	90
windows-devel	OK	86
windows-release	OK	85
windows-oldrel	OK	90
wasm-release	OK	91

Exports:alethia alethia_split alignment_loss apply_threshold as_embedder assess_models centered_separability cluster_entities cluster_records DEFAULT_METRIC_WEIGHTS DEFAULT_THRESHOLD encode get_embedding_backend hubness intrinsic_dimensionality l2_normalize match_by_embeddings match_by_strings mutual_nn_edges mutual_nn_rate n_clusters onnx_available onnx_embedder positive_pair_rank RANK_DECIMALS reference_separability retrieval_margin set_embedding_backend token_sort token_sort_ratio true_accuracy uniformity_loss use_onnx_backend

Dependencies:stringdist

Quickstart

Install | Match against a reference list | Match on meaning | Choose a model without labels | Cluster without a reference list

Rendered fromquickstart.Rmdusingknitr::rmarkdown

Last update: 2026-08-26
Started: 2026-08-26

Citation

Development and contributors

Readme and manuals

Help Manual

Help page	Topics
Align messy entity strings against a reference list	alethia
Match composite entries by splitting them on a separator first	alethia_split
Alignment loss	alignment_loss
Blank out matches scoring below a threshold	apply_threshold
Coerce a model specification into an embedder	as_embedder
Score candidate embedding models on your own data, without labels	assess_models DEFAULT_METRIC_WEIGHTS
Label-free geometric metrics for an embedding set	assess-metrics
Anisotropy-corrected separability	centered_separability
Group entity strings into clusters of duplicates	cluster_entities
Cluster membership as a tidy data frame	cluster_records
Default minimum cosine similarity for embedding backends	DEFAULT_THRESHOLD
Encode text with an embedder	encode
Hubness	hubness
Intrinsic dimensionality	intrinsic_dimensionality
Row-wise L2 normalisation	l2_normalize
Match queries to references by embedding cosine similarity	match_by_embeddings
Match strings against a reference list by token-sort similarity	match_by_strings
Candidate merge edges from mutual nearest neighbours	mutual_nn_edges
Mutual nearest-neighbour rate	mutual_nn_rate
Number of clusters found	n_clusters
Is the local ONNX stack ready to use?	onnx_available
Build an embedder backed by a local ONNX model	onnx_embedder
Positive-pair rank	positive_pair_rank
Decimal places similarities are rounded to before any ranking decision	RANK_DECIMALS
Reference-set separability	reference_separability
Retrieval margin	retrieval_margin
Register a backend that turns model names into embedding functions	get_embedding_backend set_embedding_backend
Normalise a string by sorting its whitespace-separated tokens	token_sort
Token-sort similarity ratio	token_sort_ratio
Top-1 accuracy and MRR of an embedder against ground truth	true_accuracy
Uniformity loss	uniformity_loss
Resolve bare model names through the local ONNX runtime	use_onnx_backend