seqsim 0.3

Library for computing measures of similarity for sequences of hashable data types

seqsim

Python library for computing measures of distance and similarity for sequences of hashable data types.

While developed as a general-purpose library, seqsim is mostly designed for usage in research within the field of cultural evolution, and particularly of the cultural evolution of textual traditions. Some methods act as a thin-wrapper to either the standard Python library or of to other libraries such as textdistance.

Installation

In any standard Python environment, seqsim can be installed with:

$ pip install seqsim

Usage

The library offers different methods to compare sequences of arbitrary hashable elements. It is possible to mix sequence and element types.

Full documentation is offered at ReadTheDocs and code with almost complete coverage is offered in the tests. For most common usages, a wrapper .distance() function can be used.

>>> import seqsim
>>> seqsim.edit.levenshtein_dist("kitten", "string")
5
>>> seqsim.edit.levenshtein_dist("kitten", "string", normal=True)
>>> 0.8333333333333334
>>> seqsim.sequence.ratcliff_obershelp([1,2,3,4], [2,4,3,5])
0.5
>>> seqsim.compression.entropy_ncd([1,2,3,4], [2,4,3,5])
0.08333333333333333

Demonstration

The core of the library are the metrics for sequence distance/similarity on arbitrary data types, as in the table below.

Method	"kitten" / "sitting"	(1, 2, 3, 4) / (3, 4, 2, 1)
arith_ncd	1.25	0.888889
arith_ncd_normal	1.25	0.888889
birnbaum_simil	10	5
birnbaum_simil_normal	0.3125	0.5
birnbaun	0.565217	0.5
birnbaun_normal	0.565217	0.5
bulk_delete	3	3
bulk_delete_normal	0.428571	0.75
damerau	3	4
damerau_normal	0.428571	1
entropy	0.101341	0
entropy_normal	0.101341	0
fast_birnbaum	0.666667	0.7
fast_birnbaum_normal	0.666667	0.7
fast_birnbaum_simil	7	3
fast_birnbaum_simil_normal	0.25	0.3
fragile_ends_simil	3	3.5
fragile_ends_simil_normal	0.5	1
jaccard	0.7	0
jaccard_normal	0.7	0
jaro	0.253968	0.5
jaro_normal	0.253968	0.5
jaro_winkler	0.253968	0.5
jaro_winkler_normal	0.253968	0.5
levenshtein	3	4
levenshtein_normal	0.428571	1
mmcwpa	0.538462	0.387628
mmcwpa_normal	0.538462	0.387628
ratcliff_obershelp	0.384615	0.5
ratcliff_obershelp_normal	0.384615	0.5
sorensen	0.384615	0
sorensen_normal	0.384615	0
stemmatological_simil	3	3
stemmatological_simil_normal	0.428571	0.75
subseq_jaccard	0.751556	0.547008
subseq_jaccard_normal	0.751556	0.547008

Changelog

Version 0.3:

• Improvements to code quality, documentation, and references
• Added new methods and scaffolding for future expansions

Version 0.2:

• First release for new roadmap supporting sequences of any hashable Python datatype, importing code from other projects (mostly from titivillus)

Community guidelines

While the authors can be contacted directly for support, it is recommended that third parties use GitHub standard features, such as issues and pull requests, to contribute, report problems, or seek support.

Contributing guidelines, including a code of conduct, can be found in the CONTRIBUTING.md file.

Authors and citation

The library is developed in the context of "Cultural Evolution of Text", project, with funding from the Riksbankens Jubileumsfond (grant agreement ID: MXM19-1087:1).

If you use seqsim, please cite it as:

Tresoldi, Tiago; Maurits, Luke; Dunn, Michael. (2021). seqsim, a library for computing measures of distance and similarity for sequences of hashable data types. Version 0.3. Uppsala: Uppsala universitet. Available at: https://github.com/evotext/seqsim

In BibTeX:

@misc{Tresoldi2021seqsim,
  author = {Tresoldi, Tiago; Maurits, Luke; Dunn, Michael},
  title = {seqsim, a library for computing measures of distance and similarity for sequences of hashable data types. Version 0.3},
  howpublished = {\url{https://github.com/evotext/seqsim}},
  address = {Uppsala},
  publisher = {Uppsala universitet},
  year = {2021},
}

References

The image at the top of this file is derived from Yves de Saint-Denis, Vie et martyre de saint Denis et de ses compagnons, versions latine et française. It is available in high resolution from Bibliothèque nationale de France, Département des Manuscrits, Français 2090, fol. 12v.

References to the various implementation are available in the source code comments and in the online documentation.