Jboynyc Textnets Versions Save

• Sets language_level compiler directive for Cython. (In response to #63).
• Updates to spaCy 3.7.4 and igraph 0.11.3.
• Removes references to conda-forge package as it has become difficult to keep updated.
• Adds dependency on pyarrow to satisfy future pandas releases.

• Updates to spaCy 3.7.2 and to igraph 0.10.5.
• Adds Python 3.12 compatibility. (In response to #61.)
• Fixes a bug in how tf-idf is calculated without sublinear scaling.
• If the corpus contains long documents, adjust spaCy's maximum document length parameter as needed.

:rocket:

• Updates to igraph 0.10.5.
• Updates to spaCy 3.6.0, bringing support for Slovenian.
• Add dependency on spacy-lookups-data for better support of languages without pre-trained models (e.g., Turkish).

• Improves documentation.
• Updates to scipy 1.10.
• Fixes how the compiled extension is built.
• Builds binary wheels for a greater number of platforms.

No substantive change from previous release.

• Improves documentation.
• Parallelizes execution of NLP and backbone extraction for large corpora and networks.
• Uses sparse matrices for some operations to avoid running out of memory.
• Makes other improvements to efficiency.
• Introduces a NodeType enum to differentiate document and term nodes.
• Updates to pandas 2.0 and spaCy 3.5.3.
• Now also tested for compatibility with PyPy 3.9!

• Fixes bug in disparity filter.
• Updates to spaCy 3.5.1, along various other dependencies.
• Improves testing

• Updates to spaCy 3.5 and igraph 0.10.4.
• When initializing Corpus with data that includes duplicated document labels, issue a warning, and concatenate documents with shared labels. (Implemented in response to #54).
• Various code quality improvements.

• Improves documentation.
• Progress bar for long-running operations (NLP and disparity filter). The progress bar is not shown during non-interactive use. To disable, set tn.params["progress_bar"] = False.
• Python 3.11 compatibility now confirmed on Windows, too.

• Python 3.11 compatibility! (YMMV if using Windows.)
• Adds an optional parameter remove_weak_edges when initializing a Textnet to create a sparser graph.
• Adds ProjectedTextnet.m property to access one-mode graphs' adjacency matrices.
• Adds Stoltz and Taylor's (2019) textual spanning measure.
• Adds TextnetBase.cluster_strength and TextnetBase.cluster_local_cc to calculate the weighted degree and local clustering coefficient of nodes within the subgraph formed by the cluster they belong to.
• Improves display of top nodes per cluster (top_cluster_nodes).
• Adds Korean language code.
• Fixes several bugs that occurred when setting document attributes.
• Updated and expanded documentation.