Sosia

~~~~~

2023-11-05

* Rename `.create_fields_sources_list()` to `.get_field_source_information()`.
* Use `tqdm` to print progress bars.
* Adopt improved format of field-source assignment and source information.
* Upgrade third-party code usage.
* Improve documentation, add copy-code button, update code examples.

~~~

2023-04-23

* Drop usage of configuration file and recommend project-specific databases.
* In class `Scientist(), rename properties: `.country` -> `.affiliation_country`, `.affiliation` -> `.affiliation_name`.
* In class `Scientist(), create property `.affiliation_type`.
* In `.create_fields_sources_list()`, make use of parameter "verbose".
* Make retrieval of affiliation related information robust to missing information (404 error).
* Pass on "refresh" parameter from `inform_matches()`.
* Use pyproject.toml for packaging, drop `pbr` (PEP 621).
* Use XDG compliant file storage for support files in `~/.cache/sosia/`.
* Improve various methods and functions for stability and speed.

~~~

2022-01-20

* In `.inform_matches()`, remove abstract similarity and reference list similarity computation and corresponding keywords "abstract_sim" and "reference_sim"; do not require `nltk` and `scikit-learn` anymore.
* In `.inform_matches()`, add "num_cited_refs" as number of jointly cited references up until provided year.
* Increase robustness to Scopus server problems.
* Allow to refresh downloaded results when using very large stacked source-based searches.
* Fix bug with integer conversion when using pandas > 1.1.5.
* Require pybliometrics >= 3.2.0.

~~~~~

2020-12-08

* Fix bug when creating a new config.ini.
* Require pybliometrics >= 2.7.2.

~~~

2020-12-04

* End support for Python 3.5.
* Reorganize config.ini.
* In `Original()`, add parameters "sql_name" and "first_year_search", and rename parameters: "year": "treatment_year", "year_margin": "first_year_margin", "search_affiliations": "affiliations".
* In `Original().find_matches()`, remove parameter "ignore_first_id".
* In `Original()`, change default values for parameters: "year_margin": 0.2, "pub_margin": 0.2, "cits_margin": 0.2, "coauth_margin": 0.2.
* Rename function `create_cache()` to `make_database()`.
* Rename tables in MySQL database: `author_size` becomes `author_pubs`, `author_cits_size` becomes `author_ncits`.
* In `Original().find_matches()` remove parameters "information", "stop_words", "tfidf_kwds"; always create a plain list.
* Add property `.matches` to `Original()`.
* Create new method `Original().inform_matches()` to add additional information to matches.
* Use externally provided list of sources and their fields.
* Remove unused property `Original().city`.
* Raise warning if there are too few publications to determine a field.
* Allow integer values for "refresh" in all instances, require pybliometrics >= 2.7.
* Require numpy.
* Fix bug originating from missing reference EIDs.
* Fix bug originating from missing source IDs.
* Improve documentaton, add tutorial.
* Add citation dunder.

~~~~~

2020-03-17

* Update docs w.r.t. the usage of pybliometrics.
* Add support for Python 3.8 and Python 3.9.
* Add missing required package lxml and require sklearn>=0.22.1.
* Correct verbose output of `.find_matches()` w.r.t. completeness of reference lists and abstracts of matches.
* Check for existence of the search group in `.find_machtes()`.
* In `.find_matches()`, fix bug when attempting to compute the cosine similarity when reference lists or abstracts are completely missing.
* In `.get_publication_language()`, fix bug resulting from bad downloads of abstracts.
* In `Original()`, fix bug resulting from unclean source ID information in Scopus search results.