Sentibank

**Full Changelog**: https://github.com/socius-org/sentibank/compare/0.2.2.1...0.2.3

Extensive preprocessing was undertaken to transform diverse sentiment representation schemes into standardised formats, enabling rapid utilisation and seamless integration. The primary objective was to harmonise fuzzy or vector representations into well-defined unidimensional frameworks. Thus, the change was mainly about standardising the representations of preprocessed dictionaries, with a few other minor changes.

**Renamed Predefined Identifiers**:

- `NoVAD_v2013_bidimensional`, originally a bidimensional vector, was further processed as a vector norm. The predefined identifier was renamed to `NoVAD_v2013_norm` to reflect this transformation.
- `NoVAD_v2013_adjusted` was renamed to `NoVAD_v2013_boosted` for clarity and consistency.
- `HarvardGI_v2000` was renamed to `GeneralInquirer_v2000` for better discoverability and alignment with the dictionary's name.

**Standardising Binary Labels**:

- The binary labels ["Positive", "Negative"] in `GeneralInquirer_v2000` were converted to lowercase ["positive", "negative"], ensuring consistent casing across all preprocessed dictionaries with binary labels.

**Full Changelog**: https://github.com/socius-org/sentibank/compare/0.2.2...0.2.2.1

Argument `package_data` added to `setup.py` to exclude it from being treated as a package.

**Full Changelog**: https://github.com/socius-org/sentibank/compare/0.2.1...0.2.2

Improved the functionality of the `analyze().sentiment` module with the integration of the `spellcheck` class. This update *simply corrects spelling of words with three or more consecutive identical alphabets (e.g. "happppyyyy" to "happy")*. This optimization ensures more precise linguistic processing within the analysis pipeline, contributing to more reliable sentiment analysis outcomes.

**Full Changelog**: https://github.com/socius-org/sentibank/compare/0.2...0.2.1

The modified code ensures that when counting matches in a text, it considers only the longest n-gram for each occurrence. It uses a set (`matched_positions`) to keep track of the positions of the matched n-grams and checks for overlaps, ensuring that only the longest n-gram at each position contributes to the total score. This way, `utils.analyze().sentiment()` avoids counting overlapping n-grams and accurately calculates the total score based on the longest matching n-gram at each position in the text.

**Full Changelog**: https://github.com/socius-org/sentibank/compare/0.1.2...0.2

**Main Change**: Able to analyse sentiment of a given text, utilising a bag-of-words approach.