2023
Papyrus:
Fixed minor invalid CEFR scores for English.
Alexandria:
Fixed minor classification issues.
Papyrus:
Fixed page reload issue.
Added an FAQ.
Improved outcome by training new Dutch language models.
Fixed language detection issue.
Fixed German word difficulty issue.
Fixed German keyword issue.
Alexandria:
Improved the keyword extractor.
Created new analysis models.
Added a GPT model option for Dutch language.
Added the latest model and classifications.
Knowble:
Security updates.
Fixed word support function issue.
Fixed word pronunciation issue.
Fixed Missing sample sentences and pronunciation for frequent words.
General:
Bug fixes.
Papyrus:
The Papyrus user interface automatically translates into Dutch and German.
Alexandria:
As a user, I can train a classification model to classify my content and see a list of all previous models trained. I also can select a trained model, to see detailed classification accuracy statistics.
Alexandria:
As a user, I can see the classification accuracy for a statement and I can see which keywords I should add to my statement to improve classification results.
Papyrus:
Papyrus now also supports Dutch and German for text readability analysis.
Alexandria:
The screen for inspecting classified content items has received a makeover to improve overall usability. In addition, keywords extracted from a content item can now be highlighted in the text.
The confidence threshold for filtering classified content items can now be configured in more granular steps en is set by default to the maximum range (0-1)
Alexandria:
We now can handle very large word lists.
Text keyword similarities between content items and taxonomy items have been visualized.
Taxonomy subjects are now used as primers for keyword extraction.
We now allow an url for text input.
Papyrus:
Papyrus now stores texts in the browser's local database.
Subscription error fixes.
2022
General:
Bug fixes.
Alexandria:
Content items are now shown by relevance.
Users are allowed to configure the keyword extractor to increase perceived relevance/accuracy results.
We added salience score as relevance metric.