Rosette API 1.5 リリース
Rosette API version1.5 のリリースをお知らせします。昨春、このクラウドAPIの提供を開始して以来の大幅な機能拡張となりました。
Rosette API 1.5 Released
January 11, 2017
Today we’re pleased to announce the launch of Rosette API version 1.5!
Updates include new targeted relationship extraction (replacing the previous “open” relationship extraction), changes to entity linking and extraction, improved text embeddings, and expanded support for Chinese, Japanese, Korean, and Vietnamese, including sentiment analysis for Japanese text (beta).
The /relationships endpoint now returns targeted relationships, specific relationships that connect two entities (think “Bill Gates was educated at Harvard). Check out the full list of extractable relationships and don’t worry, if you’re interested in “open” relationships, Rosette can still return that information as well, just set “discoveryMode” to “true”.
We’ve fully removed our (previously deprecated) entity linking endpoint. As a reminder, entity linking is automatically included with every /entities call!
Our long text entity linker has been replaced by our fast short string entity linker. You will now see entity linking results from our short string linker by default. This removes linking support for Arabic.
We made some improvements to our algorithm for calculating text embeddings across multi-word input, so you should see more accurate results for document and sentence-level vectors.
In addition to English and Spanish, we’ve added document and entity-level sentiment analysis support for Japanese language input (beta).
Expanded Language Support
The Rosette API now supports entity extraction in Vietnamese, short string social media entity linking in Chinese and Japanese, and Japanese-Chinese/Japanese-Korean/Korean-Chinese name matching and translation. We’ve also made some significant upgrades to our Japanese language support in general, which should be directly reflected in your results!
Did you catch the #25daysofrosette in December? Take a look to see Rosette in action with Star Wars scripts, WikiLeaks, and protecting against fake news. Have a cool project you’d like us to feature on our blog? Let us know.