Kensho Classify Support
Do you have questions, or are you finding something tricky with our API? There are several ways to get help:
- Check for an answer in our FAQ
- Send us an email
We may already have an answer to your question! Please take a look through our FAQ below:
If you have a question not covered here, please don't hesitate to reach out to us at firstname.lastname@example.org.
Classify is Kensho’s market-leading natural language processing solution for finding the concepts represented within documents. Classify lets you find whether and where specific, customizable concepts occur in text and is specifically tuned for the needs of business and finance.
Classify is unique in its ability to quickly create highly performant custom concept models from a few dozen data points. Simply define your concept, give Classify 30 labeled data points, and you'll have a highly performant model ready for work in minutes. Classify takes all the difficulty out of text classification, letting you focus on what matters most.
For a product overview please visit our profile page.
Classify is a collection of advanced machine learning models that represent the conceptual content of documents to look for your specific topics. Kensho combines the latest advances in machine learning with S&P Global’s unparalleled data universe to train the models that underlie Classify.
Classify is designed for standard English documents comprising complete sentences. It will perform optimally on documents written with conventional grammatical structure, but is by no means limited to formal documents and performs well on everything from emails to research reports.
Classify’s performance is best in class, accelerating time to value and driving superior outcomes in any application. It trains faster, annotates faster, and produces better annotations. It more than meets the needs of serious workloads.
- Prediction Quality: Classify F1, precision, and recall is substantially better than open source and other vendor text classification solutions.
- Scalability: Classify can handle million-character payloads and concurrent requests without any impact on performance or reliability.
- Custom Model Creation: Classify requires a trivial amount of labeled data and needs no machine learning expertise. Classify offers performant models in hours, rather than weeks or months.
Classify is accessed via a simple REST API. Easily define your concepts, submit your text, and receive a list of annotated concepts in response in the form of a JSON. For more detailed information, please visit the API reference.
We have built and helped customers build other ways to consume Classify, such as extensions for applications such as Microsoft Office. If you are interested in such options, please contact us.
Names of any custom concepts should be descriptive and concisely capture the intent of the concept.
Yes! You can continuously update your Classify model with new custom concepts or refinements to existing concepts. Even better, Classify maintains a history of these model updates, so you can revert any model you created.
For an overview of usage, including parsing and interpretation of output, see our usage quick start
There is no limit to the number of documents or concepts you submit to Classify, but Classify currently limits requests to at most 20MiB each.
Kensho takes your privacy and security seriously. Data submitted to Classify is encrypted in transit and at rest, inaccessible to anyone but you, the user. Contact us to learn more.