Hyperscience Introduces Text Classification (Preview)

Bastiane Huang | Posted on July 4, 2022

Customers can now derive valuable insights from texts in documents, emails, and more.

Today, Hyperscience announced Text Classification in preview. This feature enables customers to analyze and organize unstructured texts by user intent, sentiment, topic, or any custom labels based on their own business rules. For example, insurance companies can classify disability descriptions by medical conditions and route to the most appropriate teams for claim processing. Banks can classify emails by client intent (e.g., account balance, change of address) and sentiment so they can prioritize urgent customer complaints. Or a customer support organization can categorize inbound requests by problem type based on the way users have described their issues.

Since its inception in 2014, Hyperscience has helped customers in finance, insurance, and public sectors including TD Ameritrade, Department of Veterans Affairs, Social Security Administration, and the US Air Force to extract printed and handwritten text from documents. With the introduction of Text Classification, we are taking it to the next level by helping customers make sense of the data extracted and uncover actionable insights.

In the standard workflow of today’s email triage, organizations typically dedicate teams to manually review emails or documents to understand context, enabling prioritization and routing of tasks  for downstream processing. However, the immense amount of documents and email is making it difficult for organizations to keep up. Some organizations attempt to automate the process with simple keyword search which, most of the time, falls short of the required accuracy as well as flexibility and requires constant updates every time business rules change.

With Hyperscience Text Classification, organizations can now automatically categorize and route texts from documents, emails, and chats by user intent, sentiment, topic, and business-specific labels. Using Hyperscience’s workflow automation and classification results, organizations can even go a step further and automate validation, decision making, and database updates. As an example, when a bank processes inbound requests with Hyperscience, the platform can first run Text Classification and categorize an email by intent, e.g. “change of address,” and then automatically extract the new address as well as client name, and verify that all documents required to process this request are attached.

Text Classification can also be used to categorize free-text descriptions of symptoms by medical condition, accelerating the health claims approval process. Demo videos for medical condition classification and email intent classification are available upon request.

Hyperscience Text Classification is available today in preview. To get access to Text Classification (Preview), please contact your Hyperscience representative or send us an email at preview@hyperscience.com.

To learn more about Text Classification, please read our documentation or email your questions to preview@hyperscience.com.