Amazon Web Services’ (AWS) artificial intelligent search service Amazon Kendra has launched in the cloud giant’s Sydney region.
The managed service utilises machine learning (ML) to provide websites and applications with enterprise-level search functionality for data stored over multiple locations and content repositories as different file types.
Amazon Kendra also uses natural language queries, which AWS claims “is at the core of its search engine”, and can return direct answers to specific questions, while general queries can bring up related documents ranked by relevancy as well as relevant passages.
The deep learning models for natural language understanding can be adapted to properly identify industry-specific jargon, which currently includes terminology from the IT, financial services, insurance, pharmaceuticals, industrial manufacturing, oil and gas, legal, media and entertainment, travel and hospitality, health, HR, news, telecommunications, mining, food and beverage, and automotive industries.
“For example, a user searching for HR answers could enter 'deadline for filing HSA form' and Kendra would also search for 'deadline for filing health savings account form' for broader coverage to get the most accurate answer,” AWS claimed.
The relevancy of certain responses can also be altered to promote specific documents based on timeliness or popularity.
Features labelled as coming soon include the option to provide a specific list of relevant synonyms, incremental learning based on end user searches, query autocompletion and analytics.
This is the fourth launch region for Amazon Kendra, following releases in AWS’s North Virginia, Oregon and Ireland regions.
The managed service is available as part of AWS’s free tier, as well as a paid version.
In the free tier, Amazon Kendra’s developer edition is available for up to 750 hours for the first 30 days, offering storage for up to 10,000 documents, 4,000 maximum queries per day, five data sources and one availability zone. Past the 30 days, the developer edition is charged at US$2.50 per hour.
Unavailable as part of the free tier in the developer edition is connector usage, with document scanning running up US$0.000001 per document and conner usage when syncing costing US$0.35 per hour.
Meanwhile, the enterprise edition, offers document storage and maximum daily query allotments of 500,000 and 40,000, respectively, 50 data sources and three availability zones at a cost of US$7 per hour.
The enterprise edition also has the option of adding additional document storage or maximum daily query allotments at a price of US$3.50 each.
Connector usage for the enterprise edition meanwhile is priced identically to that of the developer edition.
Additionally, discounted pricing is available for users that make certain minimum volume commits, which is available directly through AWS.