Amazon Kendra

It sifts through various sources like manuals, research reports, FAQs, HR documents, and customer service guides spread across platforms such as Amazon Simple Storage Service (S3), Microsoft SharePoint, Salesforce, ServiceNow, RDS databases, or Microsoft OneDrive.

How Amazon Kendra works

Amazon Kendra indexes documents either directly or from third-party repositories, then intelligently presents relevant information using Natural Language Understanding (NLU) to our users. It can create an updatable index of diverse document types and seamlessly integrates with other services. For instance, we can leverage Amazon Kendra to power Amazon Lex chatbots, delivering helpful responses to user inquiries. Additionally, we can utilize an Amazon Simple Storage Service (S3) bucket as a data source, allowing Amazon Kendra to connect and index our documents.

Amazon Kendra has the following major components:

• Index: Holds our documents and makes them searchable.

• Data Source: Stores our documents. Amazon Kendra connects to the data source, and we can automatically synchronize a data source with an Amazon Kendra index so that our index stays updated with our source repository.

• Document Addition API: Adds documents directly to an index.

Use cases

Amazon Kendra is ideal for ever-evolving scenarios. It accelerates research and development by enabling scientists and developers in charge of new research and development to access data from previous projects buried deep within their corporate data sources. It can also be used on websites to easily find products. Additionally, it can be used internally for research.

Amazon Kendra supports the following common types of queries:

• Factoid questions (who, what, when, where): “Who is Google’s CEO?” or “When is the US’s Independence Day?” These questions require fact-based answers that may be returned as a single word or phrase.

• Descriptive question: “How do I connect my Echo Plus to my network?” The response might take the form of a single statement, a paragraph, or even a whole document.

• Keyword searches: “Health Benefits” or “IT Help Desk.” In cases where the intent and scope are unclear, Amazon Kendra will use its deep learning models to return relevant documents.

Accelerating research and development

A prominent application of Amazon Kendra is in expediting research and development efforts. It aids scientists and developers by granting access to data from past projects buried deep within their organization’s data sources.

Amazon Kendra addresses this issue by providing a powerful search capability to index and search vast amounts of data across different repositories. It uses machine learning to understand the context and content of the documents, enabling it to return highly relevant search results.

For example, a researcher working on a new project could use Amazon Kendra to quickly find relevant data from past projects, such as experiment results, research papers, and technical specifications. This not only speeds up the research process but also helps organizations leverage their existing knowledge to drive innovation.