What is Amazon Kendra?
Amazon Kendra is an enterprise search capability powered by machine learning. It is highly accurate and easy to implement. Kendra delivers powerful natural language search capabilities and you do not need to have Machine learning skills to set it up. Natural language enables you to get you more specific answers anywhere from your data. Currently available solutions are all keyword engines, which means they look for a keyword and return all the documents where they find that word, which means a lot of false positives and having to search through a lot of data before getting the answer you are looking for.
- AWS account
- Data if you have any (on which you want to index your search).
How does it Work?
- Create an index (takes up to 30 minutes).
- Add data sources.
- Test and Deploy.
What types of questions or queries can Kendra answer?
Factoid: Who, what, when, or where questions. Questions that have a specific answer based on fact and can have resolutions in a single word or phrase. Ex: Who is the first president of India?
Descriptive: Questions whose answers could be in a phrase or sentence or a paragraph from a document. Ex: How to install python on mac?
Keyword: Questions that have limited scope or intent, where no index or direct reference in documents isn’t found. Ex: Essentials skills for Devops.
Why use Amazon Kendra?
Ease of use: It is simple to set up, and you can bring your documents in various formats (ppt, pdf, text files, etc.) and implement search capabilities into multiple materials with a few clicks.
Accurate: Amazon Kendra is powered by Machine learning, which improves your search results over time by deep learning.
Seamless integration: Connects with 3rd party data sources such as OneDrive, Salesforce Online, to name a few.
Kendra comes in 2 editions.
- Developer Edition
- Enterprise Edition