Overview of Aito

Aito is a database integrated with machine learning functions accessible through aito's APIs.

After you populated your data into the aito's database, aito analyzes the patterns of the data and help you to find interesting information that exists in the data or information that does not exist in the data but share similar patterns. You can better utilize aito's machine learning function by providing aito context which is some known information. Aito will use this context to find a more fitting pattern. For example, you can give aito the context of an user's information and ask aito to recommend the most suitable product for this specific user instead of a generic user.

Data modeling

Concepts

aito is a machine learning integrated database that serves documents in JSON format to the clients. aito provides REST APIs to interact with the database, whether to extract the data like traditional databases or to perform machine learning functions. The APIs uses queries in JSON format.

Data modeling

The aito database models your data, similar to the traditional relational databases, as a collection of tables. A table contains records (rows) that is a composite of field (column) and value pair. Aito supports nullable data which means it can handle missing data and missing link. Aito supports linking similar to relational database, however, the link between tables are used as information for machine learning functions and aito can handle linkage to missing data. Aito only supports structured data which means that you have to define aito's schema before populating the data and using the APIs. The schema contains information about tables' names, fields' names, and fields' data type.

Data format

Aito accepts 2 data format:

• JSON array for the batch upload API
• Ndjson for the File upload API as a ndjson file.

How does aito work?

At the core, aito is a statistical machine that analyzes the populated data into features with statistical properties. Aito uses these statistical properties to perform machine learning operations.

Handling documents

Documents in aito is analyzed into features based on an analyzer given by the user. The analyzer gives aito generic rules on how to processing the textual data.

Each feature is processed and is associated with statistical properties. You can access some of the statistical properties through the RELATE API, for instance, the frequency of the feature, the frequency given some condition, relationships with other features, etc.

What machine learning problems can aito solve?

Aito helps you to solve machine learning problems through its API whether it is getting analytics for decision makings, making predictions and suggestions, or automate tasks.

When using aito, instead of creating models that can only be used to solve a limited problem, you breakdown your problem into three parts, and translate it into query:

This creates advantages comparing to the traditional machine learning workflow

With aito, you don't have to create a specialized model and change the model if there are different applications. Instead, you solve your machine learning problems by formulating it as queries and send it to aito's API to get the result.

Aito's machine learning functions is explainable, meaning that aito can always explain its results. You only need to add one extra field to the query send to aito's API. The explanation are components that contribute to the final result.