GAIuS Data Format (GDF)

The data entry format for use with the GAIuS Framework is called GAIuS Data Format (GDF). GDFs have several fields that correspond to different types of data:

gdf = \
    {'strings' : [],
     'vectors': [],
     'emotives': [],
     'metadata': {},
     'path': []
    }

Fields

The following section briefly explain the different fields present in a GDF

strings

The strings field is used to observe data that is already symbolic (in string format) on a GAIuS Agent.

Example:

gdf['strings'] = ['temperature|54', 'UV_INDEX|HIGH', 'weather|sunny', 'time|afternoon']

vectors

The vectors field is used to observe numeric data on a GAIuS Agent. For example, this field may be used for flattened image data, mathmatical data, etc.

Example:

# image loaded previously into Python
flattened_image_data = image.flatten()
gdf['vectors'] = [flattened_image_data]

emotives

The emotives field is used to provide emotional data about an observation to the GAIuS agent.

metadata

The metadata field is used to pass through additional data about an observation that will not impact the predicting of a Cognitive Processor

path

The path field is used internally to show the path taken by an observation as it travels through a GAIuS agent’s topology

Example Conversions

To illustrate the process of converting data into GDF format, here are some specific examples corresponding to potential use-cases.

CSV to GDF

CSV Data can be rapidly converted into GDF format by utilizing the strings field of the GDF. Each row corresponds to a single data record, from which an individual GDF will be created. The headers, and the individual data elements from each row can be concatenated to make feature strings for each entry. For example, if a header value is “name”, and a specific data value is “Steve”, a feature string for the GDF would be “name|Steve”.

Here is a more comprehensive example:

Example HR CSV
Name	Age	Sex	Salary
Steve	34	male	50000
Josh	25	male	65000
Gregory	52	male	43000

Corresponding GDF Conversions
Strings	Vectors	Emotives
`"Name\|Steve", "Age\|34", "Sex\|male", "Salary\|50000"`
`"Name\|Josh", "Age\|25", "Sex\|male", "Salary\|65000"`
`"Name\|Gregory", "Age\|52", "Sex\|male", "Salary\|43000"`