Skip to main content

Uniphore Help Center Portal

Extract Data from File:

When extracting data from files, enter the entity name you want from the PDF. You can pick entities by clicking "+" or hitting "Enter," listing them visually.

For adding individual entities: here are the types of entities admins can extract from a document Text

Date

Money

Text refers to any pieces of text or string of characters

Normalization: It trims leading and trailing spaces, and replace consecutive spaces with a single space.

Date entity covers a specific point in time

Normalization: Convert date values to the year, month, and date format to be more scalable

Money entities represent monetary values

Normalization: Convert monetary values into the standardized currency code, it’s suitable for handling large numbers for Float and Integer data types. The accepted range is between -9,223,372,036,854,775,808 and 9,223,372,036,854,807

Money normalization and connectors also handles the following currencies even when they do not have associated numerical values (e.g., "USD" or "$" in value field will trigger the USD currency drop-down in the normalization field)

All 3-letter currency codes in ISO 4217

Currency symbols and variations for USD, INR, BRL, EUR, CAD, AUD, and GBP

Number-float

Floats are commonly used to handle numbers that include fractions, it can handle decimal numbers up to a 7-digit precision level

When more than 7-digit accuracy is needed, use the Money type instead

Number-integer

An integer is a whole number that does not include fractions.

The accepted range is between -2,147,483,648 and 2,147,483,647

Parent entity

A parent entity has a hierarchical relationship with other entities. For example: the parent entity is ‘Product category’ and the child entity is ‘Product Name’

Notes

Notes are used to provide context for users when annotating. It does not affect the prediction quality. The “*” name prefix is reserved and enforced for Notes.

Learn more about Nested Entity

Administrators have the ability to establish hierarchical relationships between entities for defining nested structures, such as table entities or nested lists. Once a parent entity is chosen and added, administrators can proceed to include child

entities. For instance, the parent entity could be 'Product Category,' and the child entity could be 'Product Name.' This apply to both tables and lists.

Adding a nested entity in workflows

During the “Extract data from file“ step, the admin can create an entity, then select “Parent entity“ from the entity type drop-down to make it a nested entity

All entity types are supported for child entities, including the “Notes“ type

Review Mode: In assisted mode for extraction workflows, administrators have the flexibility to adjust the confidence score threshold for nested entities as well.

Bulk Upload : Admin can choose to upload CSV to create entities in bulk: Download the sample file to create new entities in bulk “entity_name”: List how you would like to name the entities

“entity_type”: select type from the types supported above

“parent_entity_name”: for nested entities, put the parent entity name in this column

Tips for defining schema

The max number for schema size is 85 for simple entities and 32 tables (nested entities).

Schema name only allows lowercase letters, numbers, underscores "_", dashes "-", and single space “ “ “*” is allowed in the beginning for Notes entities