Skip to content

Intelligent Data Extraction (IDE)

Looplex’s Intelligent Data Extraction (IDE) suite offers a range of intelligent, low-code (low code) services and functions to easily create your own data extraction process from structured documents, such as spreadsheets, forms, case cover pages, API responses, semi-structured documents, such as judicial distributor certificates, invoices, property registrations, and unstructured documents, such as opposing party petitions, judicial rulings, contracts generated in Word, etc.

How does it work?

The IDE features identify data relationships in a document as key-value pairs. For example, a petition contains various fields of variable data, such as addresses, party names, dates, case numbers, etc.

Using Looplex’s IDE functionalities, each key is mapped to a field type (variable) in the Template, and the value is assigned as the data for that variable.

The data for this variable can be assigned to a response, which is directly used in the Document construction, or it can be assigned to a hidden variable, which is then operated on to alter the document’s logic or calculated to generate other responses.

What IDE features are available?

The Intelligent Data Extraction options on the Looplex platform include:

  • Manual file import: Upload data via the interface from Excel, CSV, XML, JSON, or another previously generated document (response file).
  • Machine-to-Machine Document Interaction in Looplex: Call another document that is instantiated (template is initiated and executed), allowing the calling template to operate its entire logic, extract responses, and request response calculations based on provided parameters.
  • RPA (Robotic Process Automation): A service that allows you to configure a computer program or “robot” to emulate human actions interacting with websites, applications, or other digital systems to perform tasks on a computer. Looplex’s RPA robots use the user interface to capture data and manipulate applications just as humans do. They interpret, trigger responses, and communicate with other systems to perform a wide variety of repetitive tasks.

Looplex’s robot already has pre-trained data extraction and task execution skills. You can directly provide relevant instructions for what you want it to do in the context of a specific Template, or you can train a new Skill.

You can also access other RPA solutions from within Looplex via API, such as Microsoft Power Automate. See the API documentation for more details.

  • API Tube: An API is a computing interface that defines the interactions between different software. It specifies the types of calls or requests that can be made, how to make them, the data formats to be used, and the conventions to follow. Unlike RPA, which emulates humans interacting with a machine, an API is the native machine-to-machine communication protocol.
  • Assisted Document Analysis: A separate interface in the Document construction process in Looplex, where the user is presented with a PDF file and a sequential interview structure similar to Looplex’s guided document interview. The user marks the value for each key requested in the assisted analysis interview or validates the values found by Looplex’s artificial intelligence algorithm or text parser (human-in-the-loop technique).

The validated results are then assigned to the Document string, with the user returning to the main interview for the construction process.

  • Doc Parser: A service for analyzing and decomposing phrases and word sequences into their constituents, resulting in a parse tree that shows their syntactic relationship, which may also include semantic and other information. You can use regular expressions (RegEX), scripts, and other methods to extract desired key values from the parse tree.
  • Automated AI Extraction: Services using optical character recognition (OCR), machine learning, natural language processing (NLP), and deep learning to classify, categorize, and extract relevant information from unstructured documents.

The results of automated extraction can be reconciled and corrected by humans using the Assisted Document Analysis service, which is always recommended if the accuracy achieved is not better than the human average error rate for equivalent tasks. Mapping and result quality improve over time as the service learns from human feedback.

Where to start?

Intelligent Data Extraction offers a powerful set of tools for use in your Templates. But before starting to use any of them, individually or in combination with others, consider what you aim to achieve. For example:

  • What type of input or external documents do you want to process?
  • Who will be responsible for reconciling and correcting the results?
  • Where do you want to display the data afterward?
  • How do you want them to be used in document construction?