Motivation

In a chat-based solution, users often use acronyms that are specific to their organization. These acronyms, while convenient, can be ambiguous and may lead to incorrect document retrieval if not properly understood by the system. To enhance the accuracy of document retrieval and ensure that the language model interprets queries correctly, we implement a background process that automatically replaces these acronyms with their full names. This process is invisible to the user but significantly improves the relevance and precision of the documents retrieved.

Setup Instructions

Step 1: Define a Set of Acronyms

Each institute or company must define a unique, case-sensitive list of acronyms. These acronyms should be carefully selected to avoid overlap and confusion within the organization. The defined list should include the acronym and its corresponding full name.

Examples:

1. HRPL - Human Resources Planning (HRPL)

2. TADM - Technical Administration (TADM)

3. FINM - Financial Management (FINM)

Step 2: Create and Validate Acronyms

• Case Sensitivity: Ensure that each acronym is case-sensitive to distinguish between similar terms. For example, "HRPL" and "hrpl" must be treated as distinct if necessary.

• Uniqueness: Acronyms must be unique within the company to avoid conflicting interpretations.

Step 3: Implement Acronyms via API

The acronym replacement feature is set up through API calls see Define Acronyms . The acronyms are passed in a JSON format as shown below:

This setup ensures that when a user types a query containing an acronym, the system will silently replace the acronym with its full name in the background, aiding in accurate document retrieval without altering the user's experience.