RAG-Powered Enterprise Search Across A Database

Problem

Our client is a large company with hundreds of employees, multiple departments, and dozens of ongoing projects at any given time. Each department logs the progress on projects, approaches and work experience into a company-wide database — a collection of information on each completed and ongoing project. Employees use the database to get updates on projects from other departments and exchange work experience.

The knowledge base contains thousands of documents, making searching for relevant pieces of information difficult. Coupled with confidential data and multiple levels of data access rights, the database is an inefficient tool for data exchange.

The company approached us to develop a smart search system for the database which would help employees find information faster as well as restrict access to confidential or sensitive information based on the employee access level.

Solution

The chatbot-like search module lends itself perfectly to the use of natural language processing models. However, given the confidential nature of the information stored in the database, using models like ChatGPT is not an option due to data leaking concerns.

We have implemented a different approach to natural language processing.

Chatbot For Smart Database Search

ChatGPT is the most powerful language model in the world, but it doesn’t mean there are no other solutions to processing natural language. Taking the project requirements in mind, we have implemented multiple smaller language models which perform just as good in this particular application.

The smart search system is powered by a collection of multiple smaller language models. When a user asks a question, the system uses a retrieval language model which looks for parts of text which might contain the answer. After the texts have been located, a reranking model ranks them in order of relevance. The system can also combine text from multiple documents to provide a complete answer.

The system also restricts access to sensitive information based on employee access level or department.

Results

The smart database search decreased the time employees spend looking for information by 35%, not only making the search process more efficient, but helping employees get a deeper understanding of the company's processes and projects.