Proj S17 – Handshakes – FELIX -Financial Document Insight Agent

Proj S17 – Handshakes – Financial Document Insight Agent

Introducing Proj S17 – Handshakes – FELIX -Financial Document Insight Agent

FELIX is a transformative search application powered by advanced Retrieval-Augmented Generation (RAG) and Large Language Models that can expedite research across extensive financial documents. It streamlines complex queries and precedence searches, dramatically reducing research time, and synthesises readable and accurate answers within a familiar chat user interface. This powerful engine is set to revolutionise conflict-of-interest research and analysis in the financial sector by providing swift and reliable insights.

Team members

Dhanush Kumar Manogar (ISTD), Dylan Raharja (ISTD), Hung Chia Yu (ISTD), Janani Sureshkumar (ESD), Lee Zhong Hern (ESD), Neville Wee (ISTD), Victoria Chong (ISTD)

Instructors:

Dorien Herremans

Writing Instructors:

Bernard Tan Chee Seng

Background

In today’s business world, understanding market trends is crucial, but dealing with heaps of data can be overwhelming. Traditional methods are slow, prone to mistakes, and can’t keep up with the pace of decision-making. We need a system that can quickly sift through data, extract important insights, and help businesses stay ahead of the competition by making informed choices. This system must be fast, accurate, and efficient to ensure timely and effective decision-making.

Problems with the current methods

Time Consuming

With thousands of documents to sieve through and limitations to our reading and comprehension speed, the entire process can take a very large amount of time.

Prone to errors and inaccuracies

As many hours will be spent on the analysis, accumulated fatigue can affect the concentration. This can lead to errors and inaccuracies in the insights generated.

Unsustainable

With long hours and accumulated fatigue, such a process is not ideal in the long run.

Solution

Our solution, FELIX, aims to improve market research efficiency using advanced technology called Retrieval-Augmented Generation (RAG) and Large Language Models. By employing sophisticated algorithms, FELIX accelerates research tasks and addresses issues like slow manual processes and errors. It streamlines queries, reduces research time, and ensures accurate insights through smart searches. With its user-friendly chat interface, FELIX enhances user experience, potentially transforming financial sector analysis and empowering businesses to make faster, informed decisions to stay competitive.

Design Journey

Problem Scoping & Literature Research

Analysis of the datasets were conducted especially on the different types of financial documents. The concept and methods of Information retrieval (IR) were explored.

Further Research and Experimentation

A question bank with questions pertaining to financial research was created. Different Information retrieval methods were tested out and evaluated with the question bank. Different PDF extraction methods were also tested out with the team deciding on Handshakes’s PDF extractor.

Prototyping and Development

A Hybrid Search method was chosen and developed upon as the primary IR method for FELIX. The development of the User Interface also begun in this phase.

Testing and Finetuning

Most of the development work had been concluded. Final adjustments were made to the front end and backend after user testing.

FELIX will ensure streamlined research. greater accuracy. elevated efficiency.

Key Features of FELIX

Filter and Search

Use the Chat Bot to gain financial insights

Data Pipeline

Acknowledgements

Special thanks go out to our esteemed mentors whose guidance and support have been invaluable throughout our journey. We extend our heartfelt appreciation to our company mentor, Paul Tan, for his unwavering encouragement and insights that have fueled our progress. We are immensely grateful to Prof. Dorien Herremans and Prof. Sudipta Chattopadhyay, our dedicated capstone mentors, whose expertise and mentorship have shaped our project and enriched our learning experience. Additionally, we express our gratitude to Bernard, our esteemed writing instructor, for his invaluable guidance in refining our communication skills. Thank you all for your dedication, wisdom, and belief in our potential.

Days

Hours

Minutes

Seconds

Vote the project now!

Vote for our project at the exhibition! Your support is vital in recognizing our creativity. Join us in celebrating innovation and contributing to our success. Thank you for being part of our journey!