Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

DANA: Domain-Aware Neurosymbolic Agents for Consistency and Accuracy

Vinh Luong, Sang Dinh, Shruti Raghavan, William Nguyen, Zooey Nguyen, Quynh Le, Hung Vo, Kentaro Maegaito, Loc Nguyen, Thao Nguyen, Anh Hai Ha, Christopher Nguyen

TL;DR

Application of DANA in physical industries such as semiconductor shows that its flexible architecture for incorporating knowledge is effective in mitigating the probabilistic limitations of LLMs and has potential in tackling complex, real-world problems that require reliability and precision.

Abstract

Large Language Models (LLMs) have shown remarkable capabilities, but their inherent probabilistic nature often leads to inconsistency and inaccuracy in complex problem-solving tasks. This paper introduces DANA (Domain-Aware Neurosymbolic Agent), an architecture that addresses these issues by integrating domain-specific knowledge with neurosymbolic approaches. We begin by analyzing current AI architectures, including AutoGPT, LangChain ReAct and OpenAI's ChatGPT, through a neurosymbolic lens, highlighting how their reliance on probabilistic inference contributes to inconsistent outputs. In response, DANA captures and applies domain expertise in both natural-language and symbolic forms, enabling more deterministic and reliable problem-solving behaviors. We implement a variant of DANA using Hierarchical Task Plans (HTPs) in the open-source OpenSSA framework. This implementation achieves over 90\% accuracy on the FinanceBench financial-analysis benchmark, significantly outperforming current LLM-based systems in both consistency and accuracy. Application of DANA in physical industries such as semiconductor shows that its flexible architecture for incorporating knowledge is effective in mitigating the probabilistic limitations of LLMs and has potential in tackling complex, real-world problems that require reliability and precision.

DANA: Domain-Aware Neurosymbolic Agents for Consistency and Accuracy

TL;DR

Application of DANA in physical industries such as semiconductor shows that its flexible architecture for incorporating knowledge is effective in mitigating the probabilistic limitations of LLMs and has potential in tackling complex, real-world problems that require reliability and precision.

Abstract

Large Language Models (LLMs) have shown remarkable capabilities, but their inherent probabilistic nature often leads to inconsistency and inaccuracy in complex problem-solving tasks. This paper introduces DANA (Domain-Aware Neurosymbolic Agent), an architecture that addresses these issues by integrating domain-specific knowledge with neurosymbolic approaches. We begin by analyzing current AI architectures, including AutoGPT, LangChain ReAct and OpenAI's ChatGPT, through a neurosymbolic lens, highlighting how their reliance on probabilistic inference contributes to inconsistent outputs. In response, DANA captures and applies domain expertise in both natural-language and symbolic forms, enabling more deterministic and reliable problem-solving behaviors. We implement a variant of DANA using Hierarchical Task Plans (HTPs) in the open-source OpenSSA framework. This implementation achieves over 90\% accuracy on the FinanceBench financial-analysis benchmark, significantly outperforming current LLM-based systems in both consistency and accuracy. Application of DANA in physical industries such as semiconductor shows that its flexible architecture for incorporating knowledge is effective in mitigating the probabilistic limitations of LLMs and has potential in tackling complex, real-world problems that require reliability and precision.
Paper Structure (34 sections, 3 figures, 5 tables)

This paper contains 34 sections, 3 figures, 5 tables.

Table of Contents

  1. Background & Motivation
  2. AI System Architectures Through a Neurosymbolic Lens
  3. Review of Recent Planning & Reasoning Literature
  4. Review of Recent Neurosymbolic Literature
  5. Analyses of Current AI Architectures
  6. AutoGPT and Related Frameworks & Tools
  7. LangChain ReAct
  8. OpenAI ChatGPT-4o & Assistant API
  9. Summary and Takeaways from Current Architectures
  10. DANA: Domain-Aware Neurosymbolic Agent
  11. Knowledge Store
  12. Program Store
  13. Program Search Process with Program Finder & Program Creator
  14. Knowledge Capture Process & Knowledge Representations
  15. Knowledge Capture
  16. ...and 19 more sections

Figures (3)

  • Figure 1: DANA Architecture. CAPTURE domain knowledge from experts. APPLY that to solve problems. Mediated by neurosymbolic building blocks.
  • Figure 2: Comparison of DANA and other agents. OpenSSA DANA outperforms alternative agent frameworks and tools on both Accuracy and Consistency.
  • Figure 3: Typical recipe analysis & recommendation from DANA-based Semiconductor Etching Advisor