Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Hierarchical Tree-based approach for creating Configurable and Static Deep Research Agent (Static-DRA)

Saurav Prateek

TL;DR

The paper tackles the limitations of static, two-stage RAG-based research pipelines by proposing Static-DRA, a configurable, tree-based static workflow that uses a Supervisor/Independent/Worker hierarchy to perform multi-hop information retrieval and parallel sub-topic exploration. It introduces two user-tunable parameters, Depth and Breadth, to balance research quality against LLM interaction costs. The approach is evaluated against the DeepResearch Bench via the RACE framework, demonstrating that larger depth/breadth settings yield more thorough research and larger reports, with competitive scores (e.g., 34.72 at depth=2, breadth=5 using gemini-2.5-pro). The work also provides a transparent, open-source implementation and a formal method to compute the number of topics assigned to workers, highlighting its practical utility for resource-aware deep research tasks.

Abstract

The advancement in Large Language Models has driven the creation of complex agentic systems, such as Deep Research Agents (DRAs), to overcome the limitations of static Retrieval Augmented Generation (RAG) pipelines in handling complex, multi-turn research tasks. This paper introduces the Static Deep Research Agent (Static-DRA), a novel solution built upon a configurable and hierarchical Tree-based static workflow. The core contribution is the integration of two user-tunable parameters, Depth and Breadth, which provide granular control over the research intensity. This design allows end-users to consciously balance the desired quality and comprehensiveness of the research report against the associated computational cost of Large Language Model (LLM) interactions. The agent's architecture, comprising Supervisor, Independent, and Worker agents, facilitates effective multi-hop information retrieval and parallel sub-topic investigation. We evaluate the Static-DRA against the established DeepResearch Bench using the RACE (Reference-based Adaptive Criteria-driven Evaluation) framework. Configured with a depth of 2 and a breadth of 5, and powered by the gemini-2.5-pro model, the agent achieved an overall score of 34.72. Our experiments validate that increasing the configured Depth and Breadth parameters results in a more in-depth research process and a correspondingly higher evaluation score. The Static-DRA offers a pragmatic and resource-aware solution, empowering users with transparent control over the deep research process. The entire source code, outputs and benchmark results are open-sourced at https://github.com/SauravP97/Static-Deep-Research/

A Hierarchical Tree-based approach for creating Configurable and Static Deep Research Agent (Static-DRA)

TL;DR

The paper tackles the limitations of static, two-stage RAG-based research pipelines by proposing Static-DRA, a configurable, tree-based static workflow that uses a Supervisor/Independent/Worker hierarchy to perform multi-hop information retrieval and parallel sub-topic exploration. It introduces two user-tunable parameters, Depth and Breadth, to balance research quality against LLM interaction costs. The approach is evaluated against the DeepResearch Bench via the RACE framework, demonstrating that larger depth/breadth settings yield more thorough research and larger reports, with competitive scores (e.g., 34.72 at depth=2, breadth=5 using gemini-2.5-pro). The work also provides a transparent, open-source implementation and a formal method to compute the number of topics assigned to workers, highlighting its practical utility for resource-aware deep research tasks.

Abstract

The advancement in Large Language Models has driven the creation of complex agentic systems, such as Deep Research Agents (DRAs), to overcome the limitations of static Retrieval Augmented Generation (RAG) pipelines in handling complex, multi-turn research tasks. This paper introduces the Static Deep Research Agent (Static-DRA), a novel solution built upon a configurable and hierarchical Tree-based static workflow. The core contribution is the integration of two user-tunable parameters, Depth and Breadth, which provide granular control over the research intensity. This design allows end-users to consciously balance the desired quality and comprehensiveness of the research report against the associated computational cost of Large Language Model (LLM) interactions. The agent's architecture, comprising Supervisor, Independent, and Worker agents, facilitates effective multi-hop information retrieval and parallel sub-topic investigation. We evaluate the Static-DRA against the established DeepResearch Bench using the RACE (Reference-based Adaptive Criteria-driven Evaluation) framework. Configured with a depth of 2 and a breadth of 5, and powered by the gemini-2.5-pro model, the agent achieved an overall score of 34.72. Our experiments validate that increasing the configured Depth and Breadth parameters results in a more in-depth research process and a correspondingly higher evaluation score. The Static-DRA offers a pragmatic and resource-aware solution, empowering users with transparent control over the deep research process. The entire source code, outputs and benchmark results are open-sourced at https://github.com/SauravP97/Static-Deep-Research/

Paper Structure

This paper contains 17 sections, 6 figures, 4 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. Static - Deep Research Agent
  3. Design Overview
  4. Core Design and Parameter Tuning
  5. Supervisor Agent
  6. Independent Agent
  7. Worker Node & Web Search Tool
  8. Worker Node
  9. Calling a Web Search Tool
  10. Researching the Topic with LLM
  11. Generated Report structure
  12. Evaluation
  13. Conclusion
  14. Computing the number of research topics being provided to the Worker
  15. Sample Deep Research Report
  16. ...and 2 more sections

Figures (6)

  • Figure 1: Independent agent breaking down the research topic into sub-topics.
  • Figure 2: Independent agent breaking down the research topic into maximum possible sub-topics irrespective of the value of breadth (b) parameter
  • Figure 3: The depth of the research being controlled by the depth (d) parameter.
  • Figure 4: Evaluation Metrics for multiple agent configurations.
  • Figure 5: Increase in the ability to perform in-depth research with increasing value of depth (b) and breadth (b) parameters.
  • ...and 1 more figures