Skip to main content

AI RAG Agents & Academic Knowledge Graphs

EcoLab's Literature Review module is designed to fully automate the process of searching, evaluating, and synthesizing academic papers related to your research topic. This module uses Retrieval-Augmented Generation (RAG) combined with Knowledge Graphs powered by Neo4j to ensure academic accuracy and eliminate AI hallucinations.

1. Academic Search and Filtering​

The platform connects directly to large-scale open academic databases (Semantic Scholar, ArXiv) to retrieve relevant literature:

  1. Navigate to the Literature Review module.
  2. Enter your research question or topic into the search bar. The AI agent will automatically combine this with the selected project idea as the background context.
  3. Click the Chạy tổng quan (Run Review) button.
  4. Search results will populate the Bài báo (Papers) tab with detailed information:
    • Paper title, authors list, journal, year.
    • Citation count — a key metric for measuring paper authority.
    • Verification status.

2. Academic Synthesis Modules​

After retrieving the papers, EcoLab's RAG agent analyzes the abstracts and full texts (if available) to synthesize findings into 5 structured formats:

Literature Matrix​

A structured matrix that allows you to compare previous studies at a glance:

  • Author & Year: Identifies the timeline and research group.
  • Research Question: The core question addressed by the paper.
  • Econometric Methodology: The estimators used (e.g., OLS, GMM, FE/RE).
  • Key Findings: The most important quantitative results of the study.
  • Variables & Identification Strategy: Key variables included in the model.

Research Gaps​

The AI agent analyzes the collection of papers to identify remaining knowledge gaps that your study could fill:

  • Geographic/Contextual Gaps: Studies have not been conducted in developing countries or specifically in Vietnam.
  • Methodological Gaps: Lack of robustness tests or endogeneity corrections.
  • Temporal Gaps: Datasets used in previous studies are outdated and do not reflect recent economic shifts (e.g., post-pandemic or post-financial crisis).

Research Objectives & Hypotheses​

The AI automatically translates your selected idea and reference papers into a clear set of objectives:

  • General & Specific Objectives.
  • Scientific Hypotheses: Clearly defines testing hypotheses (e.g., $H_0$, $H_1$) to form the basis for econometric modeling.

Proposed Model (Variables & Estimator)​

Based on the selected papers, the AI suggests a model specification:

  • Editable Variables Table:
    • Variable Name: Name of the indicator (e.g., ROA, Leverage, Size).
    • Variable Role: Dependent, Independent, or Control.
    • Measurement: How the indicator is calculated.
    • Expected Sign: Negative ($-$), positive ($+$), or ambiguous ($?$) based on economic theory.
  • Suggested Estimator: Suggests OLS, Fixed Effects, or GMM depending on the proposed data structure.

Knowledge Graph Viz​

Uses Neo4j to draw interactive visual network maps connecting:

  • Citation relationships between papers.
  • Semantic relationships between research concepts.
  • The most frequently applied methodological clusters.