EmpiRE-Compass – A Neuro-Symbolic Dashboard for Navigating the Knowledge Landscape of Empirical Research in Requirements Engineering

EmpiRE-Compass Dashboard

GitHub Repository

Empirical research in Requirements Engineering (RE) continues to grow rapidly, accompanied by an increasing number of secondary studies that attempt to synthesize this expanding body of evidence. Yet, despite this growth, literature reviews (LRs) in RE and Software Engineering (SE) often remain isolated and difficult to update. Their underlying data and artifacts are rarely shared, limiting replication, reuse, and long‑term sustainability. The rise of generative AI has further amplified the production of LRs—frequently at the expense of rigor, transparency, and traceability—making the need for sustainable, data‑driven, and reproducible review practices more urgent than ever.

EmpiRE‑Compass addresses these challenges by providing a neuro‑symbolic dashboard that unifies semantically structured LR data in research knowledge graphs (RKGs) with the generative and interpretive capabilities of large language models (LLMs). Building on KG‑EmpiRE—a community‑maintainable RKG of empirical research in RE—EmpiRE‑Compass lowers the technical barriers for accessing, exploring, synthesizing, and reusing LR data. Instead of requiring users to write SPARQL queries or navigate complex graph schemas, the dashboard enables intuitive, natural‑language interaction and dynamic visual analytics.

Our long‑term goal is to maintain EmpiRE‑Compass collaboratively with the research community to provide a comprehensive, up‑to‑date, and long‑term available overview of empirical research in RE. By combining RKGs and LLMs, EmpiRE‑Compass enables replicable, reusable, and sustainable literature reviews that support transparency, cumulative knowledge building, and continuous updating.

EmpiRE‑Compass is openly available online and integrates all curated data from KG‑EmpiRE, which currently covers 776 papers published at the IEEE International Requirements Engineering Conference (1993–2025), described with 53,727 triples, 75,727 resources, and 31,727 literals. The dashboard also incorporates a second use case on empirical research in NLP4RE, demonstrating its extensibility to additional datasets. All data is maintained in the Open Research Knowledge Graph (ORKG) as part of the ORKG Observatory on Empirical Research in Software Engineering, ensuring FAIR, long‑term availability.

The dashboard provides three core capabilities:

  1. Exploratory visual analytics for curated competency questions, including predefined and dynamic visualizations enriched with interpretations and explanations.
  2. Neuro‑symbolic synthesis for custom competency questions, where LLMs generate SPARQL queries, visualizations, interpretations, and explanations directly from RKG data.
  3. Reusable knowledge, with all queries, analyses, visualizations, and underlying data openly available for replication, reuse, and sharing.

EmpiRE‑Compass supports two main workflows. In the curated workflow, users select predefined competency questions derived from prior research on empirical RE. The dashboard executes the corresponding SPARQL queries, processes the retrieved data, and presents interactive visualizations with manually validated interpretations and explanations. In the custom workflow, users pose natural‑language questions tailored to their needs. The LLM generates SPARQL queries consistent with the underlying graph schema, retrieves and visualizes the data, and provides contextual interpretations and explanations. Both workflows support iterative refinement, export and import of results, and detailed history tracking to ensure transparency and reproducibility. All source code, documentation, and configuration files are openly available in the EmpiRE‑Compass GitHub repository. This openness fosters reuse, adoption, and extension, enabling researchers to adapt the dashboard to new datasets, integrate additional competency questions, or experiment with different LLMs.

EmpiRE‑Compass demonstrates how LR data can be made transparent, reusable, and sustainable by combining semantically structured representations with neuro‑symbolic interaction. By lowering technical barriers and enabling dynamic, human‑centered access to RKG data, EmpiRE‑Compass supports collaborative, continuously updated, and reproducible literature reviews—advancing the quality, reliability, and timeliness of empirical research in RE, SE, and beyond.