Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Towards Energy-aware Requirements Dependency Classification: Knowledge-Graph vs. Vector-Retrieval Augmented Inference with SLMs

Shreyas Patil, Pragati Kumari, Novarun Deb, Gouri Ginde

Abstract

The continuous evolution of system specifications necessitates frequent evaluation of conflicting requirements, a process that is traditionally labour intensive. Although large language models (LLMs) have demonstrated significant potential for automating this detection, their massive computational requirements often result in excessive energy waste. Consequently, there is a growing need to transition toward Small Language Models (SLMs) and energy aware architectures for sustainable Requirements Engineering. This study proposes and empirically evaluates an energy aware framework that compares Knowledge Graph-based Retrieval (KGR) with Vector-based Semantic Retrieval (VSR) to enhance SLM-based inference at the 7B to 8B parameter scale. By leveraging structured graph traversal and high dimensional semantic mapping, we extract candidate requirements, which are then classified as conflicting or neutral by an inference engine. We evaluate these retrieval enhanced strategies across Zero-Shot, Few-Shot, and Chain of Thoughts prompting methods. Using a three-pillar sustainability framework measuring energy consumption (Wh), latency (s), and carbon emissions (gCO2eq) alongside standard accuracy metrics (F1 Score), this research provides a first systematic empirical evaluation and trade off analysis between predictive performance and environmental impact. Our findings highlight the effectiveness of structured versus semantic retrieval in detecting requirement conflicts, offering a reproducible, sustainability aware architecture for energy efficient requirement engineering.

Towards Energy-aware Requirements Dependency Classification: Knowledge-Graph vs. Vector-Retrieval Augmented Inference with SLMs

Abstract

The continuous evolution of system specifications necessitates frequent evaluation of conflicting requirements, a process that is traditionally labour intensive. Although large language models (LLMs) have demonstrated significant potential for automating this detection, their massive computational requirements often result in excessive energy waste. Consequently, there is a growing need to transition toward Small Language Models (SLMs) and energy aware architectures for sustainable Requirements Engineering. This study proposes and empirically evaluates an energy aware framework that compares Knowledge Graph-based Retrieval (KGR) with Vector-based Semantic Retrieval (VSR) to enhance SLM-based inference at the 7B to 8B parameter scale. By leveraging structured graph traversal and high dimensional semantic mapping, we extract candidate requirements, which are then classified as conflicting or neutral by an inference engine. We evaluate these retrieval enhanced strategies across Zero-Shot, Few-Shot, and Chain of Thoughts prompting methods. Using a three-pillar sustainability framework measuring energy consumption (Wh), latency (s), and carbon emissions (gCO2eq) alongside standard accuracy metrics (F1 Score), this research provides a first systematic empirical evaluation and trade off analysis between predictive performance and environmental impact. Our findings highlight the effectiveness of structured versus semantic retrieval in detecting requirement conflicts, offering a reproducible, sustainability aware architecture for energy efficient requirement engineering.
Paper Structure (13 sections, 5 equations, 5 figures, 5 tables, 1 algorithm)

This paper contains 13 sections, 5 equations, 5 figures, 5 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Background (Preliminaries)
  3. Methodology
  4. Total Carbon per Dataset per Retrieval Pipeline
  5. EcoScore
  6. Results
  7. KGR Vs.VSR - comparison across environmental sustainability metrics (RQ1)
  8. Impact of different prompting strategies on SLMs (RQ2)
  9. Impact of KGR/VSR-Augmented Pipeline on Sustainability-to-Performance Ratio
  10. Discussion
  11. Related work
  12. Threats to Validity
  13. Conclusion and Future work

Figures (5)

  • Figure 1: Overall study design: Our proposed KGR approach is compared with VSR and Baseline using five datasets and various evaluation measures.
  • Figure 2: Knowledge graph–based retrieval via shared entities. Two requirements, Requirement R1: The UAV shall land automatically when Pilot communication is restored and estimated flight time lapsed is more than 5 minutes and Requirement R2: The UAV shall land automatically when Pilot communication is lost and the estimated flight time remaining is more than 5 minutes, are connected through overlapping structured entities, enabling explainable graph traversal.
  • Figure 3: RQ1: Computing K for IBM-UAV performance comparison of KGR and VSR pipelines (Recall@K), the curves plateu at 20. Similar analysis was done for other datasets to choose K value for further evaluation.
  • Figure 4: RQ1 – Comparative evaluation of retrieval pipelines (KGR and VSR) across datasets.
  • Figure 5: Answering RQ2 - Macro F1-score and Total carbon emission for the three SLMs used for evaluating the five datasets. Mistral SLM fairs well on both for all the datasets