عربي БГ বাংলা ČS DA DE ΕΛ EN ES ET FI FR GA हिन्दी HR HU IT 日本語 LT LV MT NL PL PT RO РУ SK SL SV 中文

AI Challenge

Design a CPU-only, purely symbolic Question/Answering system whose knowledge graph is built automatically from natural language documents.

Green AI • CPU-only • Symbolic • Explainable

1. Challenge Summary

The challenge aims to design, implement and evaluate a question/answering (QA) system running exclusively on CPU, based on a purely symbolic architecture (lexicon, grammars, knowledge graph, formal logic). The entire system development is driven by quantised open-weights LLMs, executable on consumer hardware (i9 CPU, RTX 3090).

This project falls within the Green AI movement (Schwartz et al., 2020) and Frugal AI, in opposition to the race towards ever larger models whose carbon footprint has become alarming (Kim et al., 2026; Hasan et al., 2025). It also addresses the growing demands for transparency, explainability and regulatory compliance imposed by the GDPR and the European AI Act (Wagner et al., 2025; Juliussen, 2025).

Originality: No existing work simultaneously combines the following constraints:
  1. Purely symbolic QA system (no neural network at inference)
  2. CPU-only at runtime (no GPU required)
  3. Automatic KG construction from natural language documents
  4. Development entirely supervised by quantised open-weights LLMs
  5. Native GDPR/AI Act compliance by design (intrinsic interpretability)
  6. Ethical license (Hippocratic 3.0)

2. Context and Scientific Positioning

2.1. Problem Statement

Current QA systems rely heavily on LLMs (GPT, LLaMA, etc.) that require expensive GPUs for inference, and raise issues of hallucination, reproducibility, explainability and data sovereignty. They operate as black boxes and are difficult to make compliant with the requirements of the European AI Act (Article 13) regarding the transparency of high-risk systems (Wagner, 2025; Boshnjakovski, 2025).

2.2. State of the Art

Neuro-symbolic Approaches and KG-QA

Recent literature converges towards integrating knowledge graphs (KG) and symbolic reasoning to address the limitations of purely neural approaches:

Automatic Knowledge Graph Construction

CPU Efficiency and Lightweight Approaches

Ecological Impact of AI

GDPR and AI Act Compliance

3. Objectives

Main Objective

Demonstrate the feasibility and evaluate the performance of a CPU-only, purely symbolic QA system whose knowledge graph is built automatically from natural language documents.

Secondary Objectives

  1. Achieve competitive performance on recognised KGQA benchmarks.
  2. Measure and compare the energy footprint (CPU vs GPU) to demonstrate the ecological benefit.
  3. Demonstrate the intrinsic interpretability of the system (traceability of each answer).
  4. Provide reproducible, documented software under an ethical license.

4. Architecture & Technical Specifications

4.1. Overall Architecture

The system is divided into two distinct phases:

Phase 1 — Knowledge Graph Construction (offline, GPU allowed)

NL Documents → Preprocessing → NER → Relation Extraction → Coreference Resolution
    → KG Construction (RDF/OWL triples) → Graph Storage

Phase 2 — Question/Answering (runtime, strict CPU-only)

NL Question → Syntactic/Semantic Analysis → Transformation to Formal Query
    → KG Traversal → Logical Reasoning → Answer Generation
    → Explanation Trace

4.2. Hardware Specifications

ComponentPhase 1 (KG Construction)Phase 2 (QA Runtime)
CPUIntel Core i9 (max)Intel Core i9 (max)
GPUNVIDIA RTX 3090 (24 GB VRAM)None (strict CPU-only)
RAM32 GB (max)32 GB (sufficient)
StorageNVMe SSDNVMe SSD

Justification: Phase 1 may use the GPU to run the quantised open-weights LLMs that supervise pipeline construction. Phase 2, which constitutes the deployable system, must be strictly CPU-only to guarantee accessibility, sovereignty and frugality.

4.3. Development Environment

4.4. LLM Supervision

System development is entirely driven by one or more open-weights LLMs, executable locally on the specified hardware (Phase 1). The choice of models, quantisation and inference framework is left to the participant.

5. Suggested Pipeline Architecture

The following breakdown is provided as a guideline. Participants are free to propose a different architecture, provided they meet the functional constraints (sections 3, 4 and 12) and license requirements (section 6).

5.1. Text Preprocessing

  • Sentence segmentation
  • Tokenisation
  • Lemmatisation
  • Part-of-Speech tagging
  • Dependency parsing

5.2. Named Entity Recognition (NER)

  • Entity detection (persons, locations, organisations, dates, concepts)
  • Entity typing according to a domain ontology
  • Coreference resolution

5.3. Relation Extraction (RE)

  • Triple extraction (subject, predicate, object) from text
  • Open or pattern-guided extraction

5.4. Knowledge Graph Construction

  • Entity normalisation and merging (entity linking)
  • Graph construction and storage (triples)
  • Enrichment through implicit relation inference
  • Graph validation and cleaning

5.5. Question Understanding

  • Syntactic analysis of the natural language question
  • Question type identification (who, what, where, when…)
  • Extraction of mentioned entities and relations
  • Transformation into a formal query on the graph

5.6. Reasoning & Answer Search

  • Knowledge graph traversal (multi-hop reasoning)
  • Logical resolution and ranking of candidate answers

5.7. Answer Generation & Explanations

  • Answer formulation in natural language
  • Complete explanation trace: path in the KG, applied rules, involved entities and relations, confidence/completeness

6. License Constraints

6.1. Hippocratic License 3.0 (HL3)

The final software will be published under Hippocratic License 3.0 (firstdonoharm.dev). This license:

6.2. Dependency Compatibility

LicenseHL3 CompatibleExamples
MIT✅ Yes—
BSD (2/3-clause)✅ Yes—
Apache 2.0✅ Yes—
ISC / CC0✅ Yes—
GPL v2/v3⚠️ Potentially incompatible—
AGPL❌ Probably incompatible—

6.3. Recommended Optional HL3 Modules

7. Regulatory Compliance

7.1. GDPR

GDPR PrincipleImplementation
Data minimisation (Art. 5.1.c)The KG stores only relevant extracted triples, not raw personal data.
Right to explanation (Art. 22 + recital 71)Complete explanation trace for each answer.
Right to erasure (Art. 17)Ability to delete entities/triples from the KG.
Privacy by design (Art. 25)No sensitive data in the KG unless documented necessity.
No automated profiling (Art. 22)The system does not make automated decisions about individuals.

7.2. European AI Act

AI Act RequirementCompliance
Transparency (Art. 13)Intrinsically interpretable symbolic architecture — every reasoning step is traceable.
Explainability (Art. 13.1)Complete trace: question → formal query → KG path → answer.
Technical documentation (Art. 11)Fully documentable symbolic pipeline (grammars, rules, ontology).
Robustness and accuracy (Art. 15)No hallucination — answers come exclusively from the KG.
Human oversight (Art. 14)The KG is inspectable and modifiable by a human.
Risk management (Art. 9)Low risk: no opaque model, no emergent bias.
Key argument: A purely symbolic system is natively compliant with the transparency and explainability requirements of the AI Act, unlike LLM-based systems which require additional XAI layers (Goncalves & Correia, 2026; Wagner et al., 2025).

8. Ecological Impact

8.1. The Problem

8.2. Advantage of Symbolic CPU-only

MetricLLM System (GPU)AI Challenge (CPU-only)Estimated Gain
Instantaneous Power300-400 W (GPU)65-125 W (CPU)~3-5×
Latency per Query0.5-5 s< 100 ms (target)~5-50×
Energy per Query~0.001-0.01 kWh~0.000001-0.00001 kWh~100-1000×
Required HardwareGPU ≥ €6,000Consumer CPU < €500~12×

8.3. Mandatory Measurement

The project must measure and report:

9. Reproducibility

  1. Complete source code: published on a public repository, under HL3 license.
  2. Installation instructions: automated script or containerisation.
  3. Test data: exclusive use of public benchmarks.
  4. Random seeds: fixed and documented for any stochastic component.
  5. Dependency versioning: pinned dependency file.
  6. Containerisation: the complete system must be reproducibly executable.
  7. Documented minimum hardware: explicit minimum specifications.

The maximum required hardware (i9 + RTX 3090) is accessible to a wide audience. Phase 2 (QA runtime) requires no GPU, enabling execution on a simple laptop.

10. Sovereignty, Security & Privacy

Digital Sovereignty

  • No cloud dependency — fully local operation.
  • No external API calls — no data leaves the machine.
  • Open-weights LLMs downloadable and executable locally.
  • No vendor lock-in.

Security

  • Sandboxed execution (isolated environment).
  • No arbitrary code execution.
  • No prompt injection (no LLM at inference).
  • Input validation via formal grammar.

Privacy

  • Exclusive local processing — documents never leave the machine.
  • No telemetry.
  • KG erasable on demand.
  • Ability to instantiate separate KGs per user or per domain.

11. Benchmarks & Evaluation

11.1. Recognised KGQA Benchmarks

The system must be evaluated on at least two of the following benchmarks:

BenchmarkDescriptionSizeReference
SimpleQuestionsSimple factual questions on Freebase~100K questionsACL
WebQuestionsSPWeb questions with SPARQL annotations~4,737 questionsACL
LC-QuAD 2.0Complex questions on DBpedia/Wikidata~30K questionsISWC
ComplexWebQuestionsMulti-hop questions on Freebase~34K questionsNAACL
QALDMultilingual challenge, small size~400-500 questionsCLEF/ESWC
MINTQARecent benchmark for LLM+KG on rare entitiesVariableEMNLP 2025
SQuAD 2.0Extractive QA on paragraphs~150K questionsACL
Natural QuestionsReal Google questions~307K questionsTACL
TriviaQATrivia QA with supporting documents~95K questionsACL

Recommendation: prioritise LC-QuAD 2.0 and QALD, supplemented by SQuAD or Natural Questions for comparison with neural approaches.

11.2. Evaluation Metrics

CategoryMetricDescription
QA PerformanceF1 score, Exact Match (EM)Answer precision
QA PerformanceHits@1, Hits@5Correct answer rate in top-k
QA PerformanceMRRMean rank of correct answer
LatencyAverage/p95 response time (ms)Measured on CPU only
ThroughputQuestions/secondThroughput
EcologykWh/query, gCO₂/queryVia codecarbon/carbontracker
ExplainabilityTraceability rate% of answers with complete trace (target: 100%)
KG CoverageTriples, entities, relationsDescriptive statistics

11.3. Comparison Baselines

  1. RAG + LLM (GPU): neural baseline on GPU.
  2. KG-RAG (GPU): hybrid KG + LLM baseline.
  3. Direct SPARQL: pure symbolic baseline (manual queries on the same KG).

12. Performance Specifications

CriterionTargetJustification
QA Latency< 100 ms (p95)Acceptable user interactivity
KG Construction< 1 min / 1,000 short documentsPracticality
QA Memory Footprint< 4 GB RAMExecutable on modest PC
KG SizeScalable to 10M+ triplesSufficient coverage for benchmarks
QA Precision≥ 50% F1 on simple benchmarksInterpretability and frugality as main arguments
Traceability100% of answers explainedKey advantage of symbolic approach

13. Risks & Mitigations

RiskProbabilityImpactMitigation
Insufficient QA performance vs LLMHighMediumFocus on interpretability/frugality
Insufficient quality of automatic KGMediumHighIterate rules, LLM supervision
Multi-hop question complexityHighMediumLimit to ≤ 3 hops, document limitations
Dependency license incompatibilityLowHighSystematic license audit
Open-weights LLM insufficient for supervisionMediumMediumTest and combine multiple models
Limited KG coverage at scaleMediumMediumStart with QALD, then scale up

14. Success Criteria

  1. The QA system operates entirely on CPU with no GPU at inference.
  2. The KG is built automatically from natural language documents.
  3. Performance is measured on ≥ 2 recognised public benchmarks.
  4. 100% of answers are accompanied by an explanation trace.
  5. Energy consumption is measured and compared to a GPU baseline.
  6. Code is published under Hippocratic 3.0 license with compatible dependencies.
  7. The system is reproducible via Docker on the specified hardware.
  8. Development is entirely carried out by open-weights LLMs in a sandbox.

15. Bibliographic References

  1. Agarwal, P. & Bedathur, S. (2025). A Zero-Shot Neuro-Symbolic Approach for Complex Knowledge Graph Question Answering. EMNLP 2025 Findings.
  2. Ansar, W., Goswami, S. & Chakrabarti, A. (2024). From Transformers to LLMs: A Systematic Survey of Efficiency Considerations in NLP. arXiv:2406.16893.
  3. Awab, S. (2026). AI's Resource Gluttony and Sustainable Development. SSRN 6301563.
  4. Boshnjakovski, V. (2025). Balancing Compliance and Innovation: Addressing Explainability in AI Systems under the EU AI Act. MBA Thesis.
  5. Chaudhary, V. et al. (2025). Fusing Knowledge and Language: A Comparative Study of KG-Based QA with LLMs. arXiv:2509.09272.
  6. Chen, H. et al. (2025). Open Local Knowledge Graph Construction from Academic Papers Using Generative LLMs. TheWebConf 2025.
  7. Cheng, K. et al. (2025). Neural-symbolic methods for knowledge graph reasoning: A survey. ACM Trans. Knowl. Discov. Data.
  8. Chudasama, Y. et al. (2025). Toward interpretable hybrid AI: integrating knowledge graphs and symbolic reasoning. IEEE Access.
  9. Corradini, F. et al. (2025). State of the art and future directions of small language models. Big Data and Cognitive Computing.
  10. Goncalves, A. & Correia, A. (2026). XAI-Compliance-by-Design: A Modular Framework for GDPR- and AI Act-Aligned Decision Transparency. MDPI J. Cybersecurity and Privacy.
  11. Hasan, S.M. et al. (2025). Carbon emission quantification of ML: A review. IEEE Trans. Sustainable Computing.
  12. Juliussen, B.A. (2025). The Right to an Explanation Under the GDPR and the AI Act. Springer, Int. Conf. Multimedia Modeling.
  13. Kim, M.K. & Yoo, T.A. (2026). Toward Sustainable AI: A Scoping Review of Carbon Footprint. IEEE Access.
  14. Linders, J. & Tomczak, J.M. (2025). Knowledge graph-extended RAG for QA. Applied Intelligence.
  15. Liu, B. et al. (2025). SymAgent: A neural-symbolic self-learning agent framework for complex reasoning over KGs. TheWebConf 2025.
  16. Ma, C. et al. (2025). Large language models meet knowledge graphs for question answering. EMNLP 2025.
  17. Niu, G. et al. (2026). A Comprehensive Survey of Knowledge Graph Reasoning. IEEE Trans. Big Data.
  18. Perevalov, A. et al. (2024). Multilingual question answering systems for knowledge graphs — a survey. Semantic Web.
  19. Salman, M. et al. (2024). Doc-KG: Unstructured documents to KG construction. Expert Systems.
  20. Schwartz, R. et al. (2020). Green AI. Communications of the ACM, 63(12), 54-63.
  21. Sidhu, M.K. et al. (2025). Neurosymbolic AI for Enhanced Reasoning in Small Language Models. IEEE.
  22. Sovrano, F. et al. (2025). Legal XAI: Systematic Review and Interdisciplinary Mapping of XAI and EU Law. ResearchGate.
  23. Tunzina, T. et al. (2026). GreenAI: Environmental Efficiency in LLM-Generated Code. IEEE Access.
  24. Wagner, M. et al. (2025). A Critical Look at the EU AI Act's Requirements — Who Must Explain What? IEEE Int. Requirements Eng. Conf.
  25. Zavarella, V. (2026). Methods for KG Construction from Text Collections. arXiv:2603.25862.