Natural Language Explainability for Temporal KGs

chevron-icon
Back
Top
chevron-icon
project-presentation-img
user-profile-img
ivan reznikov
Project Owner

Natural Language Explainability for Temporal KGs

Status

  • Overall Status

    ⏳ Contract Pending

  • Funding Transfered

    $0 USD

  • Max Funding Amount

    $87,500 USD

Funding Schedule

View Milestones
Milestone Release 1
$25,000 USD Pending TBD
Milestone Release 2
$25,000 USD Pending TBD
Milestone Release 3
$25,000 USD Pending TBD
Milestone Release 4
$12,500 USD Pending TBD

Project AI Services

No Service Available

Overview

Our team is dedicated to addressing a fundamental challenge in AGI development: making temporal knowledge graphs interpretable through natural language while ensuring reasoning consistency through Truth Maintenance Systems (TMS). We're proposing a comprehensive framework for explaining complex temporal relationships and reasoning paths within knowledge graphs in human-understandable language, with particular attention to browsable, observable logic workflows compatible with OpenCog Hyperon, including MeTTa and MORK. What excites us most is the potential to bridge the gap between machine temporal reasoning and human understanding through transparent, traceable inference processes.

RFP Guidelines

Advanced knowledge graph tooling for AGI systems

Complete & Awarded
  • Type SingularityNET RFP
  • Total RFP Funding $350,000 USD
  • Proposals 39
  • Awarded Projects 5
author-img
SingularityNET
Apr. 16, 2025

This RFP seeks the development of advanced tools and techniques for interfacing with, refining, and evaluating knowledge graphs that support reasoning in AGI systems. Projects may target any part of the graph lifecycle — from extraction to refinement to benchmarking — and should optionally support symbolic reasoning within the OpenCog Hyperon framework, including compatibility with the MeTTa language and MORK knowledge graph. Bids are expected to range from $10,000 - $200,000.

Proposal Description

Our Team

We're a passionate group of experts who have worked together on complex graph problems for years:

  • 3 dedicated researchers with backgrounds in knowledge representation

  • 2 backend and 1 data engineers with extensive graph database experience

  • 2 data scientists who specialize in graph analytics

  • 1 graph scientist focused on optimization algorithms

  • 1 graph network expert

Company Name (if applicable)

SciRenovation Labs

Project details

If you ask ChatGPT, Claude or other no-internet systems who is the Pope - you'll get an outdated answer. Current graph systems can be built quite fast, but there might an issue with getting time-relevant results and "time-travel". This is especially important in fields of news, science, etc - some information can get updated quite fast and reasoning models relying on graph systems might not get actual results.

Anyone who's worked with temporal knowledge graphs knows the challenge of interpreting and explaining time-dependent relationships and reasoning paths. In complex logic systems like MeTTa, inference behavior can become opaque—particularly when forward/backward chaining, multi-rule propagation, or context-sensitive reasoning is involved. We've experienced these challenges firsthand, and they've motivated us to develop this proposal. Our vision is to create an open, extensible framework that transforms complex temporal graph structures into natural language explanations while maintaining logical consistency through TMS principles—making AGI reasoning more transparent, traceable, and trustworthy.

We're focusing on three key areas that we believe will make the most significant impact:

1. Natural Language Generation for Temporal Graph Structures with Dependency Tracking

We're determined to overcome the limitations of traditional graph visualization and technical query languages. Our approach leverages advanced natural language generation techniques to express temporal relationships in intuitive human language while incorporating TMS dependency tracking.

We're particularly excited about:

  • Developing specialized templates and linguistic patterns that accurately convey temporal intervals, sequences, and causality

  • Implementing contextual explanation strategies that adapt to different user knowledge levels and domains

  • Creating narrative frameworks that can explain complex temporal reasoning paths as coherent stories

  • Integrating TMS-based justification structures that explain not just what the system believes, but why it believes it

We're also exploring transformer-based models fine-tuned specifically for temporal relationship verbalization. Our preliminary tests show promising results in generating explanations that distinguish between different temporal representations (timestamps, intervals, durations) and accurately convey their significance.

For complex temporal reasoning chains, we're adapting sequence-to-sequence models to transform graph traversal paths into step-by-step explanations—early results suggest this approach significantly improves user comprehension of machine reasoning compared to traditional graph visualizations.

Additionally, we're investigating techniques to generate counterfactual explanations that help users understand how temporal changes would affect outcomes ("If event X had occurred 3 months earlier, Y would likely have happened instead of Z"). These counterfactual explanations will be supported by TMS contextual reasoning capabilities that can maintain multiple belief sets simultaneously.

2. Temporal Query Translation and Interpretation with Consistency Maintenance

We're frustrated by the gap between natural human questions about time and the structured query languages needed to extract answers from temporal graphs. Our goal is to create a bidirectional translation system that converts natural language temporal queries into precise graph queries and translates results back into explanatory language while ensuring logical consistency.

We're exploring:

  • Advanced temporal expression recognition that captures nuanced time references in natural language

  • Query intent classification to distinguish between point-in-time queries, interval queries, sequence queries, and causal queries

  • Natural language interfaces that allow non-technical users to explore temporal knowledge through conversation

  • **Browsable, observable logic workflows that explicitly answer crucial questions like:

    • What rules were triggered?

    • Why did a given conclusion hold—or not hold?

    • Where did a contradiction arise?**

One of our most ambitious goals is to develop explanation frameworks for temporal inference—using narrative structures to explain how the system reasoned from temporal facts to derive new knowledge. We're developing techniques to verbalize temporal logic operations and probabilistic reasoning in ways that maintain accuracy while remaining intuitive to users.

Through TMS integration, our system will provide first-class treatment of queries and their results, enabling developers and researchers to:

  • Assert rules that generate other assertions

  • Assert rules that issue queries

  • Observe query results as first-class knowledge base elements

3. Benchmarks for Temporal Graph Explainability with TMS Evaluation

We're concerned about the lack of standardized evaluation methods for assessing the quality of temporal explanations. Our team is designing a comprehensive benchmark suite that evaluates aspects that truly matter: clarity, accuracy, consistency maintenance, and utility of explanations for temporal knowledge graphs.

Our benchmarks will measure:

  • Explanation fidelity across different temporal relationship types (point-in-time, intervals, sequences)

  • Comprehension metrics based on user understanding after reading explanations

  • Explanation efficiency at different levels of temporal reasoning complexity

  • Cross-domain applicability from historical to financial to scientific contexts

  • Robustness to different temporal granularities (seconds to decades)

  • Consistency maintenance under revision of temporal facts

  • Quality of contradiction detection and resolution

Beyond simple factual explanations, we're focusing on complex temporal reasoning tasks like explaining causal chains, temporal anomalies, and counterfactual scenarios—the kinds of explanations that distinguish truly intelligent systems from mere information retrievers.

We're particularly inspired by the need for AI systems to explain their temporal reasoning in fields like medicine, finance, and historical analysis, where understanding the sequence and timing of events is crucial for trust and decision-making.

Our project naturally aligns with the Hyperon framework by providing essential explainability tools that will enhance temporal reasoning capabilities in MeTTa. We're committed to exploring compatibility with the MORK system to enable real-time generation of natural language explanations for temporal inference paths. The integration of TMS principles will significantly improve interpreter correctness, accelerate theorem prover debugging, and enable interactive, human-understandable AI development.

 

  1. Wu, H., Cheng, J., Huang, S., Ke, Y., Lu, Y., & Xu, Y. (2014). Path problems in temporal graphs. Proceedings of the VLDB Endowment, 7(9), 721-732.

  2. Paranjape, A., Benson, A. R., & Leskovec, J. (2017). Motifs in temporal networks. Proceedings of the Tenth ACM International Conference on Web Search and Data Mining.

  3. Rozemberczki, B., Scherer, P., Kiss, O., Sarkar, R., & Ferenci, T. (2022). PyTorch geometric temporal: Spatiotemporal signal processing with neural machine learning models. arXiv preprint.

  4. Casteigts, A., Flocchini, P., Quattrociocchi, W., & Santoro, N. (2012). Time-varying graphs and dynamic networks. International Journal of Parallel, Emergent and Distributed Systems, 27(5), 387-408.

  5. Rossi, E., Chamberlain, B., Frasca, F., Eynard, D., Monti, F., & Bronstein, M. (2020). Temporal graph networks for deep learning on dynamic graphs. arXiv preprint.

  6. Huang, Haixing, Jinghe Song, Xuelian Lin, Shuai Ma, and Jinpeng Huai. "Tgraph: A temporal graph data management system." In Proceedings of the 25th ACM International on Conference on Information and Knowledge Management, pp. 2469-2472. 2016.

  7. Rahman, Aniq Ur, and Justin P. Coon. "A Primer on Temporal Graph Learning." arXiv preprint arXiv:2401.03988 (2024).

  8. Sahili, Zahraa Al, and Mariette Awad. "Spatio-temporal graph neural networks: A survey." arXiv preprint arXiv:2301.10569 (2023).

  9. Rossetti, G., & Cazabet, R. (2018). Community discovery in dynamic networks: A survey. ACM Computing Surveys, 51(2), 1-37.

  10. Exploring Temporal and Spatial Graph Databases

  11. Michail, Othon. "An introduction to temporal graphs: An algorithmic perspective." Internet Mathematics 12, no. 4 (2016): 239-280.https://www.internetmathematicsjournal.com/api/v1/articles/1606-an-introduction-to-temporal-graphs-an-algorithmic-perspective.pdf

  12. ATGL: Awesome Temporal Graph Learning - GitHub, accessed May 2, 2025, https://github.com/MGitHubL/Awesome-Temporal-Graph-Learning

  13. Casteigts, Arnaud, Kitty Meeks, George B. Mertzios, and Rolf Niedermeier. "Temporal graphs: Structure, algorithms, applications (dagstuhl seminar 21171)." Dagstuhl Reports 11, no. 3 (2021): 16-46.

  14. Michail, Othon. "An introduction to temporal graphs: An algorithmic perspective." Internet Mathematics 12, no. 4 (2016): 239-280.

Open Source Licensing

MIT - Massachusetts Institute of Technology License

Background & Experience

This proposal is one of five parts of a unified toolkit to be developed in parallel across a team of 14-16 developers and scientists. We have already a case of successful grant completion for DeepFunding.

Our team includes current employees from Yandex and Intel, former Linkedin, 3 university lecturers, and 7 PhDs. We're proud of our 10+ presidential awards, several patents and 30+ publications, including multiple technical books from a well known publishing house.

If we are awarded funding for 4 out of 5 proposals, we are committed to developing the 5th one at no additional cost.

We believe it’s better to have one toolkit than a kit of tools, that need to be duct-taped together.

We aim to deliver a robust, extensible, modular, production-ready ecosystem that can evolve with future RFPs, enabling seamless adoption, innovation, and collaboration. This approach will maximize the utility of knowledge graph fundamentals and pull in other innovative features and technologies from DeepFunding.

Proposal Video

Not Avaliable Yet

Check back later during the Feedback & Selection period for the RFP that is proposal is applied to.

Reviews & Rating

New reviews and ratings are disabled for Awarded Projects

Overall Community

0

from 0 reviews
  • 5
    0
  • 4
    0
  • 3
    0
  • 2
    0
  • 1
    0

Feasibility

0

from 0 reviews

Viability

0

from 0 reviews

Desirabilty

0

from 0 reviews

Usefulness

0

from 0 reviews

Sort by

0 ratings

    No Reviews Avaliable

    Check back later by refreshing the page.

  • Total Milestones

    4

  • Total Budget

    $87,500 USD

  • Last Updated

    28 Aug 2025

Milestone 1 - Temporal Relationship Verbalization Framework

Status
😐 Not Started
Description

Develop a specialized natural language generation framework that transforms complex temporal graph structures into human-readable explanations. This framework will include configurable templates for various temporal relationships (intervals, sequences, causality) and adapt explanations based on user expertise level. Implementation will include TMS dependency tracking to explain not just what the system believes but why it believes it.

Deliverables

- Simple library for converting temporal graph structures to natural language explanations - Suite of specialized templates covering at least 5 common temporal relationship types (point-in-time, intervals, sequences, overlaps, causality) - Integration with TMS dependency tracking enabling explanation of belief justifications

Budget

$25,000 USD

Success Criterion

The framework successfully generates natural language explanations for temporal graphs that non-technical users can understand. Explanations include proper justification structures and can adapt to different technical levels.

Link URL

Milestone 2 - Temporal Query Translation System

Status
😐 Not Started
Description

Create a bidirectional translation system that converts natural language temporal queries into precise graph queries and translates results back into explanatory language. This system will recognize time references in natural language and classify query intents (point-in-time, interval, sequence, causal). The system will maintain logical consistency through TMS principles.

Deliverables

- Natural language interface for temporal queries with support for at least several types of temporal expressions - Query intent classification system with >90% accuracy for common temporal query types - Bidirectional translation mechanism with comprehensive test suite demonstrating accuracy

Budget

$25,000 USD

Success Criterion

Users without technical knowledge can successfully retrieve information from temporal knowledge graphs using natural language queries, with the system achieving >85% translation accuracy on our benchmark test set. The system maintains logical consistency when handling contradictory information.

Link URL

Milestone 3 - Temporal Explanation Benchmark Suite

Status
😐 Not Started
Description

Develop a comprehensive benchmark suite for evaluating temporal graph explanations across multiple dimensions. The benchmark will assess explanation quality, accuracy, and utility across different temporal relationship types and domains. Special attention will be given to evaluating how well explanations convey the system's reasoning process.

Deliverables

- Benchmark datasets covering at least 4 domains (medical, financial, historical, scientific) - Evaluation metrics for explanation fidelity, comprehension, and efficiency - Documentation and tools for running benchmarks and analyzing results

Budget

$25,000 USD

Success Criterion

The benchmark successfully differentiates between explanation techniques based on meaningful metrics and provides actionable insights for improvement. The benchmark is adopted by at least two other research groups working on temporal reasoning systems.

Link URL

Milestone 4 - MeTTa/MORK Integration Framework

Status
😐 Not Started
Description

Develop integration components that connect our temporal explanation system with the MeTTa language and MORK framework. This will enable real-time generation of natural language explanations for temporal inference paths in MeTTa programs. The integration will support truth maintenance capabilities, enhancing interpreter correctness and enabling interactive AI development.

Deliverables

- MeTTa and MORK extension package for accessing temporal explanation capabilities - Example programs demonstrating temporal reasoning explanation in MeTTa

Budget

$12,500 USD

Success Criterion

Developers using MeTTa can incorporate natural language explanations of temporal reasoning into their applications with minimal code changes.

Link URL

Join the Discussion (0)

Expert Ratings

Reviews & Ratings

New reviews and ratings are disabled for Awarded Projects

    No Reviews Avaliable

    Check back later by refreshing the page.

Welcome to our website!

Nice to meet you! If you have any question about our services, feel free to contact us.