![[2408.05861] Temporal Knowledge-Graph Memory in a Partially Observable Environment](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2408.05861] Temporal Knowledge-Graph Memory in a Partially Observable Environment
Summary
This paper introduces a novel temporal knowledge-graph memory system for agents operating in partially observable environments, enhancing their ability to integrate observations over time.
Why It Matters
The research addresses a significant gap in AI by providing a framework for agents to utilize persistent memory in dynamic environments, which is crucial for tasks requiring long-term reasoning and adaptability. This work could lead to advancements in AI applications that rely on knowledge graphs and temporal reasoning.
Key Takeaways
- Introduces a configurable Room Environment v3 for testing agents.
- Demonstrates that temporal qualifiers improve agent performance in knowledge retention.
- Symbolic TKG agents outperform neural baselines in question-answer accuracy.
- Provides reproducible research tools, including environment and agent implementations.
- Highlights the importance of integrating memory structures in AI for better adaptability.
Computer Science > Artificial Intelligence arXiv:2408.05861 (cs) [Submitted on 11 Aug 2024 (v1), last revised 25 Feb 2026 (this version, v4)] Title:Temporal Knowledge-Graph Memory in a Partially Observable Environment Authors:Taewoon Kim, Vincent François-Lavet, Michael Cochez View a PDF of the paper titled Temporal Knowledge-Graph Memory in a Partially Observable Environment, by Taewoon Kim and 2 other authors View PDF HTML (experimental) Abstract:Agents in partially observable environments require persistent memory to integrate observations over time. While KGs (knowledge graphs) provide a natural representation for such evolving state, existing benchmarks rarely expose agents to environments where both the world dynamics and the agent's memory are explicitly graph-shaped. We introduce the Room Environment v3, a configurable environment whose hidden state is an RDF KG and whose observations are RDF triples. The agent may extend these observations into a temporal KG when storing them in long-term memory. The environment is easily adjustable in terms of grid size, number of rooms, inner walls, and moving objects. We define a lightweight temporal KG memory for agents, based on RDF-star-style qualifiers (time_added, last_accessed, num_recalled), and evaluate several symbolic baselines that maintain and query this memory under different capacity constraints. Two neural sequence models (LSTM and Transformer) serve as contrasting baselines without explicit KG structure. Agents ...