AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

TGCA-LLM: Time-Aware Graph-Text Contrastive Alignment for Enhancing LLMs in Temporal Knowledge Graph Completion

Zexuan Wan, Bo Wang, Kuofei Fang, Bin Wu

AAAI 2026 · Large language models › Retrieval and knowledge · approach: Knowledge graph reasoning with LLMs · 5 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add structural graph tokens (static graph) to text-only input (ablation)TGCA-LLM · ICEWS14MRR: 0.347 -> 0.442Improved
use time-aware graph-text aligned embeddings instead of static graph (ablation)TGCA-LLM · ICEWS14MRR: 0.442 -> 0.549Improved
use TGCA contrastively aligned embeddings instead of RE-GCN ConvTransE (ablation)TGCA-LLM · ICEWS14MRR: 0.486 -> 0.549Improved
add stage-1 temporally-constrained graph matching tuning (ablation)TGCA-LLM · ICEWS14MRR: 0.501 -> 0.549Improved
shuffle graph token order in stage 1 (ablation)TGCA-LLM · ICEWS14MRR: 0.493 -> 0.549Improved
add stage-1 tuning (ICEWS05-15) (ablation)TGCA-LLM · ICEWS05-15MRR: 0.611 -> 0.650Improved
add stage-1 tuning (ICEWS18) (ablation)TGCA-LLM · ICEWS18MRR: 0.427 -> 0.476Improved
increase historical text facts from 10 to beyond 30 (ablation)TGCA-LLM · ICEWS14/ICEWS05-15/ICEWS18MRR: improves up to 30 then plateaus or slightly declinesMixed
increase historical graph length beyond 5 (ablation)TGCA-LLM · ICEWS14/ICEWS05-15/ICEWS18MRR: best at lengths 5 and 7No clear effect
use TGCA-LLM (Llama-2-7b) instead of DeepSeek-V3 zero-shotDeepSeek-V3 · ICEWS14MRR: 0.454 -> 0.549Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy