Document-level Event Extraction (DEE) aims to extract event records from given documents. Previous works typically process documents by constructing complex feature fusion matrices or graph structures, aiming to predict complete event records via entity selection or role filling. However, this holistic structure extraction paradigm often suffers from cascaded error propagation due to rigid entity dependency, multi-role structural omissions limited by selection mechanisms, and consequent scalability bottlenecks. In this paper, we propose a Deconstruction-Reconstruction Framework for Document-Level Event Extraction (DrDEE). This framework deconstructs the holistic event record extraction task into atomic Token-Event-Role (TER) triplet verification, and subsequently reconstructs event records from valid tokens. It realizes a paradigm shift in DEE from Many-to-Many Selection to One-to-One Verification, significantly simplifying the task formulation. Experimental results show that the proposed method outperforms state-of-the-art methods, confirming its effectiveness. Our code is available at https://anonymous.4open.science/r/DrDEE.

Recent years have witnessed remarkable advances in Large Language Models (LLMs). However, in the task of social relation recognition, Large Language Models (LLMs) encounter significant challenges due to their reliance on sequential training data, which inherently restricts their capacity to effectively model complex graph-structured relationships. To address this limitation, we propose a novel low-coupling method synergizing multimodal temporal Knowledge Graphs and Large Language Models (mtKG-LLM) for social relation reasoning. Specifically, we extract multimodal information from the videos and model the social networks as spatial Knowledge Graphs (KGs) for each scene. Temporal KGs are constructed based on spatial KGs and updated along the timeline for long-term reasoning. Subsequently, we retrieve multi-scale information from the graph-structured knowledge for LLMs to recognize the underlying social relation. Extensive experiments demonstrate that our method has achieved state-of-the-art performance in social relation recognition. Furthermore, our framework exhibits effectiveness in bridging the gap between KGs and LLMs. Our code will be released after acceptance.