[
    {
        "id": "osp-15468",
        "type": "article-journal",
        "title": "Socio-Foundation: A Model for Generalizable Individual Behavior Simulation via Hierarchical Capability Distillation",
        "author": [
            {
                "family": "Wang",
                "given": "Liang"
            },
            {
                "family": "Xie",
                "given": "Wenxuan"
            },
            {
                "family": "Mou",
                "given": "Xinyi"
            },
            {
                "family": "Luo",
                "given": "Yixin"
            },
            {
                "family": "Wei",
                "given": "Zhongyu"
            }
        ],
        "URL": "https://omanscience.com/en/articles/socio-foundation-a-model-for-generalizable-individual-behavior-simulation-via-hierarchical-capability-distillation",
        "language": "en",
        "issued": {
            "date-parts": [
                [
                    2026
                ]
            ]
        },
        "abstract": "Simulating individual behavior requires large language models (LLMs) to preserve persona traits while adapting to dynamic social contexts. However, general-purpose LLMs often flatten distinct personas, while task-specific tuning suffers from fragmentation and generalization. To overcome these challenges, we organize individual simulation into the \\textbf{FONTS Taxonomy}, comprising five complementary capability dimensions: \\emph{persona fidelity} (\\textbf{F}), \\emph{outcome realization} (\\textbf{O}), \\emph{behavioral naturalness} (\\textbf{N}), \\emph{trajectory coherence} (\\textbf{T}), and \\emph{social grounding} (\\textbf{S}). Grounded in this taxonomy, we curate a standardized training corpus library of approximately 10 million instances across 14 representative datasets and present \\textbf{Socio-Foundation}. Socio-Foundation decouples specialization from integration via a three-stage pipeline: learning task experts via DAPO, consolidating them into capability experts via off-policy distillation, and unifying them via multi-teacher on-policy distillation (MOPD). We also establish \\textbf{IndiEval}, consolidating 29 metrics across the FONTS dimensions. Experiments show that Socio-Foundation outperforms its \\textit{Qwen3-8B} base by 11.0 points and approaches frontier models such as \\textit{GLM-5.2}, with ablations and out-of-distribution evaluations further demonstrating the effectiveness and generalization of our model."
    }
]