用显式数据库构建可解释的智能系统，解决大模型幻觉与不可修改问题。
An Explicit World Model Based on Data-First Ontology: DaoQL Multimodal Storage Validation and Counterfactual Reasoning Evaluation
- 将知识存入显式多模态库DaoQL，让大模型专注推理与语言
- 在5个领域中实现94%的反事实推理可分解性，比纯大模型高49个百分点
- 适合医疗、金融等对准确性要求高的场景，也便于知识更新与验证
大型语言模型将世界模型隐式编码于神经权重中，在医学、金融等高精度领域存在幻觉、知识冻结、解释性差和难修改四大结构性风险。本文提出数据优先本体论：将大模型视为推理与语言引擎，确定性知识移入显式多模态数据库DaoQL。形式化显式世界模型，证明在规则独立、确定性评估和固定冲突解决下，显式模型提供可组合反事实分解的充分条件；而隐式模型缺乏原子读取/增量语义，无法提供类似架构保障。系统实现聚焦于DaoQL的验证存储层与显式评估路径，集成图、列式、向量与全文引擎于单进程。嵌入式同机部署下，图广度优先搜索达1.20毫秒，HNSW查询83.1微秒，混合查询105.8微秒；在LDBC SNB SF1与ANN-Benchmarks上，34/34查询覆盖，交互类查询多在亚毫秒至毫秒级，但整体仅1.8 QPS（因长尾商业智能/交互查询）；经桥接边保护修复后，ANN-Benchmarks Recall@10 ≥99%，达千级QPS。五域反事实实验（n=1250）显示，DaoQL+GPT-4o实现94%可组合反事实分解性，较GPT-4o单独提升49个百分点。论文明确区分可证明结构、初步实证证据与架构路线图主张。
原文摘要 · Abstract (English)
Large language models encode world models implicitly in neural weights, which exposes four structural risks in high-precision domains such as medicine and finance: hallucination, frozen knowledge, poor explainability, and poor modifiability. This paper proposes data-first ontology: LLMs are treated as reasoning and language engines, while deterministic knowledge is moved into an explicit multimodal database, DaoQL. We formalize an explicit world model and show that, under rule independence, deterministic evaluation, and fixed conflict resolution, explicit models provide a sufficient condition for composable counterfactual decomposability; implicit models lack atomic read/delta semantics and therefore provide no comparable architectural guarantee. The implemented system focuses on DaoQL's verified storage layer and explicit Eval path, integrating graph, column, vector, and full-text engines within one process. KVCache graph nodes, expert hot updates, and the DaoQL-Agent runtime remain future work. On an embedded same-machine setup, DaoQL reports graph BFS at 1.20 ms, HNSW at 83.1 us, and a Fluent hybrid query at 105.8 us; these results indicate engineering potential but must be interpreted with deployment-shape differences from client-server systems. Exploratory measurements on LDBC SNB SF1 and ANN-Benchmarks further show 34/34 query coverage with interactive-class queries mostly in the sub-millisecond to millisecond range, but only 1.8 QPS overall due to long-tail BI/IC queries; ANN-Benchmarks reaches Recall@10 >= 99% at thousand-level QPS after a bridge-edge protection fix. In a five-domain counterfactual experiment (n = 1250), DaoQL+GPT-4o achieves 94% composable counterfactual decomposability, 49 percentage points above GPT-4o alone. The paper explicitly separates provable structure, preliminary empirical evidence, and architectural roadmap claims.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。