大语言模型(large language model,LLM)技术热潮对数据质量的要求提升到了一个新的高度.在现实场景中,数据通常来源不同且高度相关.但由于数据隐私安全问题,跨域异质数据往往不允许集中共享,难以被LLM高效利用.鉴于此,提出了一种LLM和知识图谱(knowledge graph,KG)协同的跨域异质数据查询框架,在LLM+KG的范式下给出跨域异质数据查询的一个治理方案.为确保LLM能够适应多场景中的跨域异质数据,首先采用适配器对跨域异质数据进行融合,并构建相应的知识图谱.为提高查询效率,引入线性知识图,并提出同源知识图抽取算法HKGE来实现知识图谱的重构,可显著提高查询性能,确保跨域异质数据治理的高效性.进而,为保证多域数据查询的高可信度,提出可信候选子图匹配算法Trust HKGM,用于检验跨域同源数据的置信度计算和可信候选子图匹配,剔除低质量节点.最后,提出基于线性知识图提示的多域数据查询算法MKLGP,实现LLM+KG范式下的高效可信跨域查询.该方法在多个真实数据集上进行了广泛实验,验证了所提方法的有效性和高效性.
The deformation behavior and mechanism of Ti2AlNb-based alloy were experimentally investigated at elevated temperatures. Firstly, the stress?strain relationships at different temperatures and strain rates were investigated via uniaxial tensile testing. Then, formability data, as determined by examining the deep drawing and bending abilities, were obtained through limiting draw ratio (LDR) and bending tests. Finally, metallographic experiments and fracture morphology investigations were conducted to examine the thermal deformation mechanism of the alloy. The results showed that as the temperature increased, the total elongation increased from 13.58% to 97.82% and the yield strength decreased from 788 to 80 MPa over the temperature range from 750 to 950 °C at a strain rate 0.001 s?1. When the temperature reached 950 °C, the strain rate was found to have a great influence on the deformation properties. The plastic formability of the sheet metal was significantly improved and a microstructuraltransformation of O toB2 andα2 occurred in this temperature region, revealing the deformation mechanism of its plasticity.
A rational parametric planar cubic H spline curve is defined by a set of control vertices in a plane and percentage factors of line segments between every two control vertices. Movement of any control vertex affects three curve segments. This paper is the succession and development of reference of Tang Yuehong. We analyze the geometric features like cusps and inflection points in the curve and calculate the cusps and inflection points, then give a necessary and sufficient condition to the inflection points in the curve when it is non degenerative, and finally show that the curves have no cusps in the interval (0,1). In many applications, it is desirable to analyze the parametric curves for undesirable features like cusps and inflection points