Preliminary Conference Agenda

Aiming at the limitation of single associations in knowledge unit citation network, this paper enhances the diversity of semantic association types between network nodes through citation function, thereby proposing a citation function-aware knowledge unit citation network and conducting multi-dimensional analysis of domain knowledge. Firstly, the academic text is parsed to extract information such as knowledge units, citation links, citation context and citation objects, and identify their citation functions. On this basis, the complex network method is used to construct a citation function aware-knowledge unit citation network, and the multidimensional analysis of domain knowledge is carried out from the analysis of citation network structure, multidimensional citation relationship analysis of knowledge units, and citation function-aware knowledge community analysis. The proceedings of ACL (Association for Computational Linguistics) are used as an example for empirical research. The results verify the effectiveness of the proposed method; discover the usage, extension and comparison patterns among domain knowledge; and enrich the semantic information of knowledge communities. This study extends the research method of knowledge unit citation network, deeply revealing the semantic relationships between discipline knowledge, and providing a new path for analyzing discipline knowledge structure.

在学科界限日益模糊、学科壁垒逐渐被打破的背景之下，运用不同学科方法、理论来解决更加综合、复杂的学术问题成为了越来越多学者的研究模式。为识别此类跨学科知识交流活动中不同学者扮演角色的重要程度、不同学者间进行跨学科知识交流行为的趋势大小和跨学科知识交流网络的演化规律，本文构建起生态学与物理学交叉视角下的跨学科知识交流动力模型。首先将物理学中势能的相关概念和生态学中生态位的相关概念移植至跨学科知识交流领域，并论证其可行性与必要性；其次以知识主体的知识存量为基础，融合知识主体所处网络结构特征、知识主体的知识生态位宽度计算知识主体的跨学科知识交流势能；然后以知识主体间的跨学科知识交流势能差为基础，结合主体间的知识生态位重叠度，量化知识主体间的跨学科知识交流趋势，构建起知识主体视角的跨学科知识交流动力模型；接着以所有的知识主体所处的知识交流网络为研究对象，描述整个学术生态系统中跨学科知识交流行为的活跃程度和网络的演化进程，构建起跨学科知识交流网络视角下的网络演化动力模型；最后，以health informatics领域下学者的跨学科知识交流网络为实证数据，验证了知识主体视角的跨学科知识交流动力模型，分析了网络视角的跨学科知识交流演化进程和规律。

In the constantly evolving landscape of information and innovation, the ability to accurately predict emerging topics is vital across various fields. This study regards weak signals as the early stages of emerging topics and identifies these nascent topics through an innovative three-dimensional analysis framework for weak signals. Initially, we construct a keyword citation network, identifying novel signal collections based on changes in network structure. We then conduct a temporal analysis of the visibility and diffusion of signals with time-weighted attributes to identify weak signals, consider their social influence, and quantify the influence of weak signals from the public perception perspective using alternative metrics. Finally, an evaluation framework for emerging topics based on weak signals is built across three dimensions: visibility, diffusion, and influence. We applied the method to gene editing and underwent multi-angle method verification and comparative analysis. The results demonstrate that this approach not only enhances the capability to identify emerging topics at an early stage and grasp their development trends but also enables the identification of emerging topics with significant social impact, thereby offering valuable insights for strategic decision-making, innovation management, and future foresight.

【目的/意义】针对研究人员的研究兴趣呈阶段性变化，提出了一种基于注意传播机制和时序注意机制的科技论文推荐模型，以提升论文推荐的效果。【方法/过程】结合用户行为和论文属性特征构建协同知识图谱，利用TransR方法将图谱中的节点映射到向量空间中；基于注意传播机制从协同知识图谱中学习论文的知识增强特征表示，基于时序注意机制从阅读行为中学习研究人员的动态偏好表示，计算研究人员与论文的匹配分数得到推荐列表。【结果/结论】以“区块链实验室”数据进行大量的实验分析，结果表明所提出模型的推荐表现最佳，可以说明基于注意传播机制和时序注意机制的表征方法能够更为精准地捕获用户的动态偏好，且适用于论文推荐系统中。【创新/局限】通过融合用户行为和论文知识图谱，利用基于注意传播机制和时序注意机制的表征方法实现了科技论文动态推荐。虽然通过融合辅助信息的方式挖掘用户的兴趣演变，但辅助信息的类型数量有待提升。

汉字是我国的非物质文化遗产，其字形在数千年的传承中几经变化。要探究字形演变对汉字系统宏观规律的影响，可从汉字的字形复杂度特征入手，对其在汉字演变过程之中分布规律的变化进行探索。本文提出像素法——一种测量字形复杂度的新方法，由此测量、比较了汉字字形复杂度及其动链在不同字库与文本中的统计特征。研究发现，字形复杂度在汉字系统、汉语文本中的分布规律，以及字形复杂度动链的频次分布、长度分布规律，均没有因字形演变而改变，汉字字形复杂度统计特征的稳定性是其构形系统与汉语词长共同演化的结果。

Chinese characters are intangible cultural heritage in China, and their forms have undergone several changes over thousands of years of inheritance. To explore the impact of the evolution of Chinese characters on their usage patterns, we can start with the glyph complexity characteristics of Chinese characters and explore the changes in their distribution patterns during the evolution process. This article proposes the pixel method - a new method for measuring the complexity of glyphs, which measures and compares the statistical characteristics of Chinese character glyph complexity and its motifs in different word libraries and texts. Research has found that the distribution pattern of glyph complexity in Chinese character systems and texts, as well as the frequency and length distribution patterns of glyph complexity motifs, have not changed due to the evolution of glyphs. The stability of the statistical characteristics of glyph complexity in Chinese characters is the result of the co-evolution of their structural system and Chinese word length.