基于离群主题词跨学科组合的学术创新机会发现研究  

Research on the Discovery of Academic Innovation Opportunity Based on the Interdisciplinary Combination of Outlier Topic Words

作　　者：李秀霞 邵作运[2] Li Xiuxia

机构地区：[1]曲阜师范大学传媒学院,山东日照276826 [2]曲阜师范大学图书馆,山东日照276826

出　　处：《情报理论与实践》2023年第12期122-130,共9页Information Studies:Theory & Application

基　　金：国家社会科学基金一般项目“跨学科知识元迁移组合与学术创新机会发现研究”的成果,项目编号:22BTQ061。

摘　　要：[目的/意义]学科领域的离群主题词可为创新机会发现提供新颖、稀缺的信息,离群主题词跨学科组合能催生新的知识,产生突破性学术创新机会。[方法/过程]以情报学和政治学为例,利用LDA提取不同学科文献的主题,以概率分布低的主题词为数据对象,利用Word2Vec和PCA技术将题名和摘要中包含文本语义的主题词表示为低维稠密向量,根据主题词在二维空间的分布发现学科内的离群主题词;利用余弦相似度计算不同学科离群主题词之间的语义相似度,将相似度高的不同学科的离群主题词组合视为具有创新潜能的组合。根据设计的需求度指标进一步筛选离群主题词组合,最终确定未来具有研究潜力的学术创新机会。[结果/结论]将主题提取与语义分析相结合,充分考虑了离群主题词的价值和语义环境;将离群主题词跨学科组合的语义相似度与需求度结合,能够兼顾学术创新的新颖性和有用性特征。研究表明,此研究方法能够有效发现学术创新机会,为科研指导、知识服务提供可靠参考。[Purpose/significance] Outlier topic words play a crucial role in providing fresh insights for the exploration of innovative opportunities.By combining outlier topic words from different disciplines,the potential for generating groundbreaking academic innovations can be unlocked.[Method/process] In this study,Information Science and Political Science are taken as examples.The first step involves applying LDA to extract topics from literature across various fields.Utilizing Word2Vec,topic words are then transformed into low-dimensional dense vectors that capture their semantic meanings in titles and abstracts.Subsequently,outlier topic words within each discipline are identified based on their distribution in a two-dimensional space.The semantic similarity between outlier topic words from different disciplines is calculated using cosine similarity,aiming to identify combinations of outlier topic words that exhibit high similarity.To gauge the demand for these combinations,an index is devised.[Result/conclusion] By combining topic extraction with semantic analysis,the value and semantic context of outlier topic terms are fully considered.Integrating the semantic similarity of interdisciplinary combinations of outlier topic terms with their demand level allows for a balance between novelty and utility in academic innovation.Research indicates that the proposed method can effectively discover academic innovation opportunities and provide reliable references for research guidance and knowledge services.

关 键 词：离群主题词 语义相似度 需求度 创新机会发现 

分 类 号：G353.1[文化科学—情报学]