Doctoral Dissertation Research: The Semantics of Scientific Success: A Citation Analysis of the Semantic Structure and Bases of Success in Science

以下是语义引文分析项目的建议--一种科学引文研究的方法，通过结合心理语言学和新达尔文主义理论，有望对科学知识的组织和传播提供新的见解。该项目的目标是建立一种新的方法，利用从出版的主要科学引文索引摘要中提取的文本数据来模拟文化差异和转变。使用一种名为潜在语义分析的计算程序，科学文章摘要的含义将被定量表示，允许使用多维缩放来分析它们与索引中其他文章的语义相似性并以图形方式说明。这样做将允许根据潜在的语义相似性，将科学领域作为科学中不同的文化群体进行评估。其结果将是在实际科学工作的基础上对科学结构的示意性说明。然后，基于这一程序的分析将能够描绘出科学的结构和重叠。智力上的优点。除了这种方法将提供对科学差异化结构的宝贵见解外，该项目的智力价值源于它承诺回答文化传播领域的长期问题，如科学引用所证明的那样。特别是，它将使用来自潜在语义分析的变量来测试进化文化理论的基本原理，这是一个很有前途的多学科运动，其发展因缺乏具体研究而受阻。这种缺乏并不是因为对理论缺乏兴趣，而是因为将其付诸实施的一个令人烦恼但不可阻挡的问题--即难以找到合适的分析单位和分析工具。这项提议代表了这些问题的一种解决方案，这些问题源于从进化机制和种群生物学角度对科学进步的一长串思考的高潮。通过这样做，它将检验这样一个假设，即文化传播是一个服从于人口思维方法的进化过程；并揭示科学成功在多大程度上是人口变量的函数，这些变量描述了文章及其聚合主题所占据的语义空间的密度和多样性。由于该方法可以集成到当前的网络引文研究中，并可扩展到其他领域和领域，因此它有可能演变为一种分布式的、渐进式的研究方案。关于该项目的更广泛影响，拟议的研究有两个值得注意的实际应用。首先，将科学表现为一种差异化的文化系统，通过揭示科学空间的潜在语义结构，有助于向大学和授权管理人员提供信息。目前，科学工作被明确分类，并根据关于科学研究分布的假设提供资金--这些假设可能是错误的或过时的。使用潜在语义分析来考察基于语义内容的科学资源的实际分布，可以克服这些限制，使管理员能够适当地分配科学资源和构建学术计划。其次，它还可以通过提供语义搜索能力的原型来帮助多学科研究，克服了目前科学文档搜索中的限制，即基于词汇术语进行搜索；使用潜在语义分析将能够基于语义进行搜索，而不考虑词汇匹配。这将对在多学科领域工作的科学家特别有用，这些领域的术语差异可能会阻碍信息的获取，但它也可以通过降低查找科学文件所需的术语特殊性来增加公众获取信息的机会。

The following is a proposal for a project in Semantic Citation Analysis--an approach to science citation research promising novel insight into the organization and transmission of scientific knowledge by incorporating psycholinguistic and neo- Darwinian theory. The goal of the project is to establish a new method for modeling cultural differentiation and transformation using textual data drawn from the published abstracts of a major science citation index. Using a computational procedure called Latent Semantic Analysis, the meaning of science articles' abstracts will be quantitatively represented, allowing their semantic similarity to other articles in the index to be analyzed and graphically illustrated using multi-dimensional scaling. Doing so will allow scientific fields to be evaluated as discrete cultural groups within science on the basis of their latent semantic similarities. The result will be a schematic illustration of the structure of science on the basis of actual scientific work produced. Analyses based on this procedure will then be able to delineate the structure and overlap of the sciences. Intellectual merit. In addition to the valuable insight into the differentiated structure of science that this approach will afford, the project's intellectual merit stems from the answers it promises to longstanding questions in the area of cultural transmission as exemplified by scientific citation. In particular, it will use variables derived from the Latent Semantic Analysis to test the fundamental tenets of Evolutionary Culture Theory, a promising multi-disciplinary movement whose development has been impeded by a lack of concrete research. This paucity has stemmed not from lack of interest in the theory but rather a vexing yet inexorable problem with operationalizing it--namely the difficulty in finding suitable units of analysis and analytical tool. The proposal represents one solution to these problems derived from the culmination of a long line of thinking about scientific progress in terms of evolutionary mechanisms and population biology. In doing so, it will test the hypothesis that cultural transmission is an evolutionary process amenable to a population thinking approach; and reveal the degree to which scientific success is a function of population variables describing the density and diversity of semantic space occupied by articles and their paradigmatic themes. Because the approach can be integrated into current network citation research and is extensible to other domains and fields, it has the potential to evolve into a distributed, progressive research program.Broader Impacts. With regard to the broader impacts of the project, there are two notable practical applications of the proposed research. First, representing science as a differentiated cultural system can help inform university and grant administrators by revealing the latent semantic structure of scientific space. At present, scientific work is categorized explicitly and funded on the basis of assumptions regarding the distribution of scientific research--assumptions that may be erroneous or outdated. Using Latent Semantic Analysis to examine the actual distribution of scientific resources on the basis of semantic content can overcome these limitations and allow administrators to properly allocate scientific resources and structure academic programs. Second, it can also aid multi-disciplinary research by offering the prototype for a semantic search capability, overcoming current limitations in scientific document searches, which search on the basis of the lexical terms; using Latent Semantic Analysis would enable searching on the basis of semantics, regardless of a lexical match. This will be especially useful to scientists working in multi-disciplinary areas, where terminological differences across fields may impede access to information, but it can also increase the public's access to information by lowering terminological specificity needed to find scientific documents.