联系我们

论文查询结果	返回搜索

论文编号：	9371
作者编号：	2120152412
上传时间：	2017/6/20 20:43:19
中文题目：	高校人文社科研究机构科研方向导向作用的研究
英文题目：	A Study on the Guidance of Scientific Research Direction in University Humanities and Social Research Institutions
指导老师：	石鉴
中文关键字：	导向作用；科研方向；学术成果产出；文本相似度；系统聚类
英文关键字：	Guidance, Scientific Research Direction, Academic Achievements, Text Similarity, Hierarchical Clustering
中文摘要：	近年来，随着我国经济、社会、政治、文化、生态文明建设的飞速发展，国家对高校科研机构，在重大问题上的学术成果产出，也寄予了更高的期望。目前，高校科研机构的考核重点，仍集中于科研投入的数量、机构建设情况以及科研产出的数量和质量。对于高校科研机构科研方向是否真正发挥了导向作用，期初设立的科研方向，期末是否有与之对应的学术成果，期初的科研方向和期末的成果产出方向二者的拟合程度如何，目前还不得而知。这个问题也非常值得探讨的。 本文首先对研究背景进行了分析，提出研究的目的和意义。之后，通过对现阶段科研机构评价内容的总结，发现目前还没有专门的指标评价科研方向的导向作用，从评价结果中无法直接观测到科研机构在既定研究方向的产出情况。而后，总结了常用的科研机构的评价方法，发现对于科研方向导向作用的发挥，也就是科研方向和成果产出的拟合程度，目前还没有专门的研究和方法，现有的评价方法也不适合研究这一问题。因此本文创造性的引入文本挖掘的理论和方法，并对高校人文社科重点研究基地这一典型高校科研机构进行了实证研究，在研究该问题的同时，也弥补了这一研究领域的空白。 本文从定量和定性两个方面，对基地期初既定科研方向和期末的学术成果方向，二者的拟合程度做出综合度量。一方面，借助于文本分类中常用的文本相似度的概念，将文本特征词作为学术主题词，对拟合程度进行度量。另一方面，通过系统聚类的方法，对于成果文本的特征词进行聚类，据此划分出科研机构成果的产出方向，将其和期初基地上报的科研主攻方向进行比较，给出拟合程度定性评价的结果。但是受系统聚类方法本身特点的限制，一个特征词只能属于一个类簇，因此聚类结果与真实情况存在差异。针对这一问题，本文提出了基于共词关系频次统计的系统聚类结果优化方法，允许一个特征词属于多个类簇，改善了聚类结果的可解读性，提高成果产出方向识别的准确性。通过定性和定量两种方法的运用，本文发现部分科研机构的科研方向和成果产出方向拟合程度不高。 而后，通过将科研方向和成果产出方向拟合程度的计算结果，和目前科研考核体系的评价结果进行相关分析，发现二者之间相关性很弱。因此存在这样的现象：一些科研机构没有按照科研管理部门的要求，在指定科研方向上有学术成果产出，但在期末考核中却取得了好成绩。针对这一现象的原因，本文从科研机构和科研管理部门两个方面进行了分析，并提出了相应的建议。
英文摘要：	In recent years, with the rapid development of economy, society, politics, culture and ecological civilization, our country has a higher expectation on the academic achievements about major problems from university research institutions. However, the amount of invested resources, the construction of university research institutions, quantity and quality of the academic achievements are still key assessment items. People pay little attention to the guidance of scientific research direction, or the fitting degree between scientific research direction and output. Thus, this question is worth discussing. First of all, this paper will make an introduction for relevant backgrounds, put forward the aim and signification. After that, based on the summary of the evaluation of scientific research institutions at present, it is found that there is no specific index to evaluate the guidance of scientific research direction. We can’t know whether there are academic achievements related to the scientific research direction. Moreover, the existing methods which are widely used in the evaluation of university research institutions are not suitable to solve this problem. Therefore, this paper introduces the theory and method of text mining creatively, which makes up the blank of this research field. This paper will make use of qualitative and quantitative methods to measure the fitting degree between scientific research direction and output. On one hand, based on the definition of text similarity commonly used in text classification, this paper measures the fitting degree from word frequency dimension. On the other hand, by using the method of hierarchical clustering, the characteristic words of the academic achievements text are divided into several clusters which corresponds with different output directions. Then, we can compare these output directions with scientific research directions and measure the fitting degree qualitatively. However, for hierarchical clustering method, a feature word can only belong to one cluster. This is unrealistic obviously. Therefore, based on Co-word analysis, this paper puts forward a method that allowing a feature words belong to multiple clusters to improve the clustering results. By using qualitative and quantitative methods, the fitting degree of some university research institutions between scientific research direction and output is low. Then, by comparing the fitting degree with the evaluation results of the current scientific research evaluation system, it is found that their correlation is weak. So there are some scientific research institutions which don’t have academic achievements in specified scientific research direction, but they get better results in the final assessment. This paper analyses the phenomenon from two aspects, scientific research institutions and scientific research management departments, then puts forward some corresponding suggestions.
查看全文：	预览  下载(下载需要进行登录)