论文总字数：23134字

摘 要

学术创新是一种创新研究，它是指在某领域一定的学术成果积累的基础上，对所涉及到的内容、理论、方法进行开拓式研究。在此过程中所展现出的创新能力与素质，则称为学术创新力。由此，可以认为学术创新力是一种体现在学术研究过程中的能力。学术期刊作为一个科学研究成果的展示平台，在学术传播链中占据了重要的位置。作为一道连接受众与学界的桥梁，学术期刊及时有效地传达学术界的研究新风向与新成果。同时能够迅速将成果在现实中的转化效果反馈给学界。这种逆反馈将会对学术创新研究产生现实的指导作用。然而现阶段对于学术创新力的研究主要以论文或学科为载体，较少涉及针对学术期刊的创新力评价。本文针对学术期刊，对学术创新力的评价方法展开进一步的探究。

本文主要研究内容包括：第一，研究创新力评价相关理论基础，对现有的几种评价角度：文献计量法、同行评议法、综合法等进行分析评价。第二，单独提取可能对学术创新力数值有影响的指标，初步构建期刊学术创新力指标模型。第三，探究单个创新力指标与创新力数值之间的线性关系，分析单个指标对创新力数值的影响。构建并提出模型，对特征指标进行筛选，建立最有效的期刊学术创新力模型。

本文拟通过数据挖掘的方式,运用神经网络，决策树，回归模型等分析法进行分析研究。实证研究表明，基于神经网络模型的期刊学术创新力模型是有实证作用的。对初始特征指标进行分析后，可以剔除三个无效指标，从而得出最有效的指标体系，为期刊学术创新力的而构建提供一些有价值的指导。

关键词：学术创新力 指标特征 数据挖掘

Research on the Evaluation of Journal Academic Innovation Based on Data Mining——Taking the Field of Information Security as an Example

Abstract

Academic innovation is an innovative research. It refers to the pioneering research on the content, theory, and methods involved on the basis of the accumulation of certain academic achievements in a certain field. The innovative capabilities and qualities shown in this process are called academic innovation. From that, we can think of innovation as a potential inherent in research. As a display platform for scientific research achievements, academic journals occupy an important position in the academic communication chain. As a bridge connecting the audience and the academic world, academic journals timely and effectively convey the new research trends and new achievements of the academic world. At the same time, it can quickly feed back the transformation effect of the achievements in reality to the academic world. This kind of inverse feedback will have a practical guiding role in academic innovation research. However, at the present stage, the research on academic innovation mainly takes papers or disciplines as carriers, and rarely involves evaluation of the innovation of academic journals. This article aims at academic journals and further explores the evaluation methods of academic innovation.

The main research contents of this article include: first, research the relevant theoretical basis of innovation evaluation, and analyze and evaluate several existing evaluation angles: document measurement method, peer review method, comprehensive method, etc. Second, extract characteristic indexes related to innovation in journals, and initially build an original evaluation index system for innovation. Third, explore the linear relationship between a single index of innovation and the value of innovation, and analyze the impact of a single index on the value of innovation. Build and propose a model, screen feature indicators, and establish the most effective model of academic academic innovation.

This paper is intended to use information mining for analysis and research through muscle networks, decision trees, retrospective models, and other methods of analysis. Practical research shows that innovation in journalism is based on a network of muscles that have tangible effects. After analyzing the initial characteristic indexes, three invalid indexes can be eliminated, so as to obtain the most effective index system, and provide some valuable guidance for the construction of journals' academic innovation.

Key words: Academic; Innovation Index; Characteristics Data mining

目 录

摘要 I

Abstract II

目录 IV

第一章 绪论 1

1.1 研究背景 1

1.2 研究意义 2

1.3 研究方法和思路 3

1.4 研究方法和思路 4

第二章 相关理论基础和文献综述 6

2.1 学术创新力相关概念 6

2.2 学术创新力测度方法 6

2.3 学术创新力评价文献综述 7

2.3.1 学术创新力研究发展的研究综述 7

2.3.2 学术创新力评价模型的研究综述评价 8

2.3.3 对现有研究成果的评述 8

第三章 学术创新力的特征指标的构建 9

3.1 学术创新力评价指标体系构建思路 9

3.2 影响学术创新力的各类指标 10

3.2.1 创新吸引力 10

3.2.2 创新力产出指标 10

3.2.3 创新扩散力指标 11

3.2.4 内在影响因素 11

3.3 本章小结 12

第四章 期刊学术创新力的构建及实证分析——以信息安全领域为例 13

4.1 数据准备 13

4.2 数据预处理 14

4.3 单个指标与学术创新力的相关性分析 14

4.3.1 创新吸引力特征 14

4.3.2 创新产出力特征 16

4.3.3 创新扩散力特征 17

4.3.4 内在影响因素 17

4.4 学术创新力模型构建 19

4.4.1 多元线性回归模型 19

4.4.2 决策树模型 20

4.4.3 神经网络模型 22

4.4.4 模型对比 23

4.5 学术创新力评价模型确认 24

4.5.1 指标分析 24

4.5.2 模型指标确认 24

4.5.3 结论 25

第五章 总结与展望 26

5.1 本文工作总结 26

5.2 研究的不足 26

参考文献 28

附录 29

致谢 31

第一章 绪论

1.1 研究背景

请支付后下载全文，论文总字数：23134字