基于引文分析和深度学习的新兴技术识别算法研究

发布时间：2018-11-16 14:48

【摘要】：新兴技术对企业的发展具有深远的影响,它可以引导企业跟上技术发展的趋势。众所周知利用新兴技术发展趋势可以帮助企业做出重要的技术和业务决策。因此完成新兴技术的识别工作就显得尤为重要,而现有的模型大多数是针对某些特定的技术领域进行识别,这通常会带有一定的局限性,无法满足实际的要求。鉴于此,本研究重点如下:首先,对美国1975-2009年的专利引用数据进行了深入的分析,提取了一些对新兴技术识别具有代表性的特征项。根据以上提取的所有特征对专利引用数据进行了索引和聚类的预处理,接下来应用本研究提出的新兴技术和非新兴技术类别自动标注算法,实验结果表明类别自动标注算法取得了较好的效果。然后结合深度信念网络和逻辑斯蒂克回归模型构建了本研究提出的基于深度学习的新兴技术识别算法。其次,使用了美国专利商标局提供的专利引用数据来测试和评估基于深度学习的新兴技术识别算法。该算法通过自动选取专利引文数据中最佳的特征组合来进行新兴技术的识别。测试表明,该方法能够准确和稳定地识别新兴技术,通过与其它算法进行对比,发现新算法的准确率比其它算法要高1.05%。
[Abstract]:Emerging technology has a profound impact on the development of enterprises, it can guide enterprises to keep up with the trend of technological development. It is well known that the use of emerging technology trends can help enterprises to make important technical and business decisions. Therefore, it is very important to complete the identification of emerging technologies, and most of the existing models are aimed at some specific technical fields, which usually have some limitations and can not meet the actual requirements. In view of this, the focus of this study is as follows: firstly, the patent citation data from 1975 to 2009 in the United States are analyzed in depth, and some representative features for the identification of emerging technologies are extracted. According to all the features extracted above, the index and clustering of patent reference data are preprocessed, and then the automatic labeling algorithm of emerging and non-emerging technology categories is applied in this paper. The experimental results show that the algorithm is effective. Then a new technology recognition algorithm based on deep learning is constructed by combining depth belief network and logical Steak regression model. Secondly, the patent reference data provided by the U.S. Patent and Trademark Office is used to test and evaluate the emerging technology recognition algorithm based on in-depth learning. The algorithm automatically selects the best feature combination in patent citation data to identify emerging technologies. The test results show that the new algorithm can identify the emerging technology accurately and stably, and by comparing with other algorithms, the accuracy of the new algorithm is 1.0555% higher than that of other algorithms.
【学位授予单位】：湘潭大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP181;TP311.13

【相似文献】