基于多通道注意力的科技需求大数据实体关系抽取
首发时间:2020-12-18
摘要:随着科学技术的发展,利用科技需求大数据中地实体关系来进行演进规律分析成为了一个重要的研究方向。通过对科技需求大数据文本信息进行实体关系抽取,人们可以更加详细的对科学技术之间的关系进行分析,全面地了解科技的发展态势,从而分析科技需求地发展趋势和规律。然而科技需求数据存在实体数量密集、实体之间关系复杂以及特征属性繁多等研究难点,导致科技需求大数据实体关系抽取准确率不高。针对以上问题本文提出一种基于多通道注意力机制的科技需求大数据实体关系抽取算法(Big Data of Science And Technology Demand Based on Multi-Channel Attention Mechanism)来进行实体关系抽取。试验结果表明:在不用外部资源的情况下,本算法在科技需求大数据中实体关系抽取的准确率、召回率以及F1值比其他对比算法均有所提升,准确率提升值为4.6%、召回率提升值为5.1%、F1值的提升值为4.3%。
关键词: 科技需求大数据 实体关系抽取 多通道注意力机制 演进规律分析
For information in English, please click here
Entity relationship extraction of technology demand big data based on multi-channel attention
Abstract:With the development of science and technology, it has become an important research direction to analyze the evolution law by using the relationship between land entities in big data of science and technology demand. Through the extraction of entity relationship of text information of big data of science and technology demand, people can analyze the relationship between science and technology in more detail, comprehensively understand the development trend of science and technology, and analyze the development trend and law of science and technology demand. However, there are many difficulties in the research of science and technology demand data, such as the number of entities is dense, the relationship betwEntity relEntity relationship extraction of science and technology demand big data based on multi-channel attentionationship extraction of science and technology demand big data based on multi-channel attentioneen entities is complex, and the feature attributes are various, which lead to the low accuracy of entity relationship extraction of science and technology demand big data. To solve the above problems, this paper proposes a big data of science and technology demand based on multi-channel attention mechanism to extract entity relationship. The experimental results show that: in the case of no external resources, the accuracy rate, recall rate and F1 value of this algorithm in the extraction of entity relationship in the big data of science and technology demand are improved compared with other comparative algorithms, the improvement value of accuracy rate is 4.6%, recall rate is 5.1%, and F1 value is 4.3%.
Keywords: Big data of science and technology demand Entity relation extraction Multi channel attention mechanism Analysis of evolution law
基金:
引用
No.****
动态公开评议
共计0人参与
勘误表
基于多通道注意力的科技需求大数据实体关系抽取
评论
全部评论0/1000