【摘要】：針對中文專利侵權(quán)檢測中關(guān)鍵詞特征表達能力弱以及句子結(jié)構(gòu)特征容易引起噪聲干擾的問題,提出了一種通過抽取三元組特征來改進中文專利侵權(quán)檢測的方法。該方法將專利權(quán)利要求書抽取為三元組特征的集合,并結(jié)合詞向量技術(shù)和How Net計算三元組特征間的語義相似度,從而有效提高對疑似侵權(quán)專利的識別能力。實驗結(jié)果表明,該方法取得了較好的檢測效果,且在準確率上要高于其他方法。
[Abstract]:Aiming at the problem of weak expression of keyword features in Chinese patent infringement detection and noise interference caused by sentence structure features, this paper proposes a method to improve Chinese patent infringement detection by extracting triple features. In this method, the patent claim is extracted as a set of triple features, and the semantic similarity between the features of the triple is calculated by combining word vector technology and How Net, so as to improve the ability to identify the suspected patent infringement effectively. The experimental results show that this method has better detection effect and is more accurate than other methods.
【作者單位】： 江蘇大學計算機科學與通信工程學院;南京審計大學工學院;
【基金】：國家自然科學基金資助項目(71271117) 江蘇省六大人才高峰項目(2013-WLW-005) 江蘇省自然科學基金資助項目(BK20150531)
【分類號】：G306;TP391.1
，

本文編號：2161603