首页> 外文会议>Portland International Conference on Management of Engineering and Technology >Cross-Domain Academic Paper Recommendation by Semantic Linkage Approach Using Text Analysis and Recurrent Neural Networks
【24h】

Cross-Domain Academic Paper Recommendation by Semantic Linkage Approach Using Text Analysis and Recurrent Neural Networks

机译:基于文本分析和递归神经网络的语义链接方法跨领域学术论文推荐

获取原文

摘要

In this digital age, free-flow and exchange of knowledge and information are of paramount importance. This is the prime reason why we decided to tackle cross-domain linkage. Firstly, we build a system which recommends scholarly academic papers based on the content of news article a user is reading using text analysis techniques. We perform a human expert evaluation to test the system for relevance. Our judges show good agreement with a kappa value of 0.869. To improve the quality of recommendations further, we use an RNN-LSTM model trained on Wikipedia to measure document relevance. We reorder a list of academic papers based on their semantic similarity with the input document using our RNN-LSTM model. Our model achieves a slightly better performance than one of the best document embedding techniques doc2vec (paragraph vector). To the best of our knowledge, ours is the first study linking the domains of News Media and Academic landscape, and bridging the knowledge-gap.
机译:在这个数字时代,知识和信息的自由流通和交流至关重要。这就是我们决定解决跨域链接的主要原因。首先,我们建立了一个系统,该系统根据用户使用文本分析技术阅读的新闻文章的内容来推荐学术论文。我们执行人类专家评估,以测试系统的相关性。我们的法官表现出很好的一致性,kappa值为0.869。为了进一步提高建议的质量,我们使用了在Wikipedia上训​​练的RNN-LSTM模型来衡量文档的相关性。我们使用RNN-LSTM模型根据与输入文档的语义相似性对学术论文列表进行重新排序。与最好的文档嵌入技术doc2vec(段落向量)之一相比,我们的模型实现了略微更好的性能。据我们所知,我们是第一个将新闻媒体和学术领域联系起来并弥合知识鸿沟的研究。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号