Workflow Extraction from Textual Process Descriptions
Workflow Extraction from Textual Process Descriptions
复制标题
从文本流程描述中提取工作流程
DOI：
--
复制
发表时间：
2016
期刊：
影响因子：
--
通讯作者：
Pol Schumacher
中科院分区：
文献类型：
--
作者：
Pol Schumacher
Wikipedia、WordNet和词共现是文本处理中用来解决各种问题的知识源。在文本分类中，这些知识源被用来丰富文档表示，提高分类精度。简单的文章可以通过查看标题或作者来正确分类，而其他文章可能需要更多的处理才能正确分类。这种方法与人类如何在浅层方法失败后逐渐诉诸更复杂的处理过程有着有趣的对应关系。作为博士研究的一部分，我们试图提出一种方法，我们试图为每个文档建议一个原则性的知识源组合，使用最少的知识来解决问题。这是通过估计处理文档的难度并建议可以降低难度的正确知识来源来实现的。
Wikipedia, WordNet and word co-occurrences are knowledge sources that have been used in text processing to solve various problems. In text classification, these knowledge sources are used to enrich the document representation and increase classification accuracy. Easy articles can be classified correctly by looking at just the title or author while others might require more processing to classify correctly. Such an approach has interesting correspondences to how humans progressively resort to more involved processing once shallower approaches fail. As part of doctoral research, we attempt to propose a method, where we try to suggest a principled combination of knowledge sources for each document that uses minimal knowledge to solve the problem. This is done by estimating the difficulty in processing the documents and suggesting the right knowledge source that would reduce this difficulty.