Blake C, Pratt W
Information and Computer Science Department, University of California, Irvine, USA.
Proc AMIA Symp. 2000:81-5.
The number of publications available to physicians and patients is increasing at an alarming rate. Although users can use tools to assist in reformulating their query, this approach is ineffective when their information needs are imprecise or many documents are relevant. The ranked list presentation of documents provides little or no information relating documents to the initial query or to each other. We argue that information overload can be reduced if documents are placed in categories that (a) relate to the initial query and (b) contain a manageable number of documents. Dynamic Categorization is a knowledge-based approach that satisfies (a), however the number of documents in a category may still be large. We demonstrate that using the same approach to re-categorize the documents in large categories reduces the number of documents in the subcategory, and maintains a clear relationship to the initial query.
可供医生和患者使用的出版物数量正以惊人的速度增长。尽管用户可以使用工具来辅助重新表述他们的查询，但当他们的信息需求不精确或许多文档都相关时，这种方法是无效的。文档的排名列表展示几乎没有提供与初始查询或彼此相关的文档信息。我们认为，如果将文档分类为（a）与初始查询相关且（b）包含数量可控的文档，那么信息过载是可以减少的。动态分类是一种基于知识的方法，它满足（a），然而一个类别中的文档数量可能仍然很多。我们证明，使用相同的方法对大类别中的文档进行重新分类可以减少子类别中的文档数量，并与初始查询保持明确的关系。
