Indexes for Document Retrieval with Relevance
Indexes for Document Retrieval with Relevance
复制标题
相关性文档检索索引
DOI：
--
复制
发表时间：
2013
期刊：
影响因子：
--
通讯作者：
J. Vitter
中科院分区：
文献类型：
--
作者：
W. Hon;Manish Patil;Rahul Shah;Sharma V. Thankachan;J. Vitter
文档检索是一种特殊类型的模式匹配，与信息检索和Web搜索密切相关。在这个问题中，数据由文本文档的集合组成，并且给定查询模式P，我们需要报告出现此模式的所有文档（而不是所有实例）。此外，相关性的概念通常用于对满足查询的所有文档进行排名，并且仅返回具有最高相关性的那些文档。这种相关性的概念已经成为当今搜索引擎（如Google，Bing，Yahoo或Ask）的有效性和可用性的核心。当考虑相关性时，查询具有附加的输入参数k，并且任务是仅报告与P具有最高相关性的k个文档，而不是找到包含P的所有文档。例如，一个这样的相关性函数可以是文档中的查询模式的频率。在信息检索文献中，这一任务最好通过使用倒排索引来实现。但是，如果查询包含任意字符串（可以是部分单词、多单词短语或更一般的任何字符序列），则无法利用单词边界，需要采用不同的方法。
Document retrieval is a special type of pattern matching that is closely related to information retrieval and web searching. In this problem, the data consist of a collection of text documents, and given a query pattern P, we are required to report all the documents (not all the occurrences) in which this pattern occurs. In addition, the notion of relevance is commonly applied to rank all the documents that satisfy the query, and only those documents with the highest relevance are returned. Such a concept of relevance has been central in the effectiveness and usability of present day search engines like Google, Bing, Yahoo, or Ask. When relevance is considered, the query has an additional input parameter k, and the task is to report only the k documents with the highest relevance to P, instead of finding all the documents that contains P. For example, one such relevance function could be the frequency of the query pattern in the document. In the information retrieval literature, this task is best achieved by using inverted indexes. However, if the query consists of an arbitrary string—which can be a partial word, multiword phrase, or more generally any sequence of characters—we cannot take advantages of the word boundaries and we need a different approach.