Context-aware web search using dynamically weighted information fusion
Context-aware web search using dynamically weighted information fusion
复制标题
DOI：
10.1002/cpe.1805
复制
发表时间：
2013-04-01
影响因子：
2
通讯作者：
Anderson, Nicole
中科院分区：
文献类型：
--
作者：
Anderson, Nicole
网络搜索已经成为互联网用户的基石工具。一个人能够用商业工具快速检索到的信息令人敬畏，但普通用户仍然对找到真正与他们相关的信息感到沮丧。在我们的研究中，我们试图表明，如果我们利用关于用户及其当前上下文的关键信息片段，我们可以提供更准确的结果。我们通过探索上下文数据、使用查询增强技术并适当地融合结果来实现这一点，以便在向用户呈现结果时提供最佳排序。上下文数据包括用户的个人日历、位置、偏好、个人词汇表和同行推荐。查询增强涉及利用相关的上下文信息来修饰用户的查询文本以产生更有针对性的查询。信息融合允许使用上下文数据通过使用乘积和、贝叶斯技术或两者的组合来比较和重新排序查询结果。我们发现，通过将这些方法结合起来，我们能够显著提高搜索结果。我们已经进行了一个案例研究，并在这里展示了结果。版权所有(C)2011 John Wiley&Sons，Ltd.
Web search has become a cornerstone tool for Internet users. The information one is able to retrieve quickly with commercial tools is awe inspiring, yet average users still remain frustrated with finding information that is truly relevant to them. In our research, we seek to show that if we utilize key pieces of information about a user and their current context, we can provide more accurate results. We do this by exploring context data, using query enhancement techniques, and appropriately fusing the results to provide the best ordering when presenting the results to the user. Context data includes a user's personal calendar, location, preferences, personal vocabulary, and peer recommendations. Query enhancement involves utilizing relevant contextual information to decorate a user's query text to produce a more focused query. Information fusion allows the query results to be compared and re-ordered using contextual data by utilizing either a sum of products, a Bayesian technique, or a combination of the two. We find that by combining these methods, we are able to significantly enhance the search results. We have performed a case study and present the results here. Copyright (c) 2011 John Wiley & Sons, Ltd.