Can Aysu Betin, Baykal Nazife
METU Informatics Institute, Inönü Bulvari, 06531 Ankara, Turkey.
Comput Methods Programs Biomed. 2007 Apr;86(1):73-86. doi: 10.1016/j.cmpb.2007.01.007. Epub 2007 Feb 22.
We present a new next generation domain search engine called MedicoPort. MedicoPort is a medical search engine designed for the users with no medical expertise. It is enhanced with the domain knowledge obtained from Unified Medical Language System (UMLS) to increase the effectiveness of the searches. The power of the system is based on the ability to understand the semantics of web pages and the user queries. MedicoPort transforms a keyword search into a conceptual search. Through our system we present a topical web crawling technique and indexing techniques empowered by the semantics information. MedicoPort aims to generate maximum output with semantic value using minimum input from the user. Since MedicoPort is designed to help people seeking information about health on the web, our target users are not medical specialists who can effectively use the special jargon of medicine and access medical databases. Medical experts have the advantage of shrinking the answer set by expressing several terms using medical terminology. MedicoPort provides the same advantage to its users through the automated use of the medical domain knowledge in the background. The results of our experiments indicate that, expanding the queries with domain knowledge, such as using the synonyms and partially or contextually relevant terms from UMLS, increase dramatically the relevance of an answer set produced by MedicoPort and the number of retrieved web pages that are relevant to the user request.
我们展示了一种名为MedicoPort的新型下一代领域搜索引擎。MedicoPort是一款为没有医学专业知识的用户设计的医学搜索引擎。它通过从统一医学语言系统(UMLS)获取的领域知识得到增强,以提高搜索的有效性。该系统的强大之处基于理解网页语义和用户查询的能力。MedicoPort将关键词搜索转变为概念搜索。通过我们的系统,我们展示了一种由语义信息赋能的主题网络爬虫技术和索引技术。MedicoPort旨在以最少的用户输入产生具有语义价值的最大输出。由于MedicoPort旨在帮助人们在网络上查找健康信息,我们的目标用户不是能够有效使用医学专用术语并访问医学数据库的医学专家。医学专家具有通过使用医学术语表达几个术语来缩小答案集的优势。MedicoPort通过在后台自动使用医学领域知识为其用户提供同样的优势。我们的实验结果表明,利用领域知识扩展查询,例如使用UMLS中的同义词以及部分或上下文相关的术语,会显著提高MedicoPort生成的答案集的相关性以及与用户请求相关的检索网页数量。