期刊名称：Bulletin of the Technical Committee on Data Engineering
出版年度：2010
卷号：33
期号：01
出版社：IEEE Computer Society
摘要：A major hardness of processing searches issued in the form of keywords on structured data is the ambi-
guity problem. A set of keywords itself is not a complete piece of information and may imply different
information needs from different users. Although state-of-the-art keyword search engines are usually
able to automatically identify meaningful connections among keywords in the data, users may still be
frequently overwhelmed by the huge number of results and face much trouble in selecting the relevant
ones. Many search engines attempt to rank the results in order of their inferred relevance, however, it is
virtually impossible for a ranking scheme to be perfect and works properly for all users and all queries.
In this paper we discuss several post-processing methods for keyword searches on structured data,
which ease the users’ task of finding and digesting relevant results from all results returned for a keyword
query. The methods to be discussed include generating result snippets, differentiating selected results as
well as query expansion. Each of these methods helps users gracefully get the relevant information in its
own way. At last we discuss the remaining challenges in post-processing keyword searches on structured
data.
