Processing Relational Top-N Queries With Text And Numeric Attributes

In recent years, as one of the leading subject field of information retrieval and database，the researches on top-N query, which focused mainly on query processing strategies andsorting functions, have been booming. Relational database-based top-N queries can retrievethe first N tuples which match the keywords best, and then sort the results to output by thespecified sorting function. Currently, new progress of top-N queries only for numericattributes are continuously made, however, it is still relatively seldom that how to deal withtext attributes, and further to combine the two. Top-N queries that support natural semanticsnot only be able to find exact results, but also can find the same or similar answers onsemantic.The results set is sorted according to comprehensive distance of both text andnumeric, so complex queries are achieved. This paper discusses the top-N queries withsimultaneous processing of both text and numeric attribute.This paper presents an approach to achieve top-N queries.Using the WordNet system tosemanticly expand words, we establish an index that contains kinship words and numericinformation for efficient storage, pre-loading and retrieval.The index is searched firstly, to getall the identities of the results.Then, We compute semantic and numerical distance and sortaccording to sorting function, to get the candidate set.Finally, complete information can beobtained from the database through the SQL statement, resulting in top-N results.Theexperiments include the time and space cost, and accuracy, the results show that this methodis effective and efficient.