A ranking algorithm based on contents and non-key attributes for object-level keyword search over relational databases

Keyword search technique over relational databases is a research hot-spot in database field. At present, there have been many ranking correlation algorithms for object-level keyword search over relational databases. Object-level keyword search can better integrate information scattered in various tuples. OCS(Object-level Correction Sort) algorithm cannot rank results in keyword search accurately as was expected. This paper foucuses on the problems of ranking results in keyword search system for object-level over relational databases and proposes a new ranking algorithm SOCA(Sort of Correction Algorithm) which takes into consideration the content information of key attributes, and the correlation of non-key attributes. We use Weight to evaluate the content information of key attributes, and Correlation to assess the correlation of non-key attributes Finally, we give a score function about contents Correlation and Weight. Experiments demonstrate that this algorithm can effectively rank results and verify its reasonableness and effectiveness.