Evaluating Logic-Based Scoring Functions on Uncertain Relational Data

Abstract

  Nowadays, for many retrieval scenarios a strict
  query evaluation just returning
  a Boolean truth value
   is not sufficient
  anymore. We often rather need the support of a gradual query
  fulfilment expressed by a score value out of the interval [0,1].
  ProQua is a new probabilistic database
  system which combines such information retrieval
  concepts with database technologies. In contrast
  to other state-of-the-art probabilistic database
  systems ProQua facilitates logic-based 
  similarity conditions within its SQL-like
  query language by a generic similarity operator.
  In this work we formalise logic-based scoring functions
  as the underlying concept of the supported similarity conditions
  and introduce respective evaluation
  techniques implemented by relational query plans.
  Additionally, we report on their experimental verification
  on a probabilistic TPC-H database.