@techreport{TR-IC-97-09,
  number = {IC-97-09},
  author = {Nascimento, Mario A.},
  title = {B${}^+$-tree Based Approach to Index Transaction Time},
  month = {July},
  year = {1997}, 
  institution = {Institute of Computing, University of Campinas},
  note = {In English, 21 pages.
    \par\selectlanguage{english}\textbf{Abstract}
      Transaction time of a record is the time interval when the
      record is stored in the database.  In this paper we present an
      approach which provides efficient indexing of such kind of temporal
      data.  The approach makes use of two standard B${}^+$-trees with
      trivially modified node split policies---which yield a usage ratio of
      virtually 100{\%} in each tree.  We compare the proposed
      approach, which we name Two-Stage, to the Monotonic B${}^+$-tree (by
      Elmasri et al).  Our simulations show that the Two-Stage approach
      yields a structure up to 75{\%} smaller than the Monotonic
      B${}^+$-tree, and in all but one of the several investigated
      scenarios, the Two-Stage approach provided faster (or comparable)
      query processing time.  Our main contribution, however, lies in the
      fact that the Two-Stage Approach does not require novel data
      structures but well-known B${}^+$-trees.  As such, and unlike all
      previous techniques for tackling this problem, it can be implemented
      using facilitites existing on most commercial DBMSs.
  }
}