TY - GEN
T1 - Efficient support of historical queries for multiple lines of evolution
AU - Landau, Gad M.
AU - Schmidt, Jeanette P.
AU - Tsotras, Vassilis J.
PY - 1993
Y1 - 1993
N2 - Traditional approaches to history reconstruction problems assume a single line of evolution, i.e., a system (a database, a relation or a collection of objects) evolves over time by applying a set of operations on it. Historical queries are of the form: `find the employees working in a company on January 1st, 1990'. There are however, complex applications where the system's state evolves in a way that results in multiple lines of evolution (such as for example a company that splits into a number of subsidiaries, which on turn may split into new subsidiaries and so on). In general, this creates a tree (hierarchy) of evolution lines, where each node in the tree represents the evolution of a subsidiary. Consider the following (vertical) historical query: `find the employees working for a subsidiary C or one of its ancestors, on January 1st, 1990'; it may well be that C did not yet exist at that time, but it was an ancestor A of C that existed; in this case the answer to the query are the employees of A. Another (horizontal) historical query of interest is the following: given a subsidiary B, `find all the employees which were employed on January 1st, 1990, by subsidiaries which were descendants of B'. This paper focuses on the search parts of these historical queries; several reconstruction methods, designed for single evolution lines may then be used.
AB - Traditional approaches to history reconstruction problems assume a single line of evolution, i.e., a system (a database, a relation or a collection of objects) evolves over time by applying a set of operations on it. Historical queries are of the form: `find the employees working in a company on January 1st, 1990'. There are however, complex applications where the system's state evolves in a way that results in multiple lines of evolution (such as for example a company that splits into a number of subsidiaries, which on turn may split into new subsidiaries and so on). In general, this creates a tree (hierarchy) of evolution lines, where each node in the tree represents the evolution of a subsidiary. Consider the following (vertical) historical query: `find the employees working for a subsidiary C or one of its ancestors, on January 1st, 1990'; it may well be that C did not yet exist at that time, but it was an ancestor A of C that existed; in this case the answer to the query are the employees of A. Another (horizontal) historical query of interest is the following: given a subsidiary B, `find all the employees which were employed on January 1st, 1990, by subsidiaries which were descendants of B'. This paper focuses on the search parts of these historical queries; several reconstruction methods, designed for single evolution lines may then be used.
UR - http://www.scopus.com/inward/record.url?scp=0027189337&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:0027189337
SN - 0818635703
T3 - Proceedings - International Conference on Data Engineering
SP - 319
EP - 325
BT - 1993 IEEE 9th International Conference on Data Engineering
PB - Publ by IEEE
T2 - 1993 IEEE 9th International Conference on Data Engineering
Y2 - 19 April 1993 through 23 April 1993
ER -