Querying weak instances

1. INTRODUCTION The universal relation model gives the user a view of the data as though it was stored in a single relation, in which every attribute plays a unique role. Thus, in posing queries, the user does not have to navigate among the different base relations; the system performs this navigation automatically by transforming the query on the universal relation into one involving the stored base relations. For example, in a films database with relations FP (Film-Producer) and FD (Film-Diiwtor), ln response to the query rcrrieve P where D-FelJmi (which producers has Fellmi worked for), the system would " know " that diiectors are related to produc-ers' through films. Thus, the relation between producers and directors would be constructed by taking the join of relations FP and FD and projecting on the attributes P, D. The fust approach towards defmlng the relationship between the actual database (the base relations) and the universal relation is known as the pure universal instance assumption. This assumption postulates that the database is formed by projecting some universal relation (satisfying the dependencies), and the universal relation is formed by taking the join of all the base relations. This assumption places very severe restrictions on the database, since incomplete information is not allowed-each tuple of a base relation must match with tuples in all the other relations. One could allow dao-glmg tuples (tuples that don't match), but then in taking the joii of all the relations these tuples would get lost. For example, suppose that in the films database we have also a relation FA (Film-Actor). Then, if we take the join of all three relations FP, FD, FA , we loose the relationship between directors of documentaries and their

[1]  J. D. Uiiman,et al.  Principles of Database Systems , 2004, PODS 2004.

[2]  David Maier,et al.  On the foundations of the universal relation model , 1984, TODS.

[3]  Mihalis Yannakakis,et al.  Algebraic dependencies , 1980, 21st Annual Symposium on Foundations of Computer Science (sfcs 1980).

[4]  Yannis Vassiliou.  A formal treatment of imperfect information in database management , 1980 .

[5]  Mihalis Yannakakis,et al.  Equivalences Among Relational Expressions with the Union and Difference Operators , 1980, J. ACM.

[6]  Alfred V. Aho,et al.  Efficient optimization of a class of relational expressions , 1978, SIGMOD Conference.

[7]  Robert E. Tarjan,et al.  Simple Linear-Time Algorithms to Test Chordality of Graphs, Test Acyclicity of Hypergraphs, and Selectively Reduce Acyclic Hypergraphs , 1984, SIAM J. Comput..

[8]  Yehoshua Sagiv,et al.  Can we use the universal instance assumption without using nulls? , 1981, SIGMOD '81.

[9]  Jeffrey D. Ullman,et al.  The U. R. strikes back , 1982, PODS.

[10]  David Maier,et al.  Windows on the world , 1983, SIGMOD '83.

[11]  Mihalis Yannakakis,et al.  Algorithms for Acyclic Database Schemes , 1981, VLDB.

[12]  Leslie G. Valiant,et al.  Negation can be exponentially powerful , 1979, Theor. Comput. Sci..

[13]  David Maier,et al.  Testing implications of data dependencies , 1979, SIGMOD '79.

[14]  Mihalis Yannakakis,et al.  On monotone formulae with restricted depth , 1984, STOC '84.

[15]  Alfred V. Aho,et al.  Equivalences Among Relational Expressions , 1979, SIAM J. Comput..

[16]  Yehoshua Sagiv,et al.  A characterization of globally consistent databases and their correct access paths , 1983, TODS.

[17]  Alfred V. Aho,et al.  Efficient optimization of a class of relational expressions , 1979, TODS.

[18]  Alberto O. Mendelzon.  Database states and their tableaux , 1984, TODS.

[19]  Peter Honeyman,et al.  Testing satisfaction of functional dependencies , 1982, JACM.

[20]  Leslie G. Valiant,et al.  Exponential lower bounds for restricted monotone circuits , 1983, STOC.