Statistics and Computing

, Volume 11, Issue 2, pp 155–169 | Cite as

A hypergraph-theoretic analysis of collapsibility and decomposability for extended log-linear models

  • F. M. Malvestuto


Extended log-linear models (ELMs) are the natural generalization of log-linear models when the positivity assumption is relaxed. The hypergraph language, which is currently used to specify the syntax of ELMs, both provides an insight into key notions of the theory of ELMs such as collapsibility and decomposability, and allows to work out efficient algorithms to solve some problems of inference. This is the case for the three search problems addressed in this paper and referred to as the approximation problem, the selective-reduction problem and the synthesis problem. The approximation problem consists in finding the smallest decomposable ELM that contains a given ELM and is such that the given ELM is collapsible onto each of its generators. The selective-reduction problem consists in deleting the maximum number of generators of a given ELM in such a way that the resulting ELM is a submodel and none of certain variables of interest is missing. The synthesis problem consists in finding a minimal ELM containing the intersection of ELMs specified by given independence relations. We show that each of the three search problems above can be reduced to an equivalent search problem on hypergraphs, which can be solved in polynomial time.

acyclic hypergraphs collapsibility decomposability independence relation perfect decomposability total decomposability 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. Arnborg S., Corneil D.G., and Proskurowski A. 1993. Complexity of finding embeddings in a k-tree. SIAM J. Alg. Disc. Math. 9: 277-284.Google Scholar
  2. Asmussen S. and Edwards D. 1983. Collapsibility and response variables in contingency tables. Biometrika70: 367-378.Google Scholar
  3. Badsberg J.H. 1995. A guide to CoCo: An environment for graphical models. Part I: efficient algorithms for contingency tables. Ph.D. Thesis, Dept. of Math. and Computer Science, Ålborg University.Google Scholar
  4. Beeri C., Fagin R., Maier D., and Yannakakis M. 1983. On the desirability of acyclic database schemes. J. of the ACM 29: 479-514.Google Scholar
  5. Berge C. 1989. Hypergraphs. North-Holland, Amsterdam.Google Scholar
  6. Bishop Y.M., Fienberg S., and Holland P. 1975. Discrete Multivariate Analysis. M.I.T. Press, Cambridge, MA.Google Scholar
  7. Csiszar I. 1975. I-divergence geometry of probability distributions and minimization problems. Annals of Probability 3: 146-158.Google Scholar
  8. Darroch J.N., Lauritzen S.L., and Speed T.P. 1980. Markov fields and log-linear interaction models for contigency tables. Annals of Statistics 8: 522-539.Google Scholar
  9. D'Atri A. and Moscarini M. 1988. On hypergraph acyclicity and graph chordality. Information Processing Letters 29: 271-274.Google Scholar
  10. De Campos L.M. 1996. Characterization of decomposable dependency models. J. Artificial Intelligence Research 5: 289-300.Google Scholar
  11. Fagin R. 1983. Degrees of acyclicity for hypergraphs and relational database schemes. J. ACM 30: 514-550.Google Scholar
  12. Frydenberg M. 1990. Marginalization and collapsibility in graphical interaction models. The Annals of Statistics 18: 790-805.Google Scholar
  13. Gabriel K.R. 1969. Simultaneous test procedures: Some theory of multiple comparisons. Ann. Math. Statist. 40: 224-250.Google Scholar
  14. Geiger D. and Pearl J. 1993. Logical and algorithmic properties of conditional independence and graphical models. The Annals of Statistics 21: 2001-2021.Google Scholar
  15. Geng Z. 1989. Decomposability and collapsibility for log-linear models. Algorithm AS 244. Appl. Statistics 38: 189-197.Google Scholar
  16. Haberman S.J. 1974. Log-linear Models for Contingency Tables. University of Chicago Press, Chicago.Google Scholar
  17. Havranek T. 1991. Simple formal systems in data analysis. In: Proc. Conf. on Symbolic and Numeric Data Analysis and Learning, pp. 373-381.Google Scholar
  18. Ku H.H. and Kullback S. 1968. Interaction in multidimensional contingency tables: An information-theoretic approach. J. Research of the National Bureau of Standards, Mathematical Sciences 72B: 159-199.Google Scholar
  19. Lauritzen S.L. 1996. Graphical Models. Oxford Science Publications, Clarendon Press, Oxford.Google Scholar
  20. Lauritzen S.L., Speed T.P., and Vijayan K. 1984. Decomposable graphs and hypergraphs. J. of the Austral. Math. Society, Series A36:12-29.Google Scholar
  21. Lauritzen S.L. and Spiegelhalter D.J. 1988. Local computations with probabilities on graphical structures and their application to expert systems. J. of the Royal Statistical Society, Series B50: 157-224.Google Scholar
  22. Leimer H.-G. 1993. Optimal decompositions by clique-separators. Discrete Math. 113: 99-123.Google Scholar
  23. Lien Y.E. 1981. Hierarchical schemata for relational databases. ACM Transactions on Database Systems 6: 48-69.Google Scholar
  24. Lien Y.E. 1982. On the equivalence of database models. J. ACM 29: 333-362.Google Scholar
  25. Madigan D. and Mosurski K. 1990. An extension of the results of Asmussen and Edwards on collapsibility in contingency tables. Biometrika77: 315-319.Google Scholar
  26. Maier D. and Ullman J.D. 1984. Connections in acyclic hypergraphs. Theor. Comput. Sci. 32: 185-199.Google Scholar
  27. Malvestuto F.M. 1988. Existence of extensions and product extensions for discrete probability distributions. Discrete Mathematics 69: 61-77.Google Scholar
  28. Malvestuto F.M. 1992.A unique formal system for binary decompositions of database relations, probability distributions, and graphs, Information Sciences 59: 21-52; comment by Malvestuto F.M. and Studeny M. 1992. Information Sciences 63: 1-2.Google Scholar
  29. Malvestuto F.M. 1992.Recognizing submodels of loglinear hierarchical models. In: Dodge Y. and Whittaker J. (Eds.), Computational Statistics, Vol. 1 (Proc. X Symp. on Computational Statistics), Physica Verlag, Heidelberg, pp. 263-267.Google Scholar
  30. Malvestuto F.M. 1994. Statistical versus relational join dependencies. In: Proc. VII Int. Conf. on Scientific amp; Statistical Database Management, pp.64-78.Google Scholar
  31. Malvestuto F.M. 1996a. An axiomatization of loglinear models with an application to the model-search problem. In: Fisher D. and Lenz H.-J. (Eds.), Lecture Notes in Statistics,Vol. 112, Springer-Verlag, New York, pp. 175-184.Google Scholar
  32. Malvestuto F.M. 1996b. Testing implication of hierarchical log-linear models for discrete probability distributions. Statistics and Computing 6: 169-176.Google Scholar
  33. Malvestuto F.M. 1997. Computing marginals in graphical models. In: Wegman E.J. and Azen S.P. (Eds.), Computing Science and Statistics, Vol. 29(2), Proc. of the IIWorld Congress of the IASC, Pasadena, CA, pp. 52-58.Google Scholar
  34. Malvestuto F.M. 1998a. From conditional independences to decomposable models and beyond. In: Proc. VII Int. Conf. on Information Processing amp; Management of Uncertainty, pp. 795-802.Google Scholar
  35. Malvestuto F.M. 1998b. A complete axiomatization of full acyclic join dependencies. Information Processing Letters 68:133-139.Google Scholar
  36. Malvestuto F.M. and Moscarini M. 1998. A fast algorithm for query optimization in universal-relation databases.J. of Computer amp; System Sciences 56: 299-309.Google Scholar
  37. Malvestuto F.M. and Moscarini M. 2000. Decomposition of a hypergraph by partial-edge separators. Theoretical Computer Science 237:1-2: 57-79.Google Scholar
  38. Shibata Y. 1988. On the tree representation of chordal graphs. J. of Graph Theory 12: 421-428.Google Scholar
  39. Tarjan R.E. 1985. Decompositions by clique separators. Discrete Mathematics55: 221-232.Google Scholar
  40. Tarjan R.E. and Yannakakis M. 1984. Simple linear-time algorithms to test chordality of graphs, test acyclicity of hypergraphs, and selectively reduce acyclic hypergraphs. SIAM J. Computing 13: 566-591.Google Scholar
  41. Wen W.X. 1990. Optimal decomposition of belief networks, in Proc. of the VIWorkshop on Uncertainty in Artificial Intelligence, pp. 245-256.Google Scholar
  42. Wermuth N. 1976. Model search among multiplicative models. Biometrics32: 253-263.Google Scholar
  43. Yannakakis M. 1981. Computing the minimum fill-in is NP-complete. SIAM J. Alg. Disc. Math. 2: 77-79.Google Scholar

Copyright information

© Kluwer Academic Publishers 2001

Authors and Affiliations

  • F. M. Malvestuto
    • 1
  1. 1.Dip. Di Scienze Dell'InformazioneUniversità “La Sapienza” Di RomaRomaItaly

Personalised recommendations