## Neighborhood Interchangeability for Non-Binary CSPs & Application to Databases (2005)

Citations: | 1 - 1 self |

### BibTeX

@TECHREPORT{Lal05neighborhoodinterchangeability,

author = {Anagh Lal},

title = {Neighborhood Interchangeability for Non-Binary CSPs & Application to Databases},

institution = {},

year = {2005}

}

### OpenURL

### Abstract

Neighborhood Interchangeability (NI) identifies the equivalent values in the domain of a variable in a Constraint Satisfaction Problem (CSP). We introduce for the first time an algorithm for computing NI sets in the presence of non-binary constraints. We integrate this mechanism with backtrack search, in a process we call dynamic bundling. We demonstrate that, as for the binary case [Beckwith et al., 2001], dynamic bundling yields multiple robust solutions for less effort than necessary for computing a single solution. We then identify the utility of this mechanism for database applications and introduce a new algorithm based on dynamic bundling for computing a join query, which we model as a CSP. We argue that the algorithm yields a compact solution space and saves memory, disk-space, and/or network bandwidth. Finally, we discuss the application of the join algorithm to materialize views.