**Subsequence versus substring constraints in sequence pattern languages**

*Acta Informatica*( IF 0.900 )

**Pub Date : 2019-11-07**

*, DOI:*

*10.1007/s00236-019-00347-5*

Steven Engels, Tony Tan, Jan Van den Bussche

### Abstract

A family of logics for expressing patterns in sequences is investigated. The logics are all fragments of first-order logic, but they are variable-free. Instead, they can use substring and subsequence constraints as basic propositions. Propositions expressing constraints on the beginning or the end of the sequence are also available. Also wildcards can be used, which is important when the alphabet is not fixed, as is typical in database applications. The maximal logic with all four features of substring, subsequence, begin–end constraints, and wildcards, turns out to be equivalent to the family of star-free regular languages of dot-depth at most one. We investigate the lattice formed by taking all possible combinations of the above four features, and show it to be strict. For instance, we formally confirm what might intuitively be expected, namely, that boolean combinations of substring constraints are not sufficient to express subsequence constraints, and vice versa. We show an expressiveness hierarchy results from allowing multiple wildcards. We also investigate what happens with regular expressions when concatenation is replaced by subsequencing. Finally, we study the expressiveness of our logic relative to first-order logic.