This work addresses the problem of relaxing spatial constraints for pictorial queries having null answers in geographical databases. It focuses on the polygon-polyline topological relationship and proposes a computational model which is based on the notions of Operator Conceptual Neighborhood (OCN) graph and the relative 16-intersection matrix. The former is addressed to represent the conceptual topological neighborhood between pairs of Symbolic Graphical Objects and indicates how spatial constraints can be relaxed. The nodes of the OCN graph are labeled with geo-operators that have been formalized and their semantics has been enriched in order to capture user query details. The latter is a 16- intersection matrix which provides additional information about the query with respect to the well-known 9-intersection matrix proposed in the literature. It has been conceived to identify, among the approximate answers, the one closer to the user needs. In particular, it allows us to quantify the difference among the user query and the configurations corresponding to the proposed approximate answers on the basis of the OCN graph. The main characteristics of our approach are highlighted through some query examples.