A genetic algorithm based heuristic search on graphs with weighted multiple attributes

Graphs are efficient data structures used for representing connected real world entities and are very ubiquitously used in various applications. Graphs with multiple attributes describing each node and having non-negative weights assigned to these are called as Weighted Multiple Attributed Graphs (WMAGs). WMAGs are able to capture and represent the properties of real world data efficiently. Graphs such as co-authorship network, friendship network, communication network etc. belong to this category. Queries on these graphs are complex than traditional graph search. Querying WMAG helps in finding the inherent relations between users in social media, finding closely connected authors who are expertized in various research domains from a co-authorship network, etc. This work addresses the problem of graph search on weighted multiple attributed graphs by specifying attribute and structural constraints. In the proposed system, an R-Tree based index is used for efficient look ups of nodes which meets the attribute constraints. Finding nodes which spans minimum distance is considered as the structural constraint. The k-constraint query remodels the large data graph into a k partite graph. Problem of finding k-minimally separated nodes according to k-constraints from a WMAG is reduced to generalized minimum spanning tree (GMST) problem. GMST problem is NP-Hard and here a heuristic method based on genetic algorithm is proposed as the solution, which is efficient in terms of query processing and execution time.