CitizenLink: Scalable Graph-Based Modeling and Analysis of Multi-Generational Citizen Relationships

This paper contributes a graph database approach to Bangladesh’s national citizen record keeping. It uses Neo4j to store citizens as nodes and family ties as direct links. This lets the system answer deep family queries that relational databases handle poorly. The work delivers three main things. First, it builds a realistic synthetic population of over 5 million citizens across four generations. The generator follows 47 rules drawn from real Bangladesh data on marriage age, religion, fertility, and polygyny. Since the data is synthetic, others can reuse it without privacy risk. Second, it proves the speed claim at scale. Indexed lookups run in about 13 milliseconds. Relationship queries stay under 100 milliseconds on a graph of 15.3 million elements. Query cost depends on family size, not population size. Third, it introduces a relationship classifier that reads the order of steps along a path. Simple hop counting cannot separate a co-wife from a step-mother, because both sit two hops away. This classifier gets both right. It also separates half-siblings from full siblings using a parents map. The system further adds a rule based birth audit that flags biologically impossible birth gaps and duplicate registrations. The rules are open and checkable, so officials can justify any decision. Earlier AI methods were accurate but gave no auditable reasoning. This paper closes that gap by joining graph speed with legal transparency in one working prototype.