In this paper we proposed a heuristic and semi-automatic framework for inducing one-to-one mapping bilingual lexicon of intra-family languages by using pivot language and preexisting bilingual lexicon resources based on our one-to-one mapping assumption intra-family language lexicons. The framework uses heuristics and human interaction to complete induction in iterative manner. Weight of each heuristics is predefined and designated to be adjusted runtime. In the paper three basic heuristics are predefined: Probability, Pivot strength and Spelling similarity, and using them to conduct an experiment. Experimental result shows that by using our framework we can achieve (1) up to 95.3% correctness in majority of induced one-to-one mapping pairs, and (2) up to 85.2% overall correctness of entire one-to-one mapping pairs. Meanwhile number of human interaction needed for higher quality is minimized. We also designed a tool as an implementation of the framework which provided fully visualized workflow to demonstrate induction process, and its interfaces with dynamic graph components offers better user experience while users are observing data in experiment and interacting with automated induction process.