Various crime specific behaviors are widely used in criminal association analysis to detect the relationship between suspect networks. However, currently available link analysis tools for crime detection fail to utilize such behaviors for detecting various types of criminal cases. Instead, these tools generally provides generic functions for all types of crimes such as kidnapping, terrorism, drug trafficking, assaults, and robbery etc. These tools also rely on the users’ expertise on the domain knowledge of the crime for successful detection. As a result, these tools are less effective and inflexible in detecting specific patterns in certain crimes. In addition, association analysis still encounters many challenging issues. For instance, it is extremely time consuming in analyzing vast amount of available crime data and visualizing structural views of the whole criminal network. In order to address these challenges, the current trend of association analysis systems will be reviewed in this thesis. The properties of existing criminal detection system will be compared and a list of useful properties of criminal detection system will be identified. A generalized system for criminal association detection system will be proposed and developed for automated link analysis. The proposed system will also be augmented with a knowledge-base which contains money laundering specific rules and heuristics. The overall system will provides a structured event-based database to store criminal report, a module for detection of specific criminal, an association detection module, and an association visualization module to identify relationship and determine the link weight. A prototype system based on the proposed techniques will be designed and implemented.