Abstract:
i2 is one of the leading suppliers of Link Analysis tools with thousands of analysts using our products all around the world. As such, i2 is keenly aware of the problems facing analysts as they strive to cope with ever increasing amounts of information. i2 has itself demonstrated that software is a key ingredient in a modern analytical tool-set. However, the key processing of the source information still takes place in the mind, rather than the CPU. The software may collect, store, index, retrieve and display the information; but the human analyst is required to understand it, to realize its significance, to assimilate it into a 'mental model' of the scenario and to draw the appropriate inferences. Existing algorithms used in link analysis can identify areas of interest in the data and can efficiently re-organize the entities into layouts that reveal the underlying topology and key connecting entities. Analysts are now accustomed to commercial Link Analysis products, like i2’s Analyst’s Notebook, that are easy to use, robust, can integrate with existing databases and are capable of running on sensible platforms. New AI techniques need to be constructed on top of these existing proven platforms such as those now made possible by component based architectures. If new AI techniques are to be accepted, researchers need to consider carefully how they introduce AI techniques to the analytical community. They should look to 'amplify' the role of the human analysts rather than threaten them with 'automation' and utilize the existing, accepted analytical tool sets rather than seeking to replace them with immature tools. They should also work with real clients on real problems to ensure that the new techniques are truly useful and scalable. As the author’s company, i2, has perhaps the first component based Link Analysis platform in widespread use, we would be happy to hear from organizations conducting such research that would benefit from exposure to real products, problems and users.