HITIQA : A Question Answering Analytical Tool

† HITIQA (High Quality Interactive Question Answering) is currently being developed to assist analysts in finding answers to complex intelligence problems, efficiently and thoroughly. The system uses event-based, data-driven semantic processing and natural language dialogue, coupled with an advanced information visualization interface, to deliver accurate answers to the analyst's questions along with related contextual information. The first version of the system has undergone a series of preliminary evaluations with the analysts from the US Naval Reserve, producing valuable usage and performance data. These evaluations suggest that HITIQA creates a measurable cognitive augmentation effect for the analyst. The second more advanced version of the system is currently being implemented. HITIQA is an advanced question answering system that helps the analyst to produce higher quality reports for complex intelligence problems in less time and with lower cog-nitive load. The primary function of HITIQA is to supply composite answers to complex, exploratory questions such as " What is the state of development of long range missiles in North Korea? Can they reach the U.S.? " Submitting such questions to a conventional internet search produces tens of thousands of hits that include many related and unrelated documents of varying length and veracity. More questions may need to be asked along the way to fill information gaps or to explore related topics such as production capabilities or missile technology proliferation, etc. Each time a question is posed, substantial effort must be applied to retain relevant facts, note contradictions, ignore repetitions and discard unrelated and unreliable sources. The problem is not usually lack of information; more often it is too much in-formation—fragmented, indirectly related, sometimes mis-leading—and the lack of skilled assistance to help the analyst wade through it. HITIQA does not return long lists of documents, as keyword search does; instead, it retains only the most relevant passages and assembles them into a coherent composite answer. HITIQA selects its answer more carefully, too: keyword match may be a reasonable indication of potential relevance, but until we know why these words are found together, answer precision is likely to be low. In order to improve accuracy, HITIQA performs named entity extraction from candidate text passages and then uses several pro-totypical event templates (called frames) to arrive at the most likely interpretation. The text passage is thus rendered into an event frame which assigns event roles to the entities found in text. The list of available roles varies …