| The mission of the Advanced Search Technologies project is to improve the ability of people to find and organize information in an enterprise setting. The group is responsible for the Minion search engine, a flexible, scalable engine that is the basis for our research. Minion is available as an open source project under the GPL license. Previous versions of the Minion engine have shipped as part of the Sun Java System Portal Server and Sun Java System Web Server. The Minion engine is being used as a core component of the AURA Project's scalable data store. It is responsible for indexing the text and fields associated with the items that are stored in a data store. The AURA Project's interfaces and recommenders take advantage of Minion's boolean and relational search capabilities as well as Minion's document similarity measures. Minion also provides automatic document classification capabilities, light weight and full English morphologies that can be used at query time, the ability to build taxonomies of terms at indexing time, and supervised and unsupervised sense disambiguation components. Current work on the Minion engine is focused around supporting the needs of the AURA Project. In particular, we are interested in clustering approaches that can be used to cluster terms associated with documents. For example, we are interested in being able to determine that canada and canadian are highly related terms that should be treated as a single word when performing document similarity or classification computations. Web Resources for The Advanced Search Technologies Project: |
