Augmenting Information Retrieval by Knowledge Infusion

WiseWare is an operational knowledge-based system providing support for help desk analysts, users in Information Technology departments and channel partners [Delic and Lahaix, 1998]. A WiseWare document is an entity containing human readable knowledge expressed in free form English text. Documents are evaluated and annotated for complexity, quality, usefulness and age, etc. The retrieval of documents relevant to some query can be augmented by the infusion of some domain knowledge. This knowledge infusion can take place at one or both of two distinct places in the retrieval process: • before the search, knowledge of a user’s habits, preferences and query history, etc. can be used to modify a query • after the search, knowledge of a user’s preferences, skill level and viewing history, etc. can be combined with knowledge of the retrieved documents’ complexity, quality, age and source, etc. to rank and sort the results in a more relevant manner. This paper investigates the benefit of knowledge infusion after the search process.