The quantity of on-line information is growing and changing at a startling rate. The number of users seeking this information is also growing quickly. Many users are computer novices who are likely to encounter great difficulty locating the information they seek within a complex information space. This difficulty arises from two principal sources: complex, unintuitive user interfaces and the ill-structured nature of most information spaces. Our current research addresses these difficulties through the use of spoken and written natural language interfaces coupled with automated indexing techniques designed to extract semantic sense from ill-structured data.