Decentralized and individual solutions for unstructured problems are typical in many organizations. However, finding solutions often involves a duplication of effort in the solution discovery process. In addition, the solution discovery process seldom uses previous experience. Formalizing decentralized and individualized methods that provide support from current organizational knowledge about similar unstructured problems reduces the duplication of effort and creates a learning environment. Such a learning environment requires a supporting infrastructure that allows individuals to extract learned information from organizational repositories. Repository information also disseminates problem knowledge between closely related problem domains (Henninger et al, 1995).
In the learning organization, the repositories provide knowledge and grow as new problems are solved. The use of prior knowledge is a beginning point for the collection of new information. Effort is saved by not collecting 'known' information but focusing on the retrieval of value-added information.
Populating the repository needs to be continuous and quick in order to promote a learning environment. In addition, newly retrieved and previously stored information must be continually assessed for usefulness in solving the current problem Harvey et al, 1997). With a repository focused on learning support, some structure is provided for the unstructured problems.
With most knowledge management systems, it is the responsibility of the searcher to know about, access and retrieve the documents needed to complete a task. The information in the documents is generally confined to the document - there is no exchange of knowledge between documents other than by the searcher's processing. In some cases, database structures have been overlaid on document retrieval/knowledge management systems to provide a knowledge base within an organization (Liongosari et al, 1999). This knowledge base provides a source for obtaining organizational knowledge. However, knowledge can only be discovered if the organization has specifically built the data structure to accommodate the searcher's need.
There are systems that have been designed to extract relevant information from unstructured sources such as the Web. The PHOAKS (People Helping One Another Know Stuff) system searches Usenet FAQ's to identify a consensus of Web sites valid for a domain (Terveen et al, 1997). Specialized search engines and indexes have been developed for many domains (Selberg and Etzioni, 1995). Search engines have been developed to combine the efforts of other engines (Selberg and Etzioni, 1995) and select the best search engine for a domain (Howe and Dreilinger, 1997). However, these approaches do not consider the user's experience in previous searches.
User preferences have been addressed by establishing profiles. Agents search out Web sites on user stated interests (Ackerman et al, 1997) or through the joint interests of a group of users (Balabanovit and Shoham, 1997). These approaches do not consider the organization's environment or other users' experiences with specific Web sites.
Some Web search engines find information by categorizing the pages in their indexes. One of the first to create a structure as part of their Web index is Yahoo! (http://www.yahoo.com). Yahoo! has developed a hierarchy of documents, which is designed to help users find information faster. This hierarchy acts as a taxonomy of the search engine index. Yahoo! helps by directing the searcher through the levels of the taxonomy. The searcher sees document titles and summaries, which may be indexed by keywords that may not have been used in the original search. Such an organization of the contents of the Web into categories could help in an organization's search. The categories, however, need to meet the specific search requirements and then be populated with the appropriate Web pages. …