Document Search Practices

IntroductIon A large portion of the knowledge of most organizations is contained in electronic documents. For users to get pertinent information from the accumulation of stored documents, they need effective document retrieval systems. Unfortunately, electronic document management has fallen into the same trap that electronic data processing fell into: simply automating what previously was done manually. Paper documents were stored in folders in drawers in file cabinets. Electronic documents are stored in folders in directories on disk drives. The ability to find a document depends on the logic of the filing system, how familiar the individual is with the filing system, and how familiar the individual is with the problem domain of the item being sought. Some persons (e.g., research librarians) are much better than others at organizing and retrieving documents. Rarely, however, is a manager an expert at either storing or retrieving documents. Unfortunately, many electronic filing systems are set up by managers with little or no training on how to organize a filing system, and few tools, other than the Windows Search command, are available to help managers find documents that have been filed. The filing systems for libraries and knowledge management systems are more sophisticated than the filing systems of most small offices or individual managers. But even libraries and knowledge management systems predominately rely on keyword searching for retrieval. For example, if one visits the Web site for the Journal of Management Information Systems at http://jmis.bentley. edu/keywords/, one notes that the only option available for searching (other than browsing the entire collection) is a keyword search. Keyword searching has improved over the years. Knowledge seekers have benefited enormously from the ability to search remotely, the increased speed with which searches are conducted, and the ability of the search mechanism to identify variations of the keywords. Nevertheless, keyword searches have significant limitations. In particular , keyword searches cannot return all relevant documents nor can they filter out irrelevant documents. This article briefly reviews the difficulties associated with keyword searches, especially as the number of documents increases, and proposes a way to overcome those limitations.