Seeing Is Retrieving: Building Information Context From What the User Sees

As the user's document and application workspace grows more diverse, supporting personal information management becomes increasingly important. This trend toward diversity renders it difficult to implement systems which are tailored to specific applications, file types, or other information sources. The paper developed SeeTrieve, a personal document retrieval and classification system which abstracts applications by considering only the text they present to the user through the user interface. Associating the visible text which surrounds a document in time, SeeTrieve is able to identify important information about the task within which a document is used. This context enables novel, useful ways for users to retrieve their personal documents.