It's easy to imagine how the calendar is used. A user might be looking for an email message about workflow systems that he remembers printing around the time of a meeting he attended on this topic in early April. Browsing through his calendar page for this month he sees he had an appointment to "review workflow market" on April 8th. Inspection of the image thumbnails shown in this cell of the calendar reveals two images with the texture pattern characteristic of an email message. Clicking on one of these thumbnails yields the desired document.
Security is another important aspect of a system like the one described in this paper. Users would like the opportunity to share documents with one another. However, at the same time, they would like to guarantee that their confidential documents remain secret. Furthermore, they would also like to retain the full capability to search and retrieve such confidential documents.
A two-tier security model was adopted to satisfy these requirements. The simplest method allows a user to designate their document collection as either open or closed. If a collection is open, the web server allows access to any host. If a collection is closed, a user must enter the appropriate password to gain access.
Users can also encrypt individual documents by pressing the "Encrypt" link (see Figure 3 for an example). This uses the PGP algorithm to encrypt the data files for a document. Some rudimentary searching capability is retained by not modifying the full text index. Thus, full text search can still be used to find an encrypted document.
3 System Usage
The automatic document storage and retrieval system described in this paper has been in use at CRC for over two years. There are several obvious questions about such a technique. One concerns the amount of storage required. Other questions concern the usefulness of the various document sources (printers vs. copiers). Another question concerns the popularity of the various retrieval interfaces. That is, given that users are already familiar with full text search from their interaction with web search engines, will they readily adopt the additional techniques used here (10 most recent and calendar retrieval). The answers to these questions were investigated by examining the storage and access logs kept by the web server for CRC's IM³ automatic document storage and retrieval system.
