MINOS is an information system that provides integrated facilities for creating and managing multimedia objects. Implemented on a SUN workstation equipped with an addressable analog speech storage device, it accommodates documents comprising any mixture of text, image (raster and structured), and voice. Multimedia documents are created, modified, filed, retrieved, browsed, and archived within the system. All operation is online (no hard copy).

MINOS has an object-oriented model of multimedia documents that permits parts to be shared between documents. It also includes version management to assist in working with several versions of documents. Objects may be in the editing or archived states; in the latter case, they are not modifiable. MINOS includes a text editor (UNIX “vi”) and an image editor, and will include a voice editor.

The user is able to retrieve information through query specification, browsing through documents, and browsing within documents. Queries can operate not only on the text part of a document, but also on the image and voice parts; however, only the caption and data type can be searched. Queries can also specify presentation information such as position on the page.

The authors mention presentation modes such as browsing, zooming, transparencies, narration, voice segments, annotations, and process simulation. Browsing is by page number, section number, and word search; bookmarks are available. Transparencies allow annotations to be overlaid on images. Narration is different from voice segments because it begins automatically when a certain point in the document is reached, while specific action must be taken to hear voice segments. So-called “annotations” are links to independent multimedia documents. Process simulation is scripted animation by fast page turning.

Since it is not designed to fulfill any stated requirements, it is difficult to evaluate the MINOS architecture. It is by no means clear to this reviewer how, or whether, such a system would be used in practice. Despite its length, the paper does not really indicate how flexible the system is (to take a simple example, can tables--which are a kind of image--contain text? pictures?). None of the components seem particularly novel; the emphasis is on getting it all together. But the system seems to suffer from creeping featurism. What are its goals, and what does the architecture exclude? Given that all the parts exist, what are the research problems that are being addressed? This paper gives no clear answers to these fundamental questions.