predicated on the belief that information filtering can be more effective when humans are involved in the filtering process. Tapestry was designed to support both content-based filtering and collaborative filtering, which entails people collaborating to help each other perform filtering by recording their reactions to documents they read. The reactions are called annotations; they can be accessed by other people’s filters. Tapestry is intended to handle any incoming stream of electronic documents and serves both as a mail filter and repository; its components are the indexer, document store, annotation store, filterer, little box, remailer, appraiser and reader/browser. Tapestry’s client/server architecture, its various components, and the Tapestry query language are described.

This paper describes Ishmail, a program designed for people who get a lot of electronic mail. Most email programs do not address the main problem experienced by people who get a lot of email: information overload. Given a deluge of email, how does one maintain control over incoming message traffic and reduce the time required to find important messages? Some email programs support classification of messages into separate mailboxes, but this is only a partial solution. Ishmail is unique in that it not only sorts messages into mailboxes, but it orders mailboxes by a combination of user-specified priorities and alarms. While most mail programs only alert users about unread messages, Ishmail supports independent alarms on each mailbox with customizable thresholds and filters. Users control their alarms, mailboxes, and messages through customizable summaries that act as both views and interactive controls. Three additional unique features of Ishmail are 1) the ability to read messages safel...

We consider the problem of enabling users to access their mailbox files from any host on our local network, and not only on one designated "home machine". We require a solution which will not introduce any new single points of failure, force us to modify mail transfer agents and user agents, or require changes to the operating system kernels. In other words, minimize the amount of work needed by system-administrators and users. Our solution is to deliver mail into the users' home directories, which are exported via NFS[20, 25] to all of the machines on our network. We wrote a small user-level NFS server implementing a single symbolic link that references the home directory of a user, either the one who accessed it, or by name, with a fallback reference in case of failures. This enables electronic mail to be delivered directly into the user's home directory, which is already accessible from any machine on the network. Although we have used our server primarily for mail delivery redirect...

CMU-ITC-88-064 The Andrew Message System (AMS) is a portable, distributed, multi-media, multi-interface system for r_ding and sending mail and bulletin board (bboard) messages. Mail and bulletin board processing was selected as a "showcase " application to demonstrate how the Andrew file system and user interface toolkit could be applied most usefully to a user's daily tasks. The AMS supports multi-media messages, which may include line drawings, hierarchical drawings, spreadsheets, raster images, animations, and equations. It is explicitly designed to support a huge database of messages and an enormous user community. At CMU, it services over 1200 bboards, including nemews, the Dow Jones information service broadtape, and bboards on which newspaper cartoons appear as raster images. The system incorporates a B-tree based "white pages " for doing name lookx._ ups, including phonetic matching of misspelled names. In addition, the system supports a number of advanced features such as voting on multiple-choice questions, private bboards, shared mailboxes, and automatic classification of incoming mail messages. The server-based architecture makes it easy for client interfaces to be ported to or built on almost any computer. Currently, interfaces run on IBM RTs, DEC Micro-Vaxes, Suns, IBM PC's, Macintoshes, and Vax UNIXt and VMS timesharing systems. 1.

In this paper we introduce the concepts of Logical and Physical Network Locality and point out their importance to the performance of distributed systems. We then describe the design of IPwatch, a simple and inexpensive tool for monitoring logical network locality. IPwatch exploits short-term locality to enable monitoring of medium- and long-term locality of large networks using modest computational resources. We describe experiments at Carnegie Mellon University to validate our ideas and to calibrate IPwatch. The results confirm the existence of substantial short-term locality in this environment. Less than 5 percent of the possible host pairs account for 75 percent of the traffic, and less than 15 percent of them account for 90 percent. Comparative measurements on another network in our environment show even stronger short-term locality. Copyright Ó 1988 Mark J. Lorence and M. Satyanarayanan This work was supported by the National Science Foundation (Contract No. CCR-8657907), Defen...

Scale should be recognized as a primary factor influencing the architecture and implementation of distributed systems. This paper uses Andrew, a distributed environment at Carnegie Mellon University, to validate this proposition. The design of Andrew is dominated by considerations of performance, operability and security. Caching of information and placing trust in as few machines as possible emerge as two general principles that enhance scalability. The separation of con-cerns made possible by specialized mechanisms is also valu-able. Heterogeneity is a natural consequence of growth and anticipating it in the initial stages of system design is impor-tant. A location transparent shared file system considerably enhances the usability of a distributed environment. 1.

We consider the problem of enabling users to access their mailbox files from any host on our local network, and not only on one designated "home machine". We require a solution which will not introduce any new single points of failure, force us to modify mail transfer agents and user agents, or require changes to the operating system kernels. In other words, minimize the amount of work needed by system-administrators and users.