Text preview for : SSL-79-6_Computer_Science_and_Office_Information_Systems.pdf part of xerox SSL-79-6 Computer Science and Office Information Systems xerox parc techReports SSL-79-6_Computer_Science_and_Office_Information_Systems.pdf
Back to : SSL-79-6_Computer_Science | Home
Computer Science and
Office Information Systems
By Clarence A. Ellis and Gary J. Nutt
z
c
:=
Computer Science
and
Office Information Systems
BY Clarence A. Ellis and Gary J. Nutt
June 1979
ABSTRACT
Automated office systems are emerging as an interdisciplinary research area with a
strong computer science component. In this paper we define office information systems
as entities which perform document storage, retrieval, manipulation and control within a
distributed environment. Some state of the art implementations are described. We relate
the research to different areas of computer science and provide several detailed
examples.
KEY \AJORDS AND PHRASES
Office automation, distributed systems, office information systems, office modeling.
XEROX
PALO ALTO RESEARCH CENTER
3333 Coyote Hill Road / Palo Alto / California 94304
TABLE OF CONTENTS
INTRODUCTION
WHAT IS AN OFFICE INFORMATION SYSTEM
Officetalk Zero: A Prototype OIS
Goals of Officetalk Zero
Capabilities and Functions
Some Implementation Issues in Officetalk
Lim itations
SCOOP: Another Prototype OIS
The Approach
TIle SCOOP Implementation
TECHNICAL OIS RESEARCH PROBLEMS
Programming Languages
BDL: A Very High Level Business Language
BOL Capabilities and Limitations
Software Engineering
Information Control Nets
An ICN Example
Operating Systems and Databases
Office Systems Consistency
Consistency Within the ICN Model
Computer Architecture
Measurement and Evaluation
OIS Simulation
Distributed Simulation
Communications
Artificial Intelligence
Sociological Issues
Informal Communications in the Office
FUTURE TRENDS IN OIS RESEARCH
ACKNOWLEDGEMENTS
REFERENCES
1
INTRODUCfION
The automated office of the future is quickly becoming the topic of much significant computer
science research. The office machine industry, lead by Burroughs, Eastman Kodak, Exxon, IBM,
3M, and Xerox, is actively working on automating the information processing that takes place in
an office [Creative Strategies, 1978]; most of these corporations are also investing significant sums
of money into research programs for the office of the future. Active programs incorporating
computer science also exist in universities, e.g. at M.I.T. in L.C.S [Hewitt, 1979] and the Sloan
School [Hammer and Zisman, 1979], University of Pennsylvania's Wharton School [Ness, 1976-78;
Morgan, 1976, 1979; Zisman, 1977], the University of Toronto [fsichritzis, 1979] and the Harvard
Business School [BucL.lnan, 1979]. The focus of most of this attention is not traditional business
data processing, nor is it management information systems, but rather systems and facilities to aid
the office worker in the more basic aspects of his/her job. Word processors address the problem
of document preparation, but the worker must also organize, file, copy, transform, analyze and
transmit that information effectively. The automated office should mechanize all these functions as
well, thus allowing the worker to accomplish less routine work.
The need for the automated office creates a new area for applying results, techniques and
methodologies of classic computer science research. However, solutions to a large number of
difficult problems must be obtained before such systems can become a reality. Many such
. -
problems are a result of three more general problems: the complexities of distributed systems that
implement the automated office, the necessity for simple, yet complete human interfaces, and the
need for knowledge-based systems to aid the user.
We recognize that it is difficult to address the entire audience of computer science researchers at a
level which will excite each investigator into an active interest in automated office research. We
have chosen, instead, to mention as many of the appropriate topics as possible, while providing a
more complete discussion of only a few of them. Ibese detailed discussions tend to reflect the
areas of research with which we are most comfortable; one should not necessarily draw the
conclusion that these are "the most important" areas of research in office automation. For more
breadth into related topics of management science, see other Computing Surveys articles [Aron,
1969; Taggart and Tharp, 1977]. We encourage other computer science specialists to provide more
complete discussions of those topics which we have not treated in detail.
The paper deals with three major topics: some example implementations of office information
systems, a discussion of some problems from the standpoint of traditional computer science, and
2
future trends in research. The organization has been designed to allow the reader to obtain an
overview from the introduction of each major section. Additional insight is provided in each
subsection introduction, and finally, several subsections are refined to contain detailed discussions.
3
WHAT IS AN OFFICE INFOR]\tlATION SYSTElVl
The office is that part of a business that handles the information dealing with operation,
accounting, payroll, billing, etc. In particular, office work consists of activities such as document
preparation, filing, performing simple computations, checking information, intraoffice
communication and external communication. Such processing within the office is usually
stimulated by the arrival of a request for service such as an order, a bill, a complaint, a message to
order more materials, or the date changing to Friday. The office, then, can be viewed as a
mechanism that maintains the state of the business, by means of a series of activities that cause
change in state.
The computer scientist can use a number of different models to describe office activity, such as:
e A set of activities resulting from requests for service, each with a specific precedence. Each
activity requires a supporting file system.
e/\ set of people "executing their procedures" ("carrying out tasks"), communicating with and
referencing a supporting file system.
e A set of communication media with thei:- corresponding communications, such as a filled-in
fonn, a phone call, a copy of an order, or a file system query for organizing and processing
information.
e A gigantic database with users accessing and manipulating data.
An automated office information system (0 IS) attempts to perform the functions of the ordinary
office by means of a computer system. Automation in the office particl~larly aids the office worker
in document preparation, information management and decision making. Such systems may be as
modest as a group of independent word processors, or as complex as a distributed set of large,
communicating computers. Within in this spectrum is a central computer with several interactive
terminals, or a set of small interconnected computers. In either system the office worker would
use a work station to perform his work, and that work station would be capable of electronically
communicating with other work stations.
In this paper we distinguish office information systems from data processing systems both by the
autonomy of the system's parts, and by function. A data processing system is used to implement
alogotithms with a single locus of control in which there are ordinarily not collections of
4
autonomous parts; the algorithm ordinarily procedes without the need for human interaction.
Typical data processing systems compute payrolls, implement accounting systems, manage
inventories, etc. An OIS is made up of a collection of highly interactive autonomous tasks that
execute in parallel; the OIS tasks include document preparation, document management,
communication, and aids in decision making.
The terms "office of the future", "automated office", "office information system" and "integrated
office system" have been frequently applied even to small business computer or timesharing
systems. So in order to describe our view of an OIS luore exactly we will present two examples of
what we consider to be state-of-the-art office information systems.
Officetalk-Zero: A Prototype OIS
Officetalk-Zero is a prototype "first generation" office infonnation system, designed and
implemented by William Newman, Tim Mott and others from the Office Research Group at Xerox
PARC, [Newman, 1977]. The Officetalk-Zero effort began in late 1976 as a study of languages for
expressing office procedures, and subsequently evolved into an OIS emphasizing the user interface.
The prototype--operational by June, 1977--was introduced into a naive user environment within the
following year.
Goals of Oificetalk-Zero
Officetalk-Zero, or Officetalk for short, is implemented in an environment of a network of
minicomputers interconnected by a high speed communication network [Metcalfe and Boggs, 197~].
Each minicomputer, a Xerox Alto, is a 128K (16 bit) word minicomputer with a 2.5 megabyte
disk and a sophisticated CRT display [Alto, 1978]. Areas on the screen are pointed to by a cursor
under the control of an x-y coordinate input device called a mouse. The mouse is operated by a
button which is depressed, held down, then released; software can determine the state of the
button as well as the x-y coordinate addressed by the mouse. Even though the PARC
environment encourages the network approach, it is clear that many future automated office
systems will be designed around a similar physical environment [Creative Strategies, 1978].
The Officetalk designers took the position that the new OIS should be based on the data objects of
single page forms and files of forms; intercommunication is accomplished by electronically passing
forms among the work stations. The user's model of the system is that Omcetalk is merely an
electronic aid for carrying out his normal tasks. A primary difference in the user's model (as
opposed to his pre-OIS model) is the lack of real paper at the user's work station. (After all, one
5
goal of office automation is to reduce the use of paper.} Each work station provides a graphical
window onto a worker's desk, allowing the worker to manipulate electronic forms by employing
the pointing device.
Officetalk is not a decision support tool nor is it a management information system; it is intended
to be used by office workers to aid in document management, preparation and communication.
Part of the reason for restricting interest to clerical work was the desire to investigate office
procedure specification and interpretation; the designers recognized that the procedural
specification of "routine clerical work" was an unsolved problem, and that a solution to that
problem would be a step toward the solution of the more general problem.
Many of the individual facilities needed to implement the OIS described above already exist as
separate programs on several computer systems. The user must have a text editor, a graphics
package, electronic mail, a filing facility and a forms data entry capability. However, an OIS must
offer all of these facilities to the user via a simple, uniform interface. Officetalk combines all of
these facilities, plus a few others, into a single, integrated system which is currently being used by
clerical workers.
Capabilities and functions
Officetalk is_a distributed program that executes on at least one minicomputer in conjunction with
the communication network and a central file. server. Ordinarily there will be several
minicomputers, each acting as a work station for an individual user of Officetalk. The central file
server maintains a database describing all pending electronic transactions, e.g., electronic mail,
information about each authenticated user of the system, or a set of tailored blank forms to be
used in the particular applic~tion. 0 fficetalk is designed to save the majority of the user's
i...'1formation state in the central server and as little as possible in the local minicomputer.
To implement Officetalk, a set of blank forms for the application must be designed and entered
into the database. Officetalk provides a forms editor which allows one to design the artwork of a
fonn and to specify the style of each field on the form. The forms editor requires that the newly
designed forms satisfy certain rules, such as no overlapping fields; it also permits certain fields to
be designated as signature fields. (Legal signature field entries can only be filled in with the image
of the current user's signature.)
Upon. logging into office talk , the user