| Structure and Action in Distributed Organizations | | BIBA | 1-10 | |
| Peter de Jong | |||
| An organization's structure specifies the formal relationships between the objects which comprise the organization. An organization's action is initiated by the communication of messages between the organizational objects. This paper describes some of the organization's structural relationships and message passing patterns. It also explores the relationship between structure and action. An organization is described using the Ubik system. Within this system, the organizational representation is specified using a distributed semantic net consisting of objects called configurators; the message passing action is based on the Actor model of parallel and distributed processing. The structure is related to the action by the use of special configurators called constructors, tapeworms, questers and sponsors. These configurators build, maintain, monitor and reason over the distributed networks. | |||
| A Logically Distributed Approach for Structuring Office Systems | | BIBA | 11-20 | |
| Panayiotis K. Chrysanthis; David Stemple; Krithi Ramamritham | |||
| An object-oriented office model is presented. It used the object taxonomy of Booch featuring object classes based on calling patterns. Our model is motivated by that of Woo and Lochovsky, but has a number of differences, especially in the definition of the object classes and in the treatment of nested objects. An office application, setting a meeting among office workers, is defined in terms the office model. | |||
| A Computational Model for Organizations of Cooperating Intelligent Agents | | BIBA | 21-30 | |
| Alan H. Bond | |||
| A computational model is developed, which captures properties of
relationship and organization in sets of distributed intelligent agents. The
model is inspired by sociological theory, and used centrally a concept of
commitment. Symbolic interactionist concepts in sociology are reviewed from
the perspective of Distributed Artificial Intelligence. This includes the work
of Anselm Strauss and Elihu Gerson.
A formulation of the concept of commitment is proposed. The concept is propagated into notions of agent and organization. A formulation and computational model of the basic notion of commitment is given, which represents commitments as mutually agreed constraints on action, belief and world state. The issues of agent integrity, plans and resources, and their representation, are then discussed, and computational mechanisms defined. An organization is defined as a set of agents with consistent mutual commitments. A mechanism is defined to check and enforce the consistency of these commitments. | |||
| Answer Garden: A Tool for Growing Organizational Memory | | BIBA | 31-39 | |
| Mark S. Ackerman; Thomas W. Malone | |||
| Answer Garden allows organizations to develop databases of commonly asked
questions that grow "organically" as new questions arise and are answered. It
is designed to help in situations (such as field service organizations and
customer "hot lines") where there is a continuing stream of questions, many of
which occur over and over, but some of which the organization has never seen
before. The system includes a branching network of diagnostic questions that
helps users find the answers they want. If the answer is not present, the
system automatically sends the question to the appropriate expert, and the
answer is returned to the user as well as inserted into branching network.
Experts can also modify this network in response to users' problems.
Our initial Answer Garden database contains questions and answers about how to use the X Window System. | |||
| Using Latent Semantic Indexing for Information Filtering | | BIBA | 40-47 | |
| Peter W. Foltz | |||
| Latent Semantic Indexing (LSI) is an information retrieval method that organizes information into a semantic structure that takes advantage of some of the implicit higher-order associations of words with text objects. The resulting structure reflects the major associative patterns in the data while ignoring some of the smaller variations that may be due to idiosyncrasies in the word usage of individual documents. This permits retrieval based on the "latent" semantic content of the documents rather than just on keyword matches. This paper evaluates using LSI for filtering information such as Netnews articles based on a model of user preferences for articles. User judged articles on how interesting they were and based on these judgements, LSI predicted whether new articles would be judged interesting. LSI improved prediction performance over keyword matching an average of 13% and showed a 26% improvement in precision over presenting articles in the order received. The results indicate that user preferences for articles tend to cluster based on the semantic similarities between articles. | |||
| HyperMail: Treating Electronic Mail as Literature | | BIBA | 48-54 | |
| Richard K. Belew; John Rentzepis | |||
| The huge flux of electronic information surrounding the modern researcher requires new tools for its management. HyperMail is a simple tool built on the HyperCard environment for use as a retrospective aid, for relatively stable collections of Email messages concerning a relatively focused topic. By allowing users to swiftly locate, examine and compare messages across authors and across time, the original conversation can become a rich source of subsequent study. Our goal is to treat Email as a form of literature, worthy of the same preservation and augmentation that is typical of traditional printed media. | |||
| Video Workplaces | | BIB | 55 | |
| Robert W. Root | |||
| Computerization and the Quality of Working Life: The Role of Control | | BIBA | 56-68 | |
| Robert E. Kraut; Susan Dumais | |||
| Does the degree of control people have over their work environments moderate the impact of computerization in organizations? This paper addresses this question with a quantitative case study and a national survey. Results of the case study show that clerical workers with more control over the day-to-day aspects of their jobs had better experiences with computing. However, the national sample provides no evidence to support the control hypothesis. In this study, women using computers had slightly worse jobs and more stress-related symptoms than those not using computers, and these negative associations with computerization weren't reduced if they had more control. It is likely that factors other than control had important impact on how computering technology was designed, deployed and used. Good design and good implementation, however they come about, may be more important that the users' control per se. | |||
| The Work Group Manager's Role in Developing Computing Infrastructure | | BIBA | 69-78 | |
| Tom Jewett; Rob Kling | |||
| Much of the literature about the management of computing has focused on the role of computing professionals and upper level managers. We are interested in the managers of groups which use computing to support their work, for whom computer use and computing management are only adjuncts to other duties. Even when managers do not personally have high levels of computing skills or even interest in computing, they must manage the use of computing in their workplaces and provide informal continuing support for employees' use of computers. In this paper, we describe two work groups in which different management styles have shaped the integration of new computing systems. We find that the way in which these managers approach computing problems is substantially similar to their approach to non-computing aspects of the workplace. We also find the effects of the manager's approach to be more subtle and complex than "user involvement" studies have indicated. We found that managers of more dynamic computerization projects involve their subordinates in the design of work, not just the design of systems. And they help build the infrastructure of their work groups, as well as an infrastructure for computing. | |||
| Inter-Organization Computer Networks: Indications of Shifts in Interdependence | | BIBA | 79-88 | |
| Paul Hart; Deborah Estrin | |||
| As firms increasingly adopt inter-organization computer networks (IONs) to improve coordination, researchers must be concerned about the long term impact of IONs on organizational relationships. This paper reports on an exploratory study of the use of IONs in design and manufacturing activities in the semiconductor industry. We identify the potential interactions between firms that can be facilitated by IONs, and focus on the implications for customer and producer interdependence. Our analysis suggests that the long term impacts of IONs are not technologically determined, and that their use ought to be regarded differently than those of other media. | |||
| SACT: A Tool for Automating Semi-Structured Organizational Communication | | BIBA | 89-98 | |
| Carson C. Woo | |||
| To cope with a highly specialized environment, organizational workers
specialize in different areas of knowledge (e.g., taxation). Communication
plays a very important role in performing organizational tasks that require
multiple specializations. In fact, research data shows that managers and
professionals spend about 70-80% of their time in communication. To alleviate
the time organizational workers spend in communication, some of their
communication should be automated. In this paper, we propose a communication
tool for this purpose. This tool can be used essentially in the same way that
an exert system shell is used.
The function of this tool is to collect data needed for performing a task in an open system. In order to do so, the two communicating parties must understand what each other wants. The speech act taxonomy developed by Searle, then modified by Reiss, is useful for this purpose. However, their taxonomy is intended for general communication. We have to select the appropriate subset of it that is suitable for automating organizational communication and have to operationalize the theory behind it into a computer-based communication tool. | |||
| A Conversation Model for Resolving Conflicts among Distributed Office Activities | | BIBA | 99-108 | |
| Frank von Martial | |||
| In many cases, office workers perform their tasks based on certain plans. In a cooperative environment, where many individuals establish their plans, conflicts between these plans will inevitably arise. A reason for conflicts between different plans is, for instance, the competition for shared resources such as special equipment or a reservation for a meeting room. In such cases, plans may prevent each other from being executed as intended, which reduces the overall performance of an organization. There are two main problems tackled by our approach: First, to detect potential conflicts between several individual plans as early as possible and, second, to relieve the actors in an organization as much as possible from their burden to settle conflicts. In our approach, conflicts are settled by an agreement among the agents in conflict. The agreement is reached by a computer-mediated conversation (negotiation) between the agents. One key factor for the detection of conflicts among individually created plans and for guiding the conflict resolution process are the temporal relationships between actions in conflict. The role of temporal constraints for coordinating distributed activities will be described in detail. | |||
| Support of Cooperative Work by Electronic Circulation Folders | | BIBA | 109-117 | |
| B. Karbe; N. Ramsperger; P. Weiss | |||
| ProMInanD's migration system deals with cooperative office work on different types of office tasks which consist of steps to be carried out by persons playing office roles. This typical kind of cooperation is supported by ProMInanD's Electronic Circulation Folders (ECF). Instantiations of task related types of ECF's migrate automatically through an office organization. In the contents of an ECF the office workers' contribution to a task are kept as arbitrary kinds of documents. The migration specification of an ECF describes the steps to be performed, their possible sequences and the roles which have to be played for their performance. However, in many cases neither the number of steps nor the persons involved in an office task are known in advance. Often, there is a need for deviations from predefined migration routes and for exception handling. Also, there are tasks which ask for completely unformalized migrations. Moreover, in all these cases, non-deterministic human behavior has to be taken into account. These problems are all solved by ECF's. In this paper their design, processing and implementation are considered. | |||
| An Approach to Integrated Office Document Processing & Management | | BIBA | 118-122 | |
| Nelson M. Mattos; Bernhard Mitschang; Andreas Dengel; Rainer Bleisinger | |||
| We propose an approach towards an integrated document processing and management system that has the intention to capture essentially freely structured documents, like those typically used in the office domain. The document analysis system ANASTASIL is capable to reveal the structure as well as the contents of complex paper documents. Moreover, it facilitates the handling of the containing information. Analyzed documents are stored in the management system KRISYS that is connected to several different subsequent services. The described system can be considered as an ideal extension of the human clerk, making his tasks in information processing easier. The symbolic representation of the analysis results allow an easy transformation in a given international standard, e.g., ODA/ODIF or SGML, and to interchange it via global network. | |||
| An Interface for the Acquisition and Display of Office Procedures | | BIBA | 123-130 | |
| Dirk E. Mahling; W. Bruce Croft | |||
| A central problem in the design of intelligent office systems is the acquisition of knowledge about office procedures. In this paper we describe a graphical interface for the acquisition and display of office procedures from a goal- and plan-based perspective. The DACRON interface is based on a model of the office workers' view of work. DACRON supports the acquisition of plan knowledge by providing graphical representations of domain entities from the users' point of view. It allows the display and view of office procedures graphically. An experimental usability study, involving more than twenty subjects, shows that DACRON can be used to acquire plan knowledge and give relevant advice. | |||
| A Set-to-Set Linking Strategy for Hypertext Systems | | BIBA | 131-135 | |
| Yoshinori Hara; Yutaka Kasahara | |||
| This paper proposes a set-to-set linking model with the purpose of improving the trade-off between link organizing/updating costs and link utilization benefits. This model is based on a value-document linking structure in a multi-facet schema, this is, a schema-instance linking structure which helps to decrease the total number of links. It provides a series of traversing operations to create dynamically the related document set step-by-step in accord with certain specified values. A prototype fine arts encyclopedia demonstrates the usefulness of this linking strategy. | |||
| Coordinating Distributed Action via Agent Voting | | BIBA | 136-141 | |
| Arnold B. Urken | |||
| This paper investigates the use of voting policies to coordinate routing decisions in a phone network. Although the social metaphor of voting has been applied to network coordination decision tasks, this study presents the first operational example of a vote-theoretic group decision support system (GDSS) for nodes. The experimental evidence shows that a collective choice voting policy dominates a policy of individual, hierarchical voting in minimizing movement toward system saturation and promoting load balancing. This result provides a basis for using voting policies to create more complex self-correcting networks. | |||
| Design and Implementation of a Shared Workspace by Integrating Individual Workspaces | | BIBA | 142-146 | |
| Masaaki Ohkubo; Hiroshi Ishii | |||
| This paper proposes "TeamWorkStation" (TWS) as an approach to an effective shared workspace for the support of remote collaboration. There are three key design objectives in TWS: integration of virtual and actual workspaces, a simultaneously-accessible shared drawing surface, and smooth transition between individual workspaces and a shared workspace. To achieve these objectives, images of computers and/or paper are overlayed so that information and images from both are effectively combined and distributed to the group members. | |||
| Implication of the Guaranteed, Reliable, Secure Broadcast Technology to Office Information Systems | | BIBA | 147-151 | |
| L. C. N. Tseung; K. C. Yu | |||
| Guaranteed, Reliable, Secure Broadcast (GRSB) - is a protocol that provides reliable and secure broadcast/multicast communications. Four logical nodes are enforced in the network - a Central Retransmitter, a Security Controller, a Designated Acknowledger, a (many when needed) Playback Recorder(s). Through the coordinated service of the four nodes, every user node can be guaranteed to receive all broadcast messages in a secure manner and in the correct temporal order. This paper focuses on the implication of GRSB to office information systems. How GRSB coherently supports several progressive functional requirements, from small number of user nodes to complex, but integrated functions, is elaborated. | |||
| LAN Based Real Time Audio-Data Systems | | BIBA | 152-157 | |
| L. F. G. Soares; S. L. Martins; T. L. P. Bastos; N. R. Ribeiro; R. C. S. Cordeiro | |||
| One of the greatest demands for better communication services comes from the
office automation area. The functions related to electronic mail, telephony,
and teleconferencing are among these services, due to advantages it creates,
such as reduction of cost and time.
LANBRETAS is LAN-based real-time audio/data system that is being developed by the Catholic University of Rio de Janeiro and the Rio Scientific Center of IBM-Brasil and is the subject of this paper. The first part of the paper introduces the motivations to implement the system as well as the implementation environment. The second part presents the system. At first the electronic mail system LANBRETA-DMS, then the telephony system LANBRETA-DTS, and at last, the teleconferencing system LANBRETA-GCS. | |||
| PENDS: An Approach to Modeling and Retrieving 'Pending' Information | | BIBA | 158-162 | |
| Ramesh Subramanian | |||
| Recently there has been tremendous interest in temporal data modeling. The scope of most proposed temporal models has been restricted to historical, or past data. We believe that a complete temporal database must deal with historical data, as well as data relating to the future (pending data). In this paper we expand the previous work on temporal modeling by introducing the "pending" dimension. We also specify a (pending) query and update language, which is an extension of the GORDAS ER query language. We use an inventory management environment and show that our model and query language can handle complex pending situations. | |||
| Shared Hardware: A Novel Technology for Computer Support of Face to Face Meetings | | BIBA | 163-168 | |
| David Halonen; Marjorie Horton; Robert Kass; Paul Scott | |||
| This paper describes the Capture Lab, a computer supported meeting room in sue at the EDS Center for Machine Intelligence since late 1987. Most computer supported meeting environments implement a simple hardware approach, where a single computer controlled by a trained technician or facilitator is used, or else adopt a groupware approach, where each user has a personal machine and special-purpose software is used to support group activities. In contrast, the Capture Lab implements a shared hardware approach, in which each meeting participant has a personal computer, but can easily access a shared public computer as well. We discuss the advantages and limitations of this approach, based on our observations of how the room is used, and compare it to the simple hardware and groupware approaches. | |||
| Xsketch: A Multi-User Sketching Tool for X11 | | BIBA | 169-173 | |
| Jeffrey J. Lee | |||
| This is a report of work-in-progress on a multi-user sketching tool. The intent behind the project is to produce a simple tool for use in multi-user design and brainstorming sessions. The tool, xsketch, is intended to function across existing internetwork facilities and run on a variety of platforms. This paper covers the requirements and design goals for the tool. Details of a prototype implementation are given along with a list of known problems, shortcomings, and possible extensions. | |||
| Toward a Definition of Voice Documents | | BIBA | 174-183 | |
| Michael J. Muller; Jane E. Daniel | |||
| This paper develops a definition of a voice document as a combination of information, structure, and affordances (or user-executable actions or utterances) for use in a voice-I/O hypermedia system. Voice documents in our experimental prototype environment, HyperPhone, are finely-grained hypermedia objects with rich interconnections of literal and virtual links, and with certain well-defined local structures. We explore issues related to navigating, managing, and authoring such documents, and outline a set of questions for further work. | |||
| Commune: A Shared Drawing Surface | | BIBA | 184-192 | |
| Sara A. Bly; Scott L. Minneman | |||
| Careful observation of small-group design sessions suggests that the process of creating, referring to, and using drawings may be as important to the design process as the drawings themselves. Based on studies of the uses of drawing spaces, Commune was developed to allow designers working remotely to share a drawing surface and to engage in many of the interactions available in conventional face-to-face situations. The design of Commune makes marks and 2-dimensional cursor gestures visible simultaneously to all users, allows rapid transitions among drawing, writing, and gesturing, and provides a shared space with actions from multiple users occurring simultaneously. These capabilities support natural uses of the drawing surface during the interaction: the ability to interact on each other's marks, to emphasize talk with marks and gestures, to reference previous illustrations and concepts, and to interweave talk and drawing actions fluidly. | |||
| Evaluating Computer-Based Tools in Organizations | | BIBA | 193 | |
| Bruce Croft; Nicholas Belkin; Tora Bikson; Thomas Malone; Gary Marchionini | |||
| Most of the computer systems that are proposed and implemented in the OIS area are very difficult to evaluate, both in terms of the effectiveness of the functions they perform and their impact on organizations. The time and space efficiency of a tool is readily measured, but this may be only a small part of a tool's "performance". Researchers in areas such as CSCW, Information Retrieval, Hypertext, and User Interfaces have been forced to confront the issue of evaluation in order to compare systems and determine when progress has been made. There is, however, still much debate about the correct methodologies for evaluating such complex, interactive systems. System developers are also concerned with measuring the "usability" of their systems and designing systems iteratively based on user feedback. It is not clear whether these approaches compete with, or are complementary to more traditional experimental approaches. The panelists come from a variety of backgrounds and have each had experience evaluating complex systems. By comparing their experiences and the techniques used in their areas, we hope to be able to make some progress on the issue of how systems can be evaluated and compared. | |||
| A Micro-Organizational Model for Supporting Knowledge Migration | | BIBA | 194-204 | |
| Frederick H. Lochovsky; Carson C. Woo; Lawrence J. Williams | |||
| One of an organization's assets is the knowledge it has for carrying out its activities in an acceptable way. However, due to the constantly changing environment in which operates, its knowledge must evolve in order for it to survive and to stay competitive. For this reason, it is inappropriate to hard-code organizational knowledge into a computer-based organizational support system. Supporting the evolution of organizational knowledge, in general, is a very difficult problem because the domain knowledge needed to support it is open-ended. However, certain forms of organizational knowledge evolution, where the knowledge needed for the evolution already exists in some form in the organization, can be supported. We call this type of knowledge evolution knowledge migration. To evolve, in this case, is to recognize the need for change, and to rearrange or acquire the knowledge (possibly located in several different locations). In this paper, we discuss an approach for supporting this type of knowledge evolution. | |||
| Objects with Roles | | BIBAK | 205-215 | |
| Barbara Pernici | |||
| The use of object-oriented conceptual models for modeling office
applications and information systems is discussed. A model for describing
object behavior based on the concept of role is presented. Roles allow one to
describe different perspectives for object evolution. For each role, relevant
characteristics such as role properties, role states, messages, and role-state
transition rules and constraints are defined. The implications of considering
several roles in parallel for an object are discussed, and a classification of
possible role interactions is given. Keywords: Office modeling, Object-oriented conceptual modeling, Roles; Office modeling, Object-oriented conceptual modeling, Roles | |||
| An Object-Oriented Data Model for Distributed Office Applications | | BIBA | 216-226 | |
| E. Bertino; M. Negri; G. Pelagatti; L. Sbattella | |||
| The object-oriented paradigm is becoming very popular for database applications and several object-oriented DBMSs have been developed. A basic notion in this paradigm is the inheritance hierarchy that allows the users to define objects and the associated operations starting from already defined objects. However, in database applications the inheritance hierarchy must provide a conceptual modeling function, in addition to the re-usability function. Another important requirement is to provide support for data distribution in (possibly) heterogeneous environments. This means that object implementation may differ depending on the object location. This paper presents a model that decouples these two aspects, modeling vs implementation, by using the concept of abstract and implementation classes. An abstract class specifies properties and methods for a set of similar objects, like in other object-oriented data models. An abstract class is however independent of the object implementation and location. An implementation class defines the implementation of an abstract class. In our model an abstract class may have several implementations. This allows the user to provide different implementations for the same set of objects, without requiring the objects to change class. | |||
| Sharing Views and Interactions with Single-User Applications | | BIBAK | 227-237 | |
| Saul Greenberg | |||
| Although work is frequently collaborative, most computer-based activities
revolve around software packages designed to be used by one person at a time.
To get around this, people working together often talk and gesture around a
computer screen, perhaps taking turns interacting with the running
"single-user" application by passing the keyboard around. However, it is
technically possible to share these unaltered applications -- even though they
were originally designed for a single user only -- across physically different
workstations through special view-sharing software. Each person sees the same
image of the running application on their own screen, and has an opportunity to
interact with it by talking turns. This paper discusses the various roles and
responsibilities of the view-sharing software that must be considered during
its design and evaluation: view management, floor control, conference
registration by participants, and handling of meta-level communications. A
brief survey of existing shared view systems is provided for background. Keywords: Computer supported cooperative work, Real time conferencing, Shared window
systems | |||
| A Comparison of Application Sharing Mechanisms in Real-Time Desktop Conferencing Systems | | BIBA | 238-248 | |
| S. R. Ahuja; J. R. Ensor; S. E. Lucco | |||
| Desktop conferencing is a term used to describe real-time, computer-based conferences in which users may share data through their personal computers. In these conferences, the participants may access user-level programs, called application programs, which produce common displays (screens or windows) on their computers. Because each participant may give input to the application program and sees its resulting output as though the program were executing on his or her local computer, these applications retain their own look and feel as they form a shared environment for the conference. We compare some methods of sharing application programs during real-time, computer-based conferencing. In particular, we have explored different methods in three versions of Rapport, a multimedia conferencing system. We examine the qualitative and quantitative differences of these implementations using some typical shared applications and find that one method possesses good semantic characteristics as well as good performance in several network environments. | |||
| Replicated Architectures for Shared Window Systems: A Critique | | BIBA | 249-260 | |
| J. Chris Lauwers; Thomas A. Joseph; Keith A. Lantz; Allyn L. Romanow | |||
| Replicating applications in a shared window environment can significantly improve the performance of the resulting system. Compared to a completely centralized approach, a replicated architecture offers superior response time and reduces network load. To date, however, these advantages have been overshadowed by the equally significant synchronization problems associated with replication. In this paper we document these problems and show that the most frequent synchronization problems can be solved without changing existing software. We further indicate how some of the limitations of the resulting system can be removed by making applications or system servers collaboration-aware. Finally, we point out where general system support is needed to address the remaining deficiencies. | |||
| Office Documents on a Database Kernel -- Filing, Retrieval, and Archiving -- | | BIBA | 261-270 | |
| P. Zabback; H.-B. Paul; U. Deppisch | |||
| One of the main component of integrated office systems is the large central filing system. It efficiently stores, retrieves and searches office documents containing text, images, graphics, data and voice. We propose to implement a filing system on top of the Darmstadt database system (DASDBS), which is designed as a data management kernel for both standard and non-standard applications. This paper investigates the choice of appropriate storage structures for the filing system objects and the realization of the system by using the kernel. Furthermore we discuss the efficient retrieval support of office objects by signatures and introduce a new archival approach by using storage media like optical disks. | |||
| SaTellite: A Visualization and Navigation Tool for Hypermedia | | BIBA | 271-280 | |
| Xavier Pintado; Dennis Tsichritzis | |||
| SaTellite is a visualization and navigation tool for a hypermedia system.
It is based on the concept of affinity between objects; that is, a relationship
with an associated intensity. The user is presented with a two dimensional map
that provides a view of the hypermedia environment where objects lying close
together have a greater affinity than those lying further apart. The system
provides different views by allowing modification of the underlying measure of
affinity. The system is also able to track dynamically the evolution of the
objects' relationships.
Based on the affinity concept, we develop new dynamic presentation techniques that do not depend on the explicit display of links between the nodes of the graph. The dynamic layout algorithm that we present at the end of the paper is based on these techniques and it allows for the display of rapidly changing relationships between objects. | |||
| A Knowledge-Based Approach to the Design of Document-Based Retrieval Systems | | BIBA | 281-290 | |
| Hsinchun Chen; Vasant Dhar | |||
| This article presents a knowledge-based approach to the design of document-based retrieval systems. We conducted two empirical studies investigating the users' behavior using an online catalog. The studies revealed a range of knowledge elements which are necessary for performing a successful search. We proposed a semantic network based representation to capture these knowledge elements. The findings we derived from our empirical studies were used to construct a knowledge-based retrieval system. We performed a laboratory experiment to evaluate the search performance of our system. The experiment showed that our system out-performed a conventional retrieval system in recall and user satisfaction. The implications of our study to the design of document-based retrieval systems are also discussed in this article. | |||