Modern software design processes require collaboration among several specialists. These specialists have different perspectives, typically leading to many design conflicts. These conflicts, if not resolved early, create more expensive systems, delays in the development process, and compromises in the final product. Furthermore, current trends towards decentralization of operations and outsourcing of components add to the complexity of the software design process. The decentralization and the outsourcing add the differentials of time, space and organizational dimensions to an already complicated software development process. Hence, any successful software design process requires efficient collaboration mechanisms to reduce design conflict and improve design quality.

The collaboration between CICESE (Centro de Investigacion Cientifica y Estudios Superiores de Ensenada, Mexico) and the MIT CAIRO project [Pena-Mora et al 1996] include both research and educational thrusts. The research conducted at the two sites will be complimentary and will be a significant asset to both institutions. The design rationale capture (group memory) system developed at CICESE is being integrated with the CAIRO environment during the first year of the collaboration. Furthermore, the CICESE research effort is focused on the design of intelligent agent representatives that will automatically determine user preferences within the CAIRO system. Coupling the CICESE and MIT systems will provide an effective environment for distributed collaboration. The educational thrust involves the establishment of a joint collaborative software engineering course. This graduate course includes software engineering projects that involve strong interdependence among the student groups at MIT and CICESE. The class involves students both at the MEng and MS level, and will be conducted over two semesters in the traditional nine month academic year from September to May. Such a project will provide an extensive test bed for the computer supported collaboration methodologies and tools developed by both research efforts.

In January 1997 researchers at CICESE and MIT conducted a test meeting using the tools developed for this research project. Its success has motivated MIT to support for 9 months a student to look into the pedagogical aspects of the class. CICESE from its side has set aside funding for building an electronic classroom for the class to take place. Additionally, the class has drawn interest of participation from industry such as professionals from the Kajima and Shimizu Corporations of Japan, and the management staff of the Boston Central Artery and Tunnel Project. Moreover, faculty members in the Key Center for Design Computing at the University of Sydney, Australia are also interested in joining this international research initiative. In combining all these aspects, the class can be characterized as a bridge between research, education and industry by incorporating the research ideas on collaborative methodologies proposed for this research; by involving professionals from industry to participate in the class and bring their expertise and perspective on how the methodology works; and by allowing students to test all these concepts on a ìquasi-real worldî case study with their peers at other institutions.

Research

In the research thrust, the CAIRO (Collaborative Agent Interaction control and synchROnization) system is a distributed conferencing environment that allows individuals to interact over a computer network without the physical and temporal constraints experienced in a traditional meeting environment. The system provides mechanisms for automated facilitation and coordination to support the design change negotiation process in large-scale software engineering projects. Furthermore, the system has been designed to provide effective documentation of meeting proceedings to allow efficient retrieval of rationale regarding specific product or process design issues. CAIRO is the core of the Da Vinci Agent society Initiative at the Intelligent Engineering Systems Laboratory. [Pena-Mora, 1996]

CAIRO project research aims to bring together research on meeting and negotiation processes with distributed artificial intelligence concepts to explore methodologies for intelligent facilitation of distributed computer-supported meetings. The research involves the modeling of meeting control structures, group dynamics, as well as conversation elements exchanged in the meeting setting. The final goal of this research is the reduction of facilitation overhead and the removal of the same-place and same-time constraints of traditional meetings environment.

To remove the same-place constraint, a distributed conference architecture has been developed which provides synchronized multimedia communications among multiple participants on the Internet. A communication protocol has also been developed to ensure the synchronization of the multiple media channels. In addition, a standardized interface between the media devices and the synchronization/message system have been developed.

The same time constraint has been addressed through an analysis of the information exchanged in a design negotiation. The key feature necessary for eliminating this constraint is appropriate clustering and indexing of exchanged information. This provides a late or asynchronous participant with the necessary context to enable his/her interaction with the rest of the conferring group.

Another component that aids in relaxing the same time constraint is a group memory retention mechanism. In traditional meetings, an individual may retain minutes of the meeting that constitute the group memory retained. However, in some design meetings, minutes are not recorded or are not detailed enough to encompass the knowledge gained during a meeting. The CAIRO approach to support group memory retention involves a conversation indexing mechanism based on speech act conversation models and a goal oriented model for the categorization of design intent, recommendations and justifications.

Furthermore, in order to provide computer support for negotiation, the CAIRO system acts as a facilitator to maintain group focus and dissipate conflict. Effectively running software design negotiation meetings involves knowledge and experience in coordinating group effort. The CAIRO system provides a framework for encoding past experience in meeting facilitation and mediation into intelligent meeting control agents. These agents should provide ideal environments for group interaction by enforcing speech controls in accordance with the product or process design stage.

The distributed nature of CAIRO meetings allows participants to remain in their ideal working environments with access to all tools necessary for satisfying inquiries posed during the meeting.

Education

In the education thrust, the project will develop an integrated research laboratory and classroom alliance aimed at improving the skills of students and the understanding of modeling, design, implementation and collaboration in the subject domain. In this class, students at the Massachusetts Institute of Technology will collaborate with their peers at the Centro de Investigacion Cientifica y Estudios Superiores de Ensenada, located in Ensenada, Baja California, Mexico. This collaboration will take place between students from MIT and CICESE through a computerized collaborative communication system. Course content includes collaborative modeling, design and implementation of large scale software systems for civil engineering projects. The class will be taught jointly by the professors involved in both institutions trying to create an interactive, collaborative environment in which students and professors separated by distance feel as if the class was being held in one location. The curriculum has to be developed so that students can concentrate on learning the pedagogical concepts of the class without the distraction of dealing with unrelated technological issues. The curriculum also needs to address both the physical computerized communication systems, as well as the methodologies and group interaction protocols that students will adhere to during the class.

Details of the class

The class, which seeks professional diversity even from within its team of participating schools, will be sponsored by the Department of Civil and Environmental Engineering at MIT and the Department of Computer Science at CICESE. The class will involve senior undergraduate and first year graduate students, and will be conducted over two semesters in the traditional nine month academic year from September to May. Within that context, the class will have five specific educational objectives:

1. To teach the students the various strategic and tactical theories and practices of one-on-one and multi-party collaboration.
2. To help students learn how to maximize collaboration in a multi-disciplinary and multi-cultural team environment.
3. To assist students in improving software design and development skills applied to the subject domain, including improving their problem solving skills through the management of time, cost, and quality on the case study project.
4. To educate the groups on the technology associated with the distributed electronic environment through both instruction and hands-on use.
5. And finally, to place all of these elements within the framework of the large scale engineering domain by teaching the students about, and allowing them to experience, the different delivery systems, project structures, and contractual agreements that define relationships within that domain.

In order to measure the success of the class, an evaluation of the educational program will be developed. The main purposes of the evaluation are to provide feedback to faculty and industry professionals about the educational program (1) to continually improve its performance, and (2) to determine if the program is achieving its goals [Rossi and Freeman, 1993]. The evaluation will include both formative and summative assessments. The formative assessments are designed to monitor the program over time. The results of these assessments which will help to modify the program will assure that there is continuous quality improvement over time. The summative assessment conducted at the end of the academic year will help to determine whether the program should be continued.

Principal Investigatorís Current Research

The classroom/research laboratory effort represents one part of the larger Da Vinci Agent Society Initiative at the Intelligent Engineering Systems Laboratory of the Department of Civil and Environmental Engineering at MIT [Pena-Mora, 1996]. Da Vinci aims to enhance the collaboration and negotiation between professionals across distance and time through the use of a distributed infrastructure made up of several agents. Through these agents, which include a meeting structure and protocol interface [Pena-Mora et al, 1996], multiple reasoning mechanisms [Pena-Mora and Soibelman, 1996], decision rationale capture and retrieval system [Pena-Mora and Vadhavkar, 1997], game and negotiation theory protocol agent [Pena-Mora and Kennedy, 1996], and proactive cost and schedule management system [Pena-Mora and Chen, 1996], Da Vinci seeks to assist collaborators and negotiators in reaching consensus decisions that are technically efficient and organizationally sustainable.

Research Impact

The classroom/research laboratory environment envisioned, when utilized with carefully structured case studies and group exercises, offers an excellent test area for concepts that are critical to the future of the design and construction industry. The increasing globalization of the business is causing a high demand for communication solutions that help remove the restrictions of distance and time. While technologies such as teleconferencing seek to assist in such logistical problems, they can reach their self-imposed one-dimensional limitation, and fail to assist in more than just their singular plane of effectiveness. In contrast, the class provides a forum in which the impact of a methodology to address these issues can be tested within a controlled environment, so that maximum impact can be achieved before introducing the methodology to industry.

By stretching the boundaries of the classroom to include schools in other countries, the class can more accurately imitate the world into which students will move upon graduation. In that environment, students must learn to cope with multiple dimensions, including technical issues, language and cultural barriers, and competitive pressures on collaboration. The student who moves through the curriculum offered will need to deal with all of these issues, as well as complete the traditional classroom technical assignment work.

References

Pena-Mora, F., (1996), ìDa Vinci Initiative: Computer-Supported Negotiation Across Space and Time for the Life-Cycle Development of Sustainable Large-Integrated Engineering Systems in a Collaborative-Competitive, Domain-Dependent, and Strategy-Influenced Environment,î MIT Intelligent Engineering Systems Laboratory Technical Report Number IESL 96-03.

Pena-Mora, F., and Chen, H., (1996), ìCentral Artery / Third Harbor Tunnel Project Study for Better Management of Cost and Schedule Changes,î ASCE Third Congress on Computing in Civil Engineering, June 1996.

Pena-Mora, F., Hussein, K., and Sriram, D., (1996), CAIRO: ìA System for Facilitating Communication in a Distributed Collaborative Engineering Environment,î Computers in Industry, Elsevier Science Publishers, B.V. (North Holland).

Pena-Mora, F., and Kennedy, J., (1996), ìTheoretical Foundations in Computer Supported Negotiation,î ASCE Third Congress on Computing in Civil Engineering, June, 1996.

Pena-Mora, F., and Soibelman, L., (1996), ìA Geographically Distributed Multi-Reasoning Mechanism for Change Negotiation Management on Large Scale Engineering Systems,î CIB W78 Workshop, Bled, Slovenia.

Pena-Mora, F., and Vadhavkar, S.S., (1997), ìAugmenting Design Patterns with Design Rationale,î to appear in Artificial Intelligence for Engineering Design, Analysis and Manufacturing (AIEDAM) Special Issue on Design Rationale, May, 1997.

Rossi, Peter H. and Freeman, Howard E. (1993), Evaluation -- A Systematic Approach. (5th. Edition). Newbury Park, CA: Sage.