Although graphic representations have proven to be of value in computer-aided support and have received much attention in both research and practice (Goldschmidt, 1991; Goel, 1995; Achten, 1997; Do, 2002), linguistic representations presently do not significantly contribute to improve the information handling related to the computer support of a design product. During its life cycle, engineers and designers make many representations of a product. The information and knowledge used to create the product are usually represented visually in sketches, models, (technical) drawings, and images. Linguistic information is complementary to graphic information and essential to create the corporate memory of products. Linguistic information (i.e., the use of words, abbreviations, vocal comments, annotations, notes, and reports) creates meaningful information for designers and engineers as well as for computers (Segers, 2004; Juchmes et al., 2005). Captions, plain text, and keyword indexing are now common to support the communication between design actors (Lawson & Loke, 1997; Wong & Kvan, 1999; Heylighen, 2001; Boujut, 2003). Nevertheless, it is currently scarcely used to its full potential in design, maintenance, and manufacturing.

How can we characterize the promulgation of computing in design? At the outset, computing was conceived predominantly in terms of design automation. It was quickly realized that this medium offered a great deal more than automation, and computing rapidly become a mode of conception for designing. At the juncture of art, design, and computing, however, we recognize computing as becoming loaded with cultural meanings that are enacted through designed works, and designed works that reinterpret a range of assumptions about computing. Design and computing are intervening and lending each other ever-accruing layers of possibilities. The arena of design and computing is producing a new object of knowledge from which they effect: a mode of inquiry, a form of critique, a constitution of practices of subjectivities, and an articulation of objective and subjective investigations. The four papers selected for publication in this Special Issue triangulate these perspectives, providing a new vision for the direction of design and computing and emerging research themes.

The AI EDAM journal published its first issue in 1987. Now, amazingly, 20 years later, this issue starts the journal's third decade.

Readily available and affordable technologies such as the Internet, groupware, and Web conferencing mean that sharing information and data within teams is simple and affordable. However, many small- and medium-sized enterprises (SMEs) struggle to implement or perform distributed collaborative design effectively or even at all. As part of the extended design team of large multinational companies it is not uncommon for SMEs to have collaborative working tools and practice imposed on them to meet the requirements of the multinational. However, many SMEs need to develop their own working practices to support effective, collaborative team design within their own organization or their extended design team. Through a series of case studies, this paper describes how a typical SME achieved successful distributed team design within their organization. A “strategy for effective distributed team design” encompassing the processes, methods, and tools developed and implemented within the company to achieve this success, is presented. In total, four live case studies, spanning a 2-year period, are described; two initial studies focus on current distributed design team practice clearly highlighting issues and areas for improvement, leading to the development of processes, methods, and tools to support distributed collaborative team design. A strategy for effective distributed team design encapsulating these processes methods and tools is presented together with its evaluation through two further live industrial case studies. The case studies themselves, together with the processes, methods, and tools developed by this company, could be adopted by other SMEs directly to achieve the same success. Generic and transferable findings drawn from this study aimed at helping others achieve this success form the conclusion of the paper.

In 1997, in a farewell message in Volume 11, Issue 1, of AI EDAM I recounted how the journal came to be established in 1987 and how I became the Founding Editor. At that point I was handing over the reigns to a new Editor, Bill Birmingham. Five years later, Bill turned the enterprise over to our current Editor, Dave Brown. Now, a total of 20 years has passed. Twenty years, a goodly length of time in our current world. How have we done?

This paper examines the use of language, specifically verbs, as stimuli for concept generation. Because language has been shown to be important to the reasoning process in general as well as to specific reasoning processes that are central to the design process, we are investigating the relationship between language and conceptual design. The use of language to facilitate different stages of the design process has been investigated in the past. Our previous work, and the work of others, showed that ideas produced can be expressed through related hierarchical lexical relationships, so we investigated the use of verbs within these hierarchical relationships as stimuli for ideas. Participants were provided with four problems and related verb stimuli, and asked to develop concepts using the stimuli provided. The stimuli sets were generated by exploring verb hierarchies based on functional words from the problem statements. We found that participants were most successful when using lower level (more specific) verbs as stimuli, and often higher level general verbs were only used successfully in conjunction with lower level verbs. We also observed that intransitive verbs (verbs that cannot take a direct object) were less likely to be used successfully in the development of concepts. Overall, we found that the verb chosen as stimulus by the participant directly affects the success and the type of concept developed.

This paper describes research carried out to develop a parametric urban shape grammar for the Zaouiat Lakhdar quarter of the Medina of Marrakech in Morocco. The goal is to create the basis for a system that could capture some features of the existing urban fabric and apply them in contemporary urban planning and architectural design. The methodology used is described, from the initial historical analysis and fieldwork to the identification of three subgrammars necessary to encode the complexity of the urban preexistences: the urban grammar, the negotiation grammar, and the housing grammar. Top-down and bottom-up approaches to grammar design are analyzed and compared. The bottom-up urban grammar developed is then described, and a hand derivation of the existing urban fabric is proposed. Visual, symbolic, and tagged computer implementations of shape grammars are briefly discussed and a novel design generated by the tagged interpreter is presented.

A natural question is why AI in design? Although the design applications written about in the journal vary widely, the common thread is that researchers use AI techniques to implement their ideas. The use of AI techniques for design applications, at least when AI EDAM was started, was partially a reaction against the predominant design methods based on some form of optimization. Knowledge-based techniques, particularly rule-based systems of various sorts, were very popular. One of the draws of these methods, I believe, was their ability to represent knowledge that is hard or awkward to represent in traditional optimization frameworks. This mirrors my experience: at the time, I was working in configuration with components that had a large number compatibility and resource constraints. Although many constraints could be represented in mixed integer linear programming systems, it was not easy to conceptualize, write, and most importantly, maintain the constraints in those systems.

Affective engineering is being increasingly used to describe a systematic approach to the analysis of consumer reactions to candidate designs. It has evolved from Kansei engineering, which has reported improvements in products such as cars, electronics, and food. The method includes a semantic differential experiment rating candidate designs against bipolar adjectives (e.g., attractive–not attractive, traditional–not traditional). The results are statistically analyzed to identify correlations between design features and consumer reactions to inform future product developments. A number of key challenges emerge from this process. Clearly, suitable designs must be available to cover all design possibilities. However, it is also paramount that the best adjectives are used to reflect the judgments that participants might want to make. The current adjective selection process is unsystematic, and could potentially miss key concepts. Poor adjective choices can result in problems such as misinterpretation of an experimental question, clustering of results around a particular response, and participants' confusion from unfamiliar adjectives that can be difficult to consider in the required context (e.g., is this wristwatch “oppressive”?). This paper describes an artificial intelligence supported process that ensures adjectives with appropriate levels of precision and recall are developed and presented to participants (and thus addressing problems above) in an affective engineering study in the context of branded consumer goods. We illustrate our description of the entire concept expansion and reduction process by means of an industrial case study in which participants were asked to evaluate different designs of packaging for a laundry product. The paper concludes by describing the important advantages that can be gained by the new approach in comparison with previous approaches to the selection of consumer focused adjectives.

Structural optimization is usually handled by iterative methods requiring repeated samples of a physics-based model, but this process can be computationally demanding. Given a set of previously optimized structures of the same topology, this paper uses inductive learning to replace this optimization process entirely by deriving a function that directly maps any given load to an optimal geometry. A support vector machine is trained to determine the optimal geometry of individual modules of a space frame structure given a specified load condition. Structures produced by learning are compared against those found by a standard gradient descent optimization, both as individual modules and then as a composite structure. The primary motivation for this is speed, and results show the process is highly efficient for cases in which similar optimizations must be performed repeatedly. The function learned by the algorithm can approximate the result of optimization very closely after sufficient training, and has also been found effective at generalizing the underlying optima to produce structures that perform better than those found by standard iterative methods.

It is generally known that architectural practice relies heavily on the interactions between architects and other professionals. However, during their formal education, most students attending architecture schools, and engineering schools for that matter, get very little (if any) exposure to decision making in conditions that involve expertise and/or worldviews beyond those reflected and valued by their own discipline. In the past 10 years, a project-based learning initiative was developed between the University of California, Berkeley, and Stanford University in an international context involving several other universities around the world. Throughout this experience, we have identified several issues that have shown to be crucial to these interactions. This paper elaborates on three key issues: improvement of communication skills, empowerment through developing strategies of leadership, and recognition of own and others' worldviews. We also make the case to include experiential educational situations that can introduce these aspects into the academic curricula of architecture and engineering schools.

Successful realization of large-scale product development programs is challenging because of complex product and process dependencies and complicated team interactions. Proficient teamwork is underpinned by knowledge of the manner in which tasks performed by different design participants fit together to create an effective whole. Based on an extensive industrial case study with a diesel engine company, this paper first argues that the overview and experience of senior designers play an important part in supporting teamwork by coordinating activities and facilitating proactive communication across large project teams. As experts move on and novices or contractors are hired, problems are likely to occur as tacit overview knowledge is lost. If informal, overview-driven processes break down, the risk of costly oversights will increase, and greater management overhead will be required to realize successful product designs. Existing process models provide a means to express the connectivity between tasks and components thus to compensate partially for the loss of tacit overview. This paper proposes the use of design confidence, a metric that reflects the designer's belief in the maturity of a particular design parameter at a given point in the process, to address the limitations of existing models. The applicability of confidence-based design models in providing overview, as well as their shortcomings, will be demonstrated through the example of a diesel engine design process. Confidence can be used to make overview knowledge explicit and convey additional information about the design artifact, thereby informing communication and negotiation between teams.

Early research in artificial intelligence (AI) in design was “AI in the large.” Artificial intelligence promised a revolutionary approach to assisting or automating design processes, separate from other computational systems. As the field matures, “AI in the small” has the most potential. Rather than seeing large AI programs, we see the results of research in AI in design providing the glue that holds larger systems together using reasoning systems that represent or manage processes, information, and interaction devices that use conventional procedural programming; effectively blurring the boundaries between AI and non-AI. Alternatively, we see research in AI in design as playing a role in blurring the boundaries between the physical and the virtual.

This paper explores using Shannon's entropy of information to measure linkographs of 12 design sessions that involved six architects in two different experimental conditions. The aim is to find a quantitative tool to interpret the linkographs. This study examines if the differences in the design processes and the design outcomes can be reflected in the entropic interpretations. The results show that the overall entropy of one design condition is slightly higher than the other. Further, there are indications that the change of entropy might reflect design outcomes.

Because of the increasing complexity of products and the design process, as well as the popularity of computer-aided documentation tools, the number of electronic and textual design documents being generated has exploded. The availability of such extensive document resources has created new challenges and opportunities for research. These include improving design information retrieval to achieve a more coherent environment for design exploration, learning, and reuse. One critical issue is related to the construction of a structured representation for indexing design documents that record engineers' ideas and reasoning processes for a specific design. This representation should explicitly and accurately capture the important design concepts as well as the relationships between these concepts so that engineers can locate their documents of interest with less effort. For design information retrieval, we propose to use shallow natural language processing and domain-specific design ontology to automatically construct a structured and semantics-based representation from unstructured design documents. The design concepts and relationships of the representation are recognized from the document based on the identified linguistic patterns. The recognized concepts and relationships are joined to form a concept graph. The integration of these concept graphs builds an application-specific design ontology, which can be seen as the structured representation of the content of the corporate document repository, as well as an automatically populated knowledge base from previous designs. To improve the performance of design information retrieval, we have developed ontology-based query processing, where users' requests are interpreted based on their domain-specific meanings. Our approach contrasts with the traditionally used keyword-based search. An experiment to test the retrieval performance is conducted by using the design documents from a product design scenario. The results demonstrate that our method outperforms the keyword-based search techniques. This research contributes to the development and use of engineering ontology for design information retrieval.

This paper presents how the function–behavior–structure (FBS) ontology can be used to represent processes despite its original focus on representing objects. The FBS ontology provides a uniform framework for classifying processes, and includes higher level semantics in their representation. We show that this ontology supports a situated view of processes based on a model of three interacting worlds. The situated FBS framework is then used to describe the situated design of processes.

Question–answering (QA) systems have proven to be helpful, especially to those who feel uncomfortable entering keywords, sometimes extended with search symbols such as +, *, and so forth. In developing such systems, the main focus has been on the enhanced retrieval performance of searches, and recent trends in QA systems center on the extraction of exact answers. However, when their usability was evaluated, some users indicated that they found it difficult to accept the answers because of the absence of supporting context and rationale. Current approaches to address this problem include providing answers with linking paragraphs or with summarizing extensions. Both methods are believed to be sufficient to answer questions seeking the names of objects or quantities that have only a single answer. However, neither method addresses the situation when an answer requires the comparison and integration of information appearing in multiple documents or in several places in a single document. This paper argues that coherent answer generation is crucial for such questions, and that the key to this coherence is to analyze texts to a level beyond sentence annotations. To demonstrate this idea, a prototype has been developed based on rhetorical structure theory, and a preliminary evaluation has been carried out. The evaluation indicates that users prefer to see the extended answers that can be generated using such semantic annotations, provided that additional context and rationale information are made available.

After 20 years of personal experience in introducing advanced software into several engineering contexts (engineering consulting firms, construction companies, large diversified companies, and commercial research establishments), some observations may be of interest. Although some development efforts succeeded and software is still in use, not one development effort has met the expectations that were expressed at the beginning of the project. The least successful experiences are design applications.

Engineering design reviews, which take place at predetermined phases of the product development process, are fundamental elements for the evaluation and control of engineering activities. These meetings are also acknowledged as unique opportunities for all the parties involved to share information about the product and related engineering processes. For product development teams, the knowledge generated during a design review is not as secondary as it may seem; key design decisions, design experiences, and associated rationale are frequently made explicit. Useful work has been carried out on the design review process itself, but little work has been undertaken about the detailed content of the meeting activity; it is argued that understanding the transactions that take place during a meeting is critical to building an effective knowledge-oriented recording strategy. To this effect, an extensive research program based on case studies in the aerospace engineering domain has been carried out. The work reported in this paper focuses on a set of tools and methods developed to characterize and analyze in depth the transactions observed during a number of case studies. The first methodology developed, the transcript coding scheme, uses an intelligent segmentation of meeting discourse transcriptions. The second approach, which bypasses the time consuming transcribing operation, is based on a meeting capture template developed to enable a meeting observer to record the transactions as the meeting takes place. A third method, the information mapping technique, has also been developed to interpret the case study data in terms of decisions, actions, rationale, and lessons learned, effectively generating qualitative measures of the information lost in the formal records of design reviews. Overall, the results generated by the set of tools presented in this paper have fostered a practical strategy for the knowledge intensive capture of the contents of design reviews. The concluding remarks also discuss possible enhancements to the meeting analysis tools presented in this paper and future work aimed at the development of a computer supported capture software for design reviews.