Transactions on Graphics Referee's Report
Title: Designer: A Knowledge-Based Graphic Design Assistant
Author: Louis Weitzman
Reference No.: SPIS-11
Date assigned: July 22, 1986
Date due: August 22, 1986
I. General evaluation.
(If your answer to any question is no, please explain.)
A. Is the title appropriate?
Yes, but this is only a preliminary design for a graphic design assistant for the Steamer Graphics Editor. Few principles for creating such knowledge-based assistants are presented.
B. Is the abstract satisfactory?
Yes, but the paper is too superficial.
C. Are the key words well-chosen?
Yes.
D. Content
1. Does the paper have enough originality, importance and interest to warrant publication in the Research Contributions section of TOG?
No. See III.1. Research contributions.
2. Does the paper have enough importance and interest to warrant publication in the Practice and Experience section of TOG?
No. This is not an experience paper.
3. Is the paper technically sound? (List mistakes under section III.)
Maybe. See III.2. Technical soundness.
4. Does the paper make adequate reference to earlier contributions?
No. See III.3. References.
E. Presentation. (Please comment on deficiencies and/or annotate paper.)
1. Is the paper well-organized?
Yes, but too superficial.
2. Does the paper contain excess material?
Excess figures. See E.6 below.
3. Does the paper lack necessary material?
What graphic design knowledge does Designer incorporate? What corpus of graphic design knowledge is appropriate to include in an assistant? How has previous research contributions been incorporated in this work?
4. Is the presentation clear?
No. The level of detail is at times too superficial, yet at other times too involved with the implementation. There are several concepts that are insufficiently argued, such as the satisfaction of constraints and the incorporation of ATMS. This confuses the reader as to what audience is expected to understand this paper.
The audience for this paper is unclear. It might be appropriate for researchers in Artificial Intelligence, practitioners in User Interface designs, or computer graphic and graphic design folks.
5. Is the English satisfactory?
Yes
6. Are more or fewer figures necessary?
The figures lack clearly stated captions identifying the purpose of the illustrations.
One of Figure 1 or 2 appears redundant; the important images are all present in Figure 1 while Figure 2 contains an alternative organization.
Figure 3 wastes a lot of space giving details that are not supported by the written material in the paper.
Figure 5 shows a preliminary interface. It might be dispensed with in favor of more detailed presentation of the important interface components.
Figures 7, 8, 10, and 11 display screen images with considerably more detail than necessary to support the references to these figures from the paper. In particular, Figure 10 includes a long list of visual techniques that are not introduced in the paper; where did these concepts come from?
Figure 12 omits the style detail for Simplicity/Neutral/Complexity.
II. Recommendations.
Reject.
Submit to CHI+GI'87 conference.
III. Comments.
1. Research contribution
The research concepts introduced by this paper indeed are important and are of interest to TOG readers of the special issue on user interface technology. However, this paper treats the problem of creating graphic design assistants too superficially for a computer graphics research audience. Yet on the other hand, it presents too many details of the implementation suitable for a reader unfamiliar with artificial intelligence techniques.
The most important omission is the framework supporting the ideas in this research. The paper wanders through several important concepts (design constraints, and the Analysis—Critique—Synthesis structure) without sufficiently revealing the underlying principles.
2. Technical soundness
I believe that this is an important and early step in the introduction of graphic design knowledge. With only two design principles (significant difference in size or location) presented, the paper is unclear about how effective this work is in handling additional perceptual difficulties of color, typography, containment relationships, textures, and so on.
3. References
The graphic design literature is cited, but poorly integrated in the body of the paper, mainly cited as a pointer for the readers to search for their own accumulation of graphic design knowledge. I would prefer that the important contributions from the cited works be described.
Several works appear in the references but are not cited in the paper. The graphical style work by Beach and Stone relates to the variation of rendering attributes; Borning's work on constraints helps determine how constraints should be specified and satisfied; the work by Mackinlay (correct spelling) on ``Automatic design of graphical presentations'' and his criteria for assessing designs are extremely relevant to this paper; the visual design work by Roach et al. and Scholl are tantalizing in the references but not described in the paper.
4. Detailed comments
Introduction, para 2: The ``views'' in this paragraph are Steamer views, rather than more general ones. The introductory sentence should clarify this.
Introduction, para 2: The last sentence refers to ``incremental refinement.'' Does this refer to the use of the Steamer Graphical Editor without Designer?
Overview, para 1: What is the definition of ``visual expertise''? Examples?
Overview, para 1, 3rd sentence: This sentence contains awkward comma punctuation and parallel structure in ``sets of constraints for establishing a context, or style, for critiquing a design.''
Domain-Dependent ..., para 1: The references to a large set of graphic design knowledge might be better introduced by annotating what kind of knowledge can be gained from each reference. The claim that ``Designer incorporates much of this knowledge'' is unsubstantiated by the remainder of the paper and by the lack of an enumeration of the knowledge gleaned from the list of references.
Domain-Dependent ..., para 2: The material on ``MSG flavor enhancer'' is not well motivated or connected to representing graphic design knowledge. What is missing is a motivating description of why it is hard to represent graphic design knowledge.
Relationships: What are the principles that guided the inclusion of graphic-relationships knowledge in Designer? What is presented is detail from an example.
Constraints, para 1: The example principle ``Significant Difference of Size'' is discussed in terms of both size and location. Later the paper clarifies that there are two significant difference principles.
Constraints, para 1-2: Only a small number of principles and standards are considered. Is it known if the research extends gracefully to a much larger set of constraints?
Constraints, para 1-2: The representation or specification of constraints is not described. Is it unimportant?
Design Context, para 2: Where did the concepts illustrated in Figure 10 come from? How are constraints related to the design context?
Analysis, para 1: Should ``elements instance variables'' use the possessive or is it ``the instance variables of the elements class''?
Analysis, para 2: The reference to ``beautifying diagrams'' does not identify what relevant work was accomplished in the cited research.
Critique, para 1: The detail of ``comments are Flavor objects'' and ``comments, displayed in the scrolling pane'' are nits, compared to the importance of relating the critque to the objects being critiqued. The paragraph should be reworded to present the most important concepts first, then, supporting detail can be introduced.
Critique, para 1: It is unclear, from either the written material or the accompanying figures, what the designer sees of the graphical design under development. For instance, does the designer work only with the screen images in Figures 10 or 11?
Critique, para 2: It is not obvious how the statement ``It is thus possible'' relates to the detail of critiquing presented in the previous paragraph. Does the possibility exist because of the critiquing technique or because one can specify independent styles?
Synthesis, para 2: At this point in the paper, it is unclear how the designs are modified to satisfy the constraints. Only two constraints, Significant Difference of Size, and Significant Difference of Location, are identified. How would more constraints be handled, such as choice of color, choice of typography, choice of containment?
Synthesis, para 3-4: The brief introduction of ATMS raises the question of who is the audience for this paper? Is the reader expected to have sufficient insight into ATMS to determine how such a system provides support for Designer. The proof of its necessity and effectiveness are missing.