📄️ Publications


Automation in Construction

SCIE, IF = 10.517


Kihoon Son°
Seungwon Lee°
Wondeuk Yoon
Kyung Hoon Hyun*


CreativeSearch: Proactive design exploration system with Bayesian information gain and information entropy

Abstract

A spatial designer's search goal is sequentially updated during the spatial design exploration process. The design exploration process requires a proactive process that supports better-informed design decisions and refreshes the search direction to avoid being fixed; however, no study has been conducted to date. This paper describes a framework called CreativeSearch, which provides three types of guidance feedback by analyzing a spatial designer's design process. Particularly, by integrating the Bayesian Information Gain (BIG) framework and information entropy theory, CreativeSearch provides three types of guidance feedback. A user study conducted with 20 spatial designers confirmed that CreativeSearch supported them in exploring creative design ideas by lowering their mental load. Furthermore, CreativeSearch made designers proactive and supported their design decisions. The guidance feedback framework described in this paper is expected to extend the research on proactive and creative design exploration processes.


Keyword: Guidance feedback; Design exploration; Design creativity; Information theory; Information system

 



BIGexplore: Bayesian Information Gain Framework for Information Exploration

Abstract

The Bayesian information gain (BIG) framework has garnered significant interest as an interaction method for predicting a user’s intended target based on a user’s input. However, the BIG framework is constrained to goal-oriented cases, which renders it difficult to support changing goal-oriented cases such as design exploration. During the design exploration process, the design direction is often undefined and may vary over time. The designer’s mental model specifying the design direction is sequentially updated through the information-retrieval process. Therefore, tracking the change point of a user’s goal is crucial for supporting an information exploration. We introduce the BIGexplore framework for changing goal-oriented cases. BIGexplore detects transitions in a user’s browsing behavior as well as the user’s next target. Furthermore, a user study on BIGexplore confirms that the computational cost is significantly reduced compared with the existing BIG framework, and it plausibly detects the point where the user changes goals.


Keyword: Bayesian information gain; Information exploration; Design exploration; Information retrieval; Computational interaction





Automation in Construction

SCIE, IF = 10.517

Kihoon Son
Kyung Hoon Hyun*

Designer-Centric Spatial Design Support.

Abstract

Design-centric systems focus on searching or generating design alternatives using design retrieval or generation algorithms. Conversely, designer-centric systems focus on providing design feedback to refine design ideas through interactions with designers when creating design alternatives. This paper proposes a novel designer-centric design support system that analyzes user design moves and provides real-time feedback to augment the spatial design exploration process. Specifically, we developed a system that integrated sketching, design retrieval, and feedback features for tablet PCs. The user experiments involving 21 professional spatial designers verified that the proposed system helped users interact with their designs, navigate novel design spaces, reevaluate design moves, and embody design ideas. These findings provided implications for developing an interactive design exploration method for coordinating designers and computers, thereby extending design support system research. Further, we discuss applications and challenges in implementing the system on the web, revealing possibilities for data trailing of design moves.


Keyword: Computational Design; Designer-Centric Design Support System; Assisted Creativity; DesignFeedback; Design Exploration; Early Design Phase





C-Space: An Interactive Prototyping Platform for Collaborative Spatial Design Exploration.

Abstract

C-Space is an interactive prototyping platform for collaborative spatial design exploration. Spatial design projects often begin with conceptualization that includes abstract diagramming, zoning, and massing to provide a foundation for making design decisions. Specifically, abstract diagrams guide designers to explore alternative designs without thinking prematurely about the details. However, complications arise when communicating ambiguous and incomplete designs to collaborators. To overcome this drawback, designers devote considerable amounts of time and resources into searching for design references and creating rough prototypes to explicate their design concepts better. Therefore, this study proposes C-Space, a novel design support system that integrates the abstract diagram with design reference retrieval and prototyping through a tangible user interface and augmented reality. Through a user study with 12 spatial designers, we verify that C-Space promotes rapid and robust spatial design exploration, inducing collaborative discussions and motivating users to interact with designs.


Keyword Spatial Design; Design Support system; Design Collaboration; Prototyping; Tangible User Interface; Augmented Reality; Human-Computer Interaction;





Communications in Computer and Information Science
(CAAD FUTURES 2021)

SCOPUS (Computer Science, Mathematics).

Bo Hyun Park
Kihoon Son
Kyung Hoon Hyun*

Interior Design Network of Furnishing and Color Pairing with Object Detection and Color Analysis based on Deep Learning

Abstract

Furnishing is one of the most important interior design elements when decorating the space. Since every interior design element is colored, it is essential to consider the pairing of furnishing and color during the design process. Despite the importance of the furnishing and color pairing, the decision-making process of the pairings remains as a "black-box" of the interior design process. However, the advancement of social networks and online interior design platforms such as Airbnb and Today's House allows collecting large quantities of actual interior design cases shared publicly. Thus, this paper proposes a data-driven approach to unveil the secrets of furnishing and color pairing through object detection, color extraction, and network analysis. To do that, we collected a large quantity of image data (N=14,111) from an online interior design platform and extracted furnishing objects and color palettes from the collected image using object detection and color extraction algorithms. Finally, we identified distinctive patterns of furnishing and color pairing through network analysis.


Keyword Color Network; Color-Furnishing Pairing; Machine Learning; Network Analysis; Interior Style Analysis






CAADRIA 2021

YOUNG CAADRIA AWARD
(Top 14%)


Kihoon Son
Kyung Hoon Hyun*

A Framework for Multivariate Data based Floor Plan Retrieval and Generation

Abstract

Spatial designers explore various design references in the design process. These design references significantly impact the quality of design outcomes and the process. Therefore, it is crucial to provide useful designs through the retrieval or generation process to spatial designers. To do this, a methodology must be developed to identify and quantify the floor plan’s multivariate design data. Through quantifying various design data, the retrieval and generation process can provide appropriate designs in many ways. This study proposed a new floor plan design framework for retrieval and generation with newly quantified design data. For validation of this framework, we conducted a floor plan retrieval and generation process. Newly quantified design data show usability in both processes. We also compare our framework with previous studies for validation. The comparison results show that our framework utilizes the most diverse design data of the floor plan.


Keyword Design quantification; Multivariate data; Floor plan design; Design retrieva; Design generation;





Analyzing the Impact of Design Creativity on Spatial Layout Design Depending on the Level of Detail (LoD) of a Design Reference

Abstract

Space designers use existing design references for creative inspiration. However, designers' creative processes can be influenced by the representation and the amount of information that the design references contain. In this respect, researchers on design creativity conducted studies to assist designers when utilizing design references. H ow ever, previous literature does not consider the connectivity of spaces, even though it is an essential architectural element that defines space circulation. Therefore, it is important to evaluate how designers are affected by the level of details (LoD) of design references when designing spatial connectivity quantitatively. Thus, this paper aims to propose a method for quantitative evaluating the design novelty from an aspect of spatial connectivity. Also, using this method, results of experiment was analyzed the effect of space design reference LoD on the spatial connectivity design. The experiment was conducted with participants majoring in interior architectural design. Similarity of the spatial connectivity between the results and reference was calculated though the graph edit distance. Then the effect of the design reference LoD on the design of the spatial connectivity betw een the spaces w as analyzed based on the calculated results. Using the methodology of creativity evaluation of case-based design proposed in this research, three types of design references was evaluated. As a result, this research clarified the LoD of the design reference that can support the creativity.


Keyword Spatial Design; Level of Detail; Spatial Layout; Quantification Study; Design Reference Retrieval; Design Creativity





CAADRIA 2020

Kihoon Son
Hwiwon Chun
Kyung Hoon Hyun*

Ambiguous vs. Concrete: Identifying the Effect of Design References with Various Level of Details on Designer's Creativity in the Early Design Phase

Abstract

During the early design phase, spatial designers search for design references to develop design ideas. In this process, the level of detail (LoD) of design references can significantly influence the quality of design outcomes. However, previous studies have only suggested guidelines indicating that abstract references are useful in the early design phase without the degree of LoD. In response, this study aims to identify the impact of LoD of design references on design outcomes during the design concept development. To this end, we proposed three different reference types (abstract, hybrid, and concrete), and conducted experiments to assess the creativity and efficiency of the design outcome per LoD type. We also developed the FPRT (Floor Plan Retrieving Tool) system along with 7,842 existing residential floor plans for the experiments. The results of the study showed that there is a significant difference in design outcomes depending on the LoD types.


Keyword Design Reference; Design Retrieval; Spatial Design; Level of Detail; Early Design Phase





CAADRIA 2019

Poster

Kihoon Son
Seonghoon Ban
Kyung Hoon Hyun*


RAPID SPACE DESIGN PROTOTYPING TOOL BASED ON AUGMENTED REALITY AND TANGIBLE USER INTERFACE

Abstract

It is essential to explore and test a vast number of design variations through prototypes to refine the initial idea into a creative design outcome. Architects often create a scale model of their design concept by cutting out and gluing foam cores to test the visual and functional relationships among design elements. However, creating a scale model for a built environment requires time, efforts and money. This research proposes the Rapid Space Design Prototyping Tool (RSDPT) which enables designers to quickly simulate design variations of the built environment through tangible user interface and augmented reality. RSDPT consists of three major components: 1) modular parts (pillars, outer walls, inner walls) with magnets inside; 2) infrared (IR) sensing camera for automatic 3D model reconstruction; 3) projection mapping features for color and texture simulation. Thus, designers can rapidly create design prototypes with magnetic modular parts and test various materials with augmented reality using projection mapping.


KeywordPrototyping; Tangible User Interface; Physical Modeling; Level of Detail; Design Variation