Incorporating Qualitative Criteria in Multi-Objective Architectural Design

Optimization through Interaction: An Adaptive, Cognitive Framework

PHD Research

 dr. Ioannis




PhD started in: 2012


Latest graduate degree

MSc Building Technology, TU Delft


undergraduate degree MSc Architecture,

Aristotle University of Thessaloníki




Prof. Dr. I.S.Sariyildiz


Daily Supervisor(s):


Asst. Prof. Dr. Michela Turrin


Main Question:

How can qualitative and subjective

criteria be successfully

incorporated in computational

optimization for architecture?



1. Novel method

for incorporating qualitative criteria

during optimization through user


2. Digital tools and

interfaces implementing said

method and state-of-the art multi objective


3. User study

validating said method on real-world

design problem.






Keywords: Qualitative Criteria, Interactivity, Evolutionary Computation, Multi-Objective Optimization,

Cognition, Aesthetics


Area of Research: Computation & Performance



Chair of Design Informatics, Dept. of Architectural Engineering + Technology, Faculty of Architecture






































Research Summary:

Architectural design entails, to a large degree, criteria that are not easily quantifiable. Such criteria are related to psychological and cognitive aspects of design, such as spatial experience and aesthetics. Due to this fact, the use of computational optimization methods in architectural design, while having the potential to benefit design a great deal, is severely limited. The motivation behind this thesis is to facilitate the inclusion of qualitative criteria alongside quantitative objectives, in an integrated computational optimization framework. We consider as a starting point the proposition that decision maker input is invaluable in providing information regarding their preferences, and that the decision maker should be involved during the optimization process. Interactive optimization systems, such as Interactive Evolutionary Computation (IEC), are high on the modern scientific agenda, but encounter challenges dealing with human cognitive limitations. This thesis proposes an adaptive and cognition modeling framework, that can capture the preferences of the user, while complying to human cognitive limitations at hand. Furthermore, we address the issue of integration of the proposed framework to a multi-objective approach, with the aim of

simultaneously tackling qualitative and quantitative objectives, in demanding real-world problems. Finally, we validate the proposed framework on a real-world design problem and conduct a user survey to evaluate the effectiveness of the proposed method.




Research Methodology:

Identification of critical cognitive factors and limitations in decision maker involvement through review of the state of the art in Evolutionary Computation and Interaction; comparison and implementation of machine learning models for modeling user qualitative preferences; theoretical foundation, development and implementation of methods for progressive derivation of said machine learning models; integration with quantitative objectives in a multi-objective framework; identification of suitable user interaction methods; implementation of software tools and interfaces; appropriate metrics for evaluating method performance through user surveys.



Key Publications:

I. Chatzikonstantinou and S. Sariyildiz, “Approximation of simulation-derived visual

comfort indicators in office spaces: a comparative study in machine learning,” Archit. Sci. Rev., no. August

2015, pp. 1–16, Aug. 2015;

I. Chatzikonstantinou, “A Computational Intelligence Decision-Support Environment for Architectural and

Building Design: CIDEA,” in IEEE Congress on Evolutionary Computation 2016, 2016, pp. 3887–3894.











Telnr +3127892136


Room 01+.West.040


Faculty of Architecture and the

Built Environment


Building 8


Julianalaan 134


2628 BL Delft