Yifan Liu

Yifan Liu

Auto-generated profile, based on publicly available information

Yifan Liu

Google, Mountain View, California, USA

Research Focus & Metrics

Academic Career Length

9Years (2018-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 January 2020 in Journal of Computing in Civil Engineering
Verified
Journal Article
Closed access
YLYifan Liu
Yifan Liu
FCFadi Castronovo
Fadi Castronovo
JMJohn Messner
John Messner
RLRobert Leicht
Robert Leicht
With the help of virtual reality (VR), architectural design can be visualized interactively and at full scale. Viewers of VR models also gain an additional sense of spatial depth and volume. These capabilities make VR a unique design visualization media, and have attracted research interests in applying VR for design reviews. For example, previous research has studied the application of VR for evaluating the designs of courtrooms, nuclear power plants, patient rooms, and educational buildings. This research extends contributions reported in existing literature that aims to explore the value and impact of VR on design review meetings. This research performs a detailed analysis of a large number of design review meetings, including 13 meetings conducted on building construction projects. A novel design review evaluation methodology, designed to leverage video analysis techniques was adopted to allow for higher consistency of the evaluation results and reducing the impact of user bias caused by the novelty effect of VR. This methodology provides systematic rules and metrics to analyze design review meeting quality. The research has also extended the analysis of the impact of VR-based design review meetings to those that involve occupants and facility maintenance staff, which is a type of VR-based review that has not been extensively studied in the previous literature. The results of this research show that VR alone is not necessarily better than traditional media, such as drawings, in facilitating active discussions of design issues. Instead, the authors find that design review meetings yield more effective discussion and better meeting efficiency when VR is orchestrated with, and supported by, other design visualization media, such as drawings, digital models, renderings, and maps. This paper also summarizes six strengths and four weaknesses of VR to support design review meetings when compared with 2D drawings. To help project teams decide if VR should be used for their design reviews, the authors recommend the consideration of three criteria: (1) the reviewers’ exposure to the design, (2) the scope of the review, and (3) the project complexity.

8 October 2024
Verified
Conference Paper
Open access
JWJianling Wang
Jianling Wang
HLHaokai Lu
Haokai Lu
YLYifan Liu
Yifan Liu
HMHe Ma
He Ma
YWYueqi Wang
Yueqi Wang
MCMinmin Chen
Minmin Chen
Traditional recommendation systems are subject to a strong feedback loop by learning from and reinforcing past user-item interactions, which in turn limits the discovery of novel user interests. To address this, we introduce a hybrid hierarchical framework combining Large Language Models (LLMs) and classic recommendation models for user interest exploration. The framework controls the interfacing between the LLMs and the classic recommendation models through “interest clusters”, the granularity of which can be explicitly determined by algorithm designers. It recommends the next novel interests by first representing “interest clusters” using language, and employs a fine-tuned LLM to generate novel interest descriptions that are strictly within these predefined clusters. At the low level, it grounds these generated interests to an item-level policy by restricting classic recommendation models, in this case a transformer-based sequence recommender to return items that fall within the novel clusters generated at the high level. We showcase the efficacy of this approach on an industrial-scale commercial platform serving billions of users. Live experiments show a significant increase in both exploration of novel interests and overall user enjoyment of the platform.

19 April 2018 in Architectural Engineering and Design Management
Verified
Journal Article
Closed access
YLYifan Liu
Yifan Liu
JMJohn I. Messner
John I. Messner
RLRobert Leicht
Robert Leicht
Most of the previous research on design reviews focuses on management and technology innovations that improve the effectiveness of design reviews. To better evaluate the impact of those innovations, a structured view of the design review process is needed, but few previous works have provided such a view. This research studies the dynamics of usability and maintainability design review meetings that involve stakeholders in the post-occupancy phase of the building, including owners, occupants, and facility operators. The outcomes of the research include a process model, which describes the communication and interaction dynamics between the participants in those design reviews. The process model is developed and validated following grounded theory methodology through analyzing 20.5 h of the video recordings from 13 design review meetings, which are from real projects with diverse sizes and types. The process model defines the core of the design review process as three cycles of activities, which are called Understand, Validate, and Resolve Cycle respectively. The process model also defines the interactions among activity cycles and how they advance the design review process. This research contributes to the fundamental theory of design review process and provides a roadmap for other researchers to analyze the usability and maintainability design reviews. To further support future researchers’ evaluation of design review process, we also proposed a theory to derive the design review meeting performance by analyzing the time spent in each activity cycles. This theory can be used as a foundation by future researchers to evaluate and compare the impact of technology and process innovations on design reviews meetings.

22 September 2025
Verified
Conference Paper
Open access
CMChangping Meng
Changping Meng
HLHongyi Ling
Hongyi Ling
JWJianling Wang
Jianling Wang
YLYifan Liu
Yifan Liu
SZShuzhou Zhang
Shuzhou Zhang
NHNingren Han
Ningren Han