Yusheng Lu 卢渝生

Purpose

The digital revolution and the use of big data (BD) in particular has important applications in the construction industry. In construction, massive amounts of heterogeneous data need to be analyzed to improve onsite efficiency. This article presents a systematic review and identifies future research directions, presenting valuable conclusions derived from rigorous bibliometric tools. The results of this study may provide guidelines for construction engineering and global policymaking to change the current low-efficiency of construction sites.

Design/methodology/approach

This study identifies research trends from 1,253 peer-reviewed papers, using general statistics, keyword co-occurrence analysis, critical review, and qualitative-bibliometric techniques in two rounds of search.

Findings

The number of studies in this area rapidly increased from 2012 to 2020. A significant number of publications originated in the UK, China, the US, and Australia, and the smallest number from one of these countries is more than twice the largest number in the remaining countries. Keyword co-occurrence is divided into three clusters: BD application scenarios, emerging technology in BD, and BD management. Currently developing approaches in BD analytics include machine learning, data mining, and heuristic-optimization algorithms such as graph convolutional, recurrent neural networks and natural language processes (NLP). Studies have focused on safety management, energy reduction, and cost prediction. Blockchain integrated with BD is a promising means of managing construction contracts.

Research limitations/implications

The study of BD is in a stage of rapid development, and this bibliometric analysis is only a part of the necessary practical analysis.

Practical implications

National policies, temporal and spatial distribution, BD flow are interpreted, and the results of this may provide guidelines for policymakers. Overall, this work may develop the body of knowledge, producing a reference point and identifying future development.

Originality/value

To our knowledge, this is the first bibliometric review of BD in the construction industry. This study can also benefit construction practitioners by providing them a focused perspective of BD for emerging practices in the construction industry.

Online Appendix and Replication Files

 
Lu, Y. and Zhang, J. (2022). Bibliometric analysis and critical review of the research on big data in the construction industry. Engineering, Construction and Architectural Management, 29(9), 3574-3592. https://doi.org/10.1108/ECAM-01-2021-0005
 

Product recommendation systems have become an effective tool to help users make choices under information overload. For sequence recommendation, the user’s dynamic preferences and the correlations between items are essential for exploring temporal information in hidden states and latent relationships, which previous research has rarely considered. This paper proposes a novel model named MhSa-GRU that combines Multi-head Self-attention with a Gated Recurrent Unit, and integrates the prices, user behavior, and category features of items to help generate the next recommendation. In this model, an improved self-attention layer captures the items’ correlations, and multiple heads learn thorough local information about the vector. The mask and attention threshold mechanisms can exclude disturbing information commonly observed in product recommendation systems. Moreover, the GRU module introduces users’ global and local preferences to capture their behavior. Universal experiments on three real-world datasets verify that MhSa-GRU outperforms competitive baseline models, such as HR@10, HR@20, MRR@10, and MRR@20, in evaluation metrics for prediction. MhSa-GRU improves performance by 5% on the Cosmetics 2019-Nov and Cosmetics 2020-Jan datasets, and by about 10% on the Ta Feng dataset. Ablation experiments help determine the interpretability and effectiveness of the multi-head self-attention module, and show this module can strengthen efficiency by about 5%–10%. We also find that the predicted efficacy is optimal when the attention-head dimension is roughly 10% to 20% of the embedding vector. In addition, the sparse dataset should use high-dimensional attention-head vectors since the local information is too sparse to capture the relevance between items.

Online Appendix

 
Duan, Y., Liu, P. & Lu, Y. (2022). MhSa-GRU: combining user’s dynamic preferences and items’ correlation to augment sequence recommendation. Journal of Intelligent Information Systems. https://doi.org/10.1007/s10844-022-00754-0
 

Live streaming as a new shopping channel can disclose revealing information on quality preference. With this channel, viewers can obtain additional product details along with entertainment utility from streamers' product showcasing. This paper examines live streaming strategies for vertically differentiated firms that sell products of high or low quality. We obtain the optimal pricing strategies for the high-quality, low-quality firms under a particular live streaming strategy (i.e., (N, N); (S, N); (N, S); (S, S)). We identify how the competing firms should set their live streaming strategies. As the entertainment value increases, firms with a competitive edge are more willing to adopt live streaming. If the “quality-adjusted cost” is lower than 1/2, then the high-quality firm is prone to adopt live streaming; otherwise, the low-quality firm is prone to adopt live streaming. The entertainment value greatly influences the adoption decision. When the streamer is more popular and charges a medium fixed cost, the competitive firm (i.e., small marginal cost, high quality) should adopt live streaming to expand its marketplace. The other firm should refuse live streaming and emphasize improving its product quality and competitiveness. Interestingly, live streaming is not always beneficial to firms when the competitive advantages dominate the impact of live streaming and the informative degree is improved little.

Online Appendix

 
Lu, Y., Duan, Y. (2023). Strategic live streaming choices for vertically differentiated products. Journal of Retailing and Consumer Services. https://doi.org/10.1016/j.jretconser.2023.103582
 

The online content, including live streaming and short videos, provides abundant visual and textual product information to users, which offers insights into users’ multiple and changeable preferences toward product style, brand, color, etc. These dynamic preferences are helpful to the product recommender system to extract users’ purchasing patterns. However, existing sequential recommendation fails to combine with items’ multimodal characteristics and dynamic preferences. This paper proposes the Online Content-Based Sequence Recommendation method (named OCBSR), where dynamic preferences and multimodal representation such as visual images, textual descriptions, and item-ID of products are leveraged with convolutional autoencoder to map multimodal data into a unified latent space. We devise a multimodal transformer mechanism to aggregate the multimodal representation. A novel contrastive infomax loss is constructed through contrastive prototype representations and noise-contrastive sequences. The loss enhances the correlation between single-modality and multimodality preferences and learns the contrastive representation. Then, users’ dynamic preferences are extracted from the contrastive representation in each time step using the gated recurrent unit with skip connections. Experiments are conducted on two real-life multimodal datasets, H&M and KuaiRec. The hit ratio (HR) and normalized discounted cumulative gain (NDCG) results indicate that our approach noticeably outperforms the current advanced recommenders. The HR increases by above 20% and 25% on H&M and KuaiRec datasets, respectively; the NDCG rises about 12% and 24% on H&M and KuaiRec datasets, respectively. The ablation study verifies each module’s effectiveness. This research contributes to online content literature and multimodal sequential recommendation in practice.

Online Appendix

 
Lu, Y., Duan, Y. Online content-based sequential recommendation considering multimodal contrastive representation and dynamic preferences. Neural Comput & Applic, 36, 7085–7103 (2024). https://doi.org/10.1007/s00521-024-09447-x