| Robert Amor's Publications in 2023 |
|---|
 |
Kassem, M., Tagliabue, L.C., Amor, R., Sreckovic, M., Chassiakos, A., (ed.) (2023) Proceedings of the 2023 Conference of European Council for Computing in Construction (EC3) and 40th International CIB W78 Information Technology for Construction Conference, Heraklion, Crete, Greece, 10-12 July, https://doi.org/10.35490/EC3.2023. |
Abstract: The 2023 European Conference on Computing in Construction was held as a mixed Conference along with CIB W78 from July 10 - July 12, 2023. The conference had 175 attendees that presented work and exchanged ideas in the areas of the conference. This book contains the papers that were submitted to the conference and were accepted after a rigorous peer-review process. The 2023 proceedings include an illustrated review of the program, the names of organizations and persons who contributed to the technical program. The peer review process consisted of two phases. Firstly, we received 54 optional abstracts that were reviewed by the respective area chairs. Following a rigorous full paper peer review process (with each full paper being reviewed by at least two reviewers drawn from the scientific committee of international experts, and final decisions being made collectively by the corresponding track and programme chairs), 140 outstanding full papers were ultimately included in the proceedings and presentation at the conference.
|
Wang, Q., Wang, R., Zhao, K., Amor, R., Liu, B., Zheng, X., Zhang, Z., Huang, Z. (2023) Towards Legal Judgment Summarization: A Structure-Enhanced Approach, Proceedings of ECAI 2023, Kraków, Poland, 30 September - 5 October. |
Abstract: Judgment summaries are beneficial for legal practitioners to comprehend and retrieve case law efficiently. Unlike summaries in general domains, e.g., news, judgment summaries often requires a clear structure. Such a structure helps readers grasp the information contained in the summary and reduces information loss. To the best of our knowledge, none of the existing text summarizers can generate summaries aligned with the summary structure in the legal domain. Inspired by this observation, this paper introduces a Summary Structure-Enhanced (SSE) method to synthesize structured summaries for legal documents. SSE can easily be incorporated into the Encoder-Decoder framework, which is commonly adopted in state-of-the-art text summarizers. Experiments on the datasets of New Zealand and Chinese judgments show that the proposed method consistently improves the performance of state-of-the-art summarizers in terms of Rouge scores.
|
Fuchs, S., Dimyadi, J., Witbrock, M., Amor, R. (2023) Using Large Language Models for the Interpretation of Building Regulations, Proceedings of EPPM 2023, Auckland, New Zealand, 30 November - 1 December. |
Abstract: Compliance checking is an essential part of a construction project. The recent rapid uptake of building information models (BIM) in the construction industry has created more opportunities for automated compliance checking (ACC). BIM enables sharing of digital building design data that can be used for compliance checking with legal requirements, which are conventionally conveyed in natural language and not intended for machine processing. Creating a computable representation of legal requirements suitable for ACC is complex, costly, and time-consuming. Large language models (LLMs) such as the generative pre-trained transformers (GPT), GPT-3.5 and GPT-4, powering OpenAI's ChatGPT, can generate logically coherent text and source code responding to user prompts. This capability could be used to automate the conversion of building regulations into a semantic and computable representation. This paper evaluates the performance of LLMs in translating building regulations into LegalRuleML in a few-shot learning setup. By providing GPT-3.5 with only a few example translations, it can learn the basic structure of the format. Using a system prompt, we further specify the LegalRuleML representation and explore the existence of expert domain knowledge in the model. Such domain knowledge might be ingrained in GPT-3.5 through the broad pre-training but needs to be brought forth by careful contextualisation. Finally, we investigate whether strategies such as chain-of-thought reasoning and self-consistency could apply to this use case. As LLMs become more sophisticated, the increased common sense, logical coherence and means to domain adaptation can significantly support ACC, leading to more efficient and effective checking processes.
|
Zhu, Y., Deng, Z., Chen, Y., Amor, R., Witbrock, M. (2023) Chain of Propagation Prompting for Node Classification, Proceedings of 31st ACM International Conference on Multimedia, Ottawa, Canada, 29 Oct-3 November, pp. 3012-3020, https://doi.acm.org?doi=3581783.3612431. |
Abstract: Graph Neural Networks (GNN) are an effective technique for node classification, but their performance is easily affected by the quality of the primitive graph and the limited receptive field of message-passing. In this paper, we propose a new self-attention method, namely Chain of Propagation Prompting (CPP), to address the above issues as well as reduce dependence on label information when employing self-attention for node classification. To do this, we apply the self-attention framework to reduce the impact of a low-quality graph and to obtain a maximal receptive field for the message-passing. We also design a simple pattern of message-passing as the prompt to make self-attention capture complex patterns and reduce the dependence on label information. Comprehensive experimental results on real graph datasets demonstrate that CPP outperforms all relevant comparison methods.
|
Neuhäuser, T., Dimyadi, J., Eckart, C.J., Wagner, F., Hohmann, A., Amor, R., Daub, R. (2023) Systematic Merging of Building Information and Simulation Models for the Automated Evaluation of Factory Layout Variants, Proceedings of Joint CIB W78 and EC3 2023, Heraklion, Greece, 10-12 July, pp. 248-255. |
Abstract: The turbulent business environment facing manufacturing companies today has prompted the need for a more efficient decision making process in the factory planning without compromising on reliability. Additional, emerging objectives, e.g., demand for ecological sustainability and adaptability, also present challenges that must be addressed. This paper describes a systematical integration approach for Building Information Modeling and simulation data to streamline the planning process by automatically evaluate factory layout variants. An experiment has been conducted to demonstrate the viability of the approach.
|
Fuchs, S., Dimyadi, J., Witbrock, M., Amor, R. (2023) A LegalRuleML Editor with Transformer-based Autocompletion, Proceedings of Joint CIB W78 and EC3 2023, Heraklion, Greece, 10-12 July, pp. 156-163. |
Abstract: The construction industry has pursued automated compliance checking for decades, but legal requirements conveyed in natural language are not intended for machine processing. There have been numerous attempts to translate these requirements into computable representations, progressing from manual to fully-automated approaches. However, it is unclear if fully-automated translation will become reliable and interpretable enough for legal matters. We propose a LegalRuleML Editor with Transformer-based Autocompletion to facilitate a semiautomated workflow with minimal manual effort. A deep learning model generates initial translations and contextualised autocompletion options. This strategy offers experts a superior translation process, including continuous improvements approximating full automation.
|
Fuchs, S., Dimyadi, J., Witbrock, M., Amor, R. (2023) Improving the Semantic Parsing of Building Regulations through Intermediate Representations, , Proceedings of EG-ICE 2023, London, UK, 4-7 July. [best paper award] |
Abstract: Recent developments show that large transformer-based language models have the capability to generate coherent text and source code in response to user prompts. This capability can be used in the construction domain to interpret building regulations and convert them into a semantic representation usable for automated compliance checking. While base-size models can already be taught to perform semantic parsing with decent quality, this paper shows how intermediate representations (IR) can be used to improve the semantic parsing quality. With reversible IRs, the training time could be reduced to almost a quarter of the initial duration, and through adding a hierarchical parsing step, improvements of up to 6.6% on F1-Scores were reached.
|
Okakpu, I., Preston, G., Amor, R. (2023) Integrating Building Information Modelling and Health and Safety Design Phase, BIMSafe Report, University of Canterbury, 46pp. |
Abstract: Building Information Modelling (BIM) has an established role for health and safety (H&S) in the design stage. Safety in Design and Prevention through Design approaches can be adapted to the significant information which is available in a BIM model, which enhances the ability to practice these approaches. With the majority of NZ construction projects using BIM at some stage there is no reason that these common practices should not be supported with the design-level BIM model. This report describes three technological advances which the international research and case studies show can have a significant impact on H&S at the design stage. These promising technologies: provide safety rule checks based on a BIM model; allow for formalised safety in design knowledgebases based on expertise from all in the industry; and support risk assessment from the design BIM model. While these technologies require significant investment, often at the national level to provide a complete solution, they identify simpler approaches that can be implemented by all in the design stage as highlighted in the recommendations below.