Project lead: Dr Edlira Vakaj
Project Reference: CEBE-EV-02
The compliance checking process occurs constantly throughout all phases of a project lifecycle in the AECO (Architecture, Engineering, Construction, and Operation) domain to ensure buildings are fit for purpose energy efficient, and constructed in accordance with the design specifications, functional and, safe to use, and sustainable to the environment throughout its service life. The process for compliance checking is laborious and demands the interpretation of regulations and guidelines. Automating the compliance processes will transform the building design, disrupting the unavoidable tedious checking and approval step and enabling designer to work with generative design using AI more ineffectively and effectively.
The application of Natural Language Processing using a traditional approach (largely applicable for general-domain text processing) for this domain is challenging as building-code sentences typically have deeply nested syntactic and semantic structures, including recursive clauses, conjunctive and alternative obligations, and multiple exceptions. Recent efforts in NLP development have shown that semantic deep neural networks are capable of learning the complex syntactics and semantics of the natural language and thus, gives the potential for automated compliance checking. In this study, Named Entity Recognition (NER) tasks will be employed to identify a set of concepts and relations with reference to recently developed semantic models, e.g. BOT, DiCon ontology, widely adopted by W3C Linked Building Data Community Group leading the development of semantic webs for building models.
The overall aim and objective is to design and implement a consolidated NLP model for generating computer processable rules from compliance regulations of AECO that integrates deep learning, transfer learning strategies, and both target- and general domain data to extract semantic and syntactic information fully automatically
Person Specification
Qualifications Required:
- A BSc in Computer Science or similar.
- MSc in Machine Learning, Data Science, or Artificial Intelligence.
- 2-3 years of experience in full-stack machine learning engineering will be a great advantage.
- International applicants must also provide a valid English language qualification, such as International English Language Test System (IELTS) or equivalent with an overall score of 6.5 with no band below 6.0.
Required Level: Excellent
- Analytical skills and the ability to work independently as well as in a team.
- Written and oral communication skills in English.
- Programming skills in Python.
Required Level: Very Good
- A strong understanding of artificial Intelligence/ Machine Learning methods.
- Natural Language Processing libraries in Python (Spacy, SciKit-Learn, NLTK ).
Required Level: Good
- Ontologies
- Semantic Web Technologies
Other Qualifications (Desirable)
- Familiar with the Build Environment domain applications.
- Familiar with Linked Building Data.
- Publications in high impact factor journals.
To discuss this project, please contact Dr Edlira Vakaj at Edlira.vakaj@bcu.ac.uk
PhD Classic Doctoral Training Grant Funding Information
This funding model includes a 36 month fully funded PhD Studentship, in-line with the Research Council values, which comprises a tax-free stipend paid monthly (2024/5 - £19,237) per year and a Full Time Fee Scholarship for up to 3 years, subject to you making satisfactory progression within your PhD.
All applicants will receive the same stipend irrespective of fee status.
Application Closing Date:
23:59 on Tuesday 1st October 2024 for a start date of the 3rd February 2025.
How to Apply
To apply, please complete the project proposal form,ensuring that you quote the project reference (CEBE-EV-02), and then complete the online application where you will be required to upload your proposal in place of a personal statement as a pdf document.
You will also be required to upload two references, at least one being an academic reference, and your qualification(s) of entry - Bachelor/Masters certificate(s) and transcript(s).