Social Media and Heritage Management - paving the way towards inclusive approaches
This workshop explores how to use social media data for a more inclusive approach to heritage management.
About
Social media is often used to by tourists and locals alike to express their opinions about architectural buildings and urban spaces, in particular heritage sites. Many scholars have been using social media to conduct innovative research to engage people, interpret their opinions, and code values associated with heritage attributes.
The workshop is a result of a project called "Public Participation and Consensus in World Heritage management", which aims to facilitate public inclusion in heritage management. The workshop presents an introduction to coding values and attributes of text documents based on two theoretical frameworks (Veldpaus, 2015; Pereira Roders, 2007). It takes you step by step through the stages of the innovative project which codes social media data using Artificial Intelligence (including Natural Language Processing techniques and BERT model).
It took place on 2 September 2021.
Speaker
Mahda Foroughi: As an architect, I have always been interested in more inclusive approaches to the built environment. I was the head of a participation project in a rehabilitation project in a historic neighbourhood in Tehran, Iran, using different focus group meetings and interviews. I am now exploring how digital technology facilitates inclusive consensus-building in heritage management by offering social media platforms, big data analytic tools, machine learning, and Natural language processing. My Ph.D. at TUDelft entitles "consensus building and public participation in Cultural Heritage Management.” It aims to facilitate public inclusion in heritage management using available unstructured textual data sources, namely, social media, literature, and policy documents. In this study, digitalization is accrued in the data source, analytical tools, and data management. As the data source, this research uses social media posts from Instagram and Twitter. As analytical tools, various Natural language processing techniques are also used for the analysis of the study. The central part of the research is adopting the Bert model, a pre-trained neural network model, to code the posts' content (sentences and hashtags) to extract the heritage attributes and values and their typologies. The codes are implemented in Python through PyTorch. Lastly, a relational database is created in SQL, and data analytics are implemented there.