Using Topic Modelling to Reassess Heritage Values from a People-centred Perspective: Applications from the North of England

Abstract

The historic environment—comprising a palimpsest of landscapes, buildings and objects—carries meaning and plays a crucial role in giving people a sense of place, identity and belonging. It represents a repository of ever-accumulating collective and individually held values—shared perceptions, experiences, life histories, beliefs and traditions. These social or private values are mostly ascribed by people to familiar places within this environment based on the ontological security which this everyday heritage provides. However, these values are notoriously hard to capture and categorize. This makes it difficult to incorporate them into heritage-management strategies, which typically rely on objective, fact-based datasets. In this paper, we present a new methodology to capture those elusive values, by combining Topic Modelling with the principles of Grounded Theory. Results show that our novel approach is viable and replicable and that these important values can be effectively and meaningfully integrated, thus creating more inclusive approaches to heritage management than exist currently.

Debating AI in Archaeology: applications, implications, and ethical considerations

“…Recent developments and the rapid adoption of AI technology into archaeology and heritage practice, as presented here, show the need for a debate around ethical implications and sustainable applications of AI. To enable the discourse, we have presented the advantages and capabilities of the applications, which allow more time and resource efficient workflows (Tenzer 2022;Tenzer and Schofield 2023), and enable the analysis and reuse of 'big data' accumulated over decades of archaeological investigations currently dormant in archives and grey literature (Brandsen and Lippok 2021). We also provide different views on the implications of AI applications from archaeology, heritage studies, data science and philosophy, showing inherent challenges regarding limitation, bias and social impact (Bender et al 2021;Casilli 2019;Crawford 2021;Véliz 2021).…”

“…A current cultural heritage project applied NLP, in particular Topic Modelling (TM) and ML, to explore the values attributed by people to familiar cultural landscapes (Tenzer 2022;Tenzer and Schofield 2023). Social media data, online surveys, and interviews provided sufficiently large datasets to infer heritage values from a 'bottomup' or people-centred perspective.…”

“…There is also a need for explainability and transparency in the approach to data collection in qualitative research. As shown in the heritage case study (Tenzer 2022;Tenzer and Schofield 2023), AI can help analyse vast amounts of social media data or survey responses. However, generating models based on such data can introduce or reinforce biases by excluding already marginalised groups for example.…”

Debating AI in archaeology: applications, implications, and ethical considerations

“…Recent developments and the rapid adoption of AI technology into archaeology and heritage practice, as presented in this paper, show the importance of a debate around ethical implications and sustainable applications of AI. To enable the discourse, we have presented the advantages and capabilities of the applications, which allow more time and resource efficient workflows (Tenzer 2022;Tenzer and Schofield 2023), and enable the analysis and reuse of 'big data' accumulated over decades of archaeological investigations lying dormant in archives and grey literature (Brandsen and Lippok 2021). Furthermore, we provide different views on the implications of AI applications from archaeology, heritage studies, data science and philosophy, showing inherent challenges regarding limitation, bias, and social impact (Bender et al 2021;Casilli 2019;Crawford 2021;Véliz 2021).…”

“…A current cultural heritage project applied NLP and in particular Topic Modelling (TM) and ML to explore the values attributed by people to familiar cultural landscapes (Tenzer 2022;Tenzer and Schofield 2023). Social media data, online surveys, and interviews provided sufficiently large datasets to infer heritage values from a "bottom-up" or people-centred perspective.…”