Outputs

Journal and proceedings publications

The project supports a range of interdisciplinary research linking political science, computer science, web science, economics and sociology. Wide-ranging outputs link core themes of the project and evolving collaborations on new and emerging subjects for research and innovation.

Two Steps Forward, One Step Back: The Evolution of Democratic Digital Innovations in Podemos

Marco Meloni & Fabio Lupato

South European Society and Politics (2022), 27:2, 253-278, DOI: 10.1080/13608746.2022.2161973 

While the digitalisation of political parties is increasingly analysed, less attention has been paid to the evolution of digital procedures and their consequences on intra-party democracy and party change. We propose a typology for identifying different types of evolution processes (consolidation, reconfiguration, mutation, and elimination) using the Spanish party Podemos paradigmatic case. Our analysis points out the centrality of hard and soft setbacks in the evolution of the digital procedures of the party. Findings indicate the relevance of different dynamics, such as institutionalisation, personalisation, and factionalism, jointly with other internal and external factors. Studying the evolution of digital party procedures is relevant for tracing party change in digital parties and other parties that are experiencing digitalisation processes.

Benchmark Evaluation for Tasks with Highly Subjective Crowdsourced Annotations: Case study in Argument Mining of Political Debates

Rafael Mestre, Matt Ryan, Stuart E. Middleton, Richard Gomer, Masood Gheasi, Jiatong Zhu, Timothy J. Norman (2023)

This paper assesses the feasibility of using crowdsourcing techniques for subjective tasks, like the identification of argumentative relations in political debates, and analyses their inter-annotator metrics, common sources of error and disagreements. We aim to address how best to evaluate subjective crowdsourced annotations, which often exhibit significant annotator disagreements and contribute to a "quality crisis" in crowdsourcing. To do this, we compare two datasets of crowd annotations for argumentation mining performed by an open crowd with quality control settings and a small group of master annotators without these settings but with several rounds of feedback. Our results show high levels of disagreement between annotators with a rather low Krippendorf's alpha, a commonly used inter-annotator metric. This metric also fluctuates greatly and is highly sensitive to the amount of overlap between annotators, whereas other common metrics like Cohen's and Fleiss' kappa are not suitable for this task due to their underlying assumptions. We evaluate the appropriateness of the Krippendorf's alpha metric for this type of annotation and find that it may not be suitable for cases with many annotators coding only small subsets of the data. This highlights the need for more robust evaluation metrics for subjective crowdsourcing tasks. Our datasets provide a benchmark for future research in this area and can be used to increase data quality, inform the design of further work, and mitigate common errors in subjective coding, particularly in argumentation mining.

You can access the paper here.

The credibility of regional policymaking: insights from South America

Pia Riggirozzi and Matt Ryan

Globalizations (2022), 19:4, 604-619, DOI: 10.1080/14747731.2021.1893530

Regional responses to international threats and opportunities have become prevalent. Yet significant lacunae persist in our recognition of some impacts of regional policies. Where measuring impact of regional policy does exist, this tends to focus on economic impact, heavily influenced by the EU model, or focuses on impacts that we have come to expect from national level policymaking. In this paper we show that important effects of regional policymaking in South American regionalism are being overlooked. Specifically, we argue that failure to recognize the creation of normative frameworks, dynamics of regional diplomacy, and social policy outcomes from regional policymaking has negatively affected the credibility of regional organizations. Lack of understanding of the impact of regional policy risks trivializing progress in regional policy and, critically, affects the legitimacy and credibility of regional governance. We make this case by focusing on regional health and migration policies in South America.

Place-based Politics and Nested Deprivation in the U.K.: Beyond Cities-towns, ‘Two Englands’ and the ‘Left Behind’

John Boswell, John Denham, Jamie Furlong, Anna Killick, Patricia Ndugga, Beata Rek, Matthew Ryan and Jesse Shipp

Representation (2020), 1-22. DOI: https://doi.org/10.1080/00344893.2020.1751258

‘Place-based explanations’ of politics in the U.K. tell sweeping narratives about ‘Two Englands’, or of sizeable regions of the country that have been ‘Left Behind’, reinforcing popular accounts of a North–South or city-town divide. We introduce the concept of nested deprivation – deprivation that may occur in just one housing estate or even one row of flats within neighbourhoods that are otherwise affluent. We report on intensive fieldwork in 8 neighbourhoods varying in relative affluence and density of population (including urban, suburban/satellite, market town or rural village). Three key themes and consequences emerge for those living in nested deprivation in relatively affluent and geographically dispersed contexts: (a) either disconnection from or entrapment within the local economy; (b) social isolation and atomisation; and (c) powerlessness to affect politics. ‘Place-based’ explanations of rapid and radical changes to political participation in Britain need to take fine-grained geographical distinctions much more seriously. Our study provides evidence that the rising tides in affluent areas are drowning some residents rather than lifting all boats. Where deprivation is dispersed and then nested within mostly affluent constituencies it does not allow for the political mobilisation among communities of interest that is a necessary condition for pluralist representative democracies.

Cultural capital and income inequality across Italian regions

Annie Tubadji, Masood Gheasi, Alessandro Crociata, and Iacopo Odoardi

Regional Studies (2021). Volume 56, 459-475. DOI: https://doi.org/10.1080/00343404.2021.1950913

Deliberation is an essential component of a healthy democracy. Through deliberation citizens listen to, they learn from, and they engage with different discourses. Furthermore, diversity (especially inclusion of different gender, race and ethnicity) is very important in a deliberative body. Among a larger volume of research on diversity and in particular gender differences in deliberation, researchers find significant gender gaps in participation and inclusion. Interventions have been aimed at optimizing facilitation of equal voice in a deliberative discussion group. The aim of our study is to test whether the pooling of study results that are individually inconclusive, may be able to jointly explain differences in gender deliberation. There are a number of qualitative studies on gender inequality in deliberation, but this topic is relatively understudied using advanced large-N statistical methods. Furthermore, existing quantitative research articles have faced limitations in empirical estimation, causal analysis, achieving straightforwardly interpretable results etc.). Still, we succeeded in gathering 13 studies of a rich quality which yielded 207-point estimates allowing a quantitative meta-analysis on gender inequality on deliberation. In order to gather a representative set of published papers in different publication outlets, we selected from various political science literature databases all refereed articles that included an estimation of deliberation in which gender has been included as an explanatory variable. Our database only includes papers written in the English language, and we evaluate sources of bias. The set of papers were selected by means of Google search engine, Google scholar, snowballing techniques, and intensive search in political science journals. Aggregated results reveal a more nuanced picture regarding gender gaps in deliberation. Surprisingly, the aggregated effect size of gender is almost zero, but with a negative sign. However, our results confirms that the condition in which deliberation happens is more important than the critical mass. These conditions are the majority rule and the unanimous rule.   

M-Arg: MultiModal Argument Mining Dataset for Political Debates with Audio and Transcripts

Rafael Mestre, Razvan Milicin, Stuart E. Middleton, Matt Ryan, Jiatong Zhu, and Timothy J. Norman

Proceedings of the 8th Workshop on Argument Mining, 2021 https://aclanthology.org/2021.argmining-1.8/

Argumentation mining aims at extracting, analysing and modelling people's arguments, but large, high-quality annotated datasets are limited, and no multimodal datasets exist for this task. In this paper, we present M-Arg, a multimodal argument mining dataset with a corpus of US 2020 presidential debates, annotated through crowd-sourced annotations. This dataset allows models to be trained to extract arguments from natural dialogue such as debates using information like the intonation and rhythm of the speaker. Our dataset contains 7 hours of annotated US presidential debates, 6527 utterances and 4104 relation labels, and we report results from different baseline models with highest accuracy of 0.86 with a multimodal model.

Community Control in the Housing Commons: A Conceptual Typology

Catherine Durose, Liz Richardson, Max Rozenburg, Matt Ryan and Oliver Escobar

International Journal of the Commons (2021), 15(1), 291–304. DOI: http://doi.org/10.5334/ijc.1093

We advance theorizing on the governance of the commons through a configurative comparative analysis (CCA) of community control in the housing commons. We focus our analysis on community land trusts (CLTs), which are increasingly recognised as a potential governance mechanism for collective access to housing provision for low-income communities. Through systematic comparative analysis of CLTs in the US and UK, we extend the existing evidence base and develop a conceptual typology of community control in the housing commons. The typology suggests that whilst some social purposes for CLTs may align with notions of the commons – enrichment of community politics, conservation of community life, or creation of participatory governance – other CLTs focus on housing provision as a means of making a broader contribution to the social economy, or as an asset-lock to enable wider provision for affordable housing. By understanding this differentiation, we challenge the assumption that design principles or governance mechanisms are sufficient for or inherently offer a singly clear route to community control, and recognise that community control is achieved through different pathways informed by the multiple configurations of dynamics between different aspects of governance, as usefully illuminated by CCA. Our approach demonstrates the value to scholarship and activism on the commons of systematic comparative analysis in order to interrogate the expansion of the commons not only in practice but in spirit.

Augmenting pre-trained language models with audio feature embedding for argumentation mining in political debates

Mestre, R., Middleton, S. E., Ryan, M., Gheasi, M., Norman, T. J. & Zhu, J. (2023)

The integration of multimodality in natural language processing (NLP) tasks seeks to exploit the complementary information contained in two or more modalities, such as text, audio and video. This paper investigates the integration of often under-researched audio features with text, using the task of argumentation mining (AM) as a case study. We take a previously reported dataset and present an audio-enhanced version (the Multimodal USElecDeb60To16 dataset). We report the performance of two text models based on BERT and GloVe embeddings, one audio model (based on CNN and Bi-LSTM) and multimodal combinations, on a dataset of 28,850 utterances. The results show that multimodal models do not outperform text-based models when using the full dataset. However, we show that audio features add value in fully supervised scenarios with limited data. We find that when data is scarce (e.g. with 10% of the original dataset) multimodal models yield improved performance, whereas text models based on BERT considerably decrease performance. Finally, we conduct a study with artificially generated voices and an ablation study to investigate the importance of different audio features in the audio models.

You can access the paper here.