scieee AI-readable full text Open interactive document viewer

DigAiPo+: Digitalisation and Artificial Intelligence Party Positions

Guglielmo, Marco; Teruel, Andreu

Abstract

The dataset DigAiPo+ contains classification of textual data sources from parties’ official documents and statements concerned with the digitalisation processes and artificial intelligence, in the broadest possible sense. This dataset releases an initial sample comprising approximately 15% of all sentences, drawn from five countries that account for a considerable portion of the total observations. Future versions will include an extended dataset.

Full text

Dataset Digitalisation and Artificial Intelligence: Party Positions+ DigAiPo+ Codebook Version 1.0 13 Oct 2025 Introduction The dataset DigAiPo+ contains classification of textual data sources from parties’ official documents and statements concerned with the digitalisation processes and artificial intelligence, in the broadest possible sense. This codebook provides information on: • Data sources and routes for integration with existing datasets. • Coding procedures, including but not limited to the integration of human and automated classification, intercoder reliability tests and validation. • Content Analysis: codes’ descriptions, including examples of synthetic data. • Indexes of interest to the database authors. • Acknowledgement of research team and coders. The DigAiPo+ codebook and database are available open access at Zenodo. Citations When using the dataset and/or this codebook please cite: Guglielmo, M., Teruel A., (2025) ‘Digitalisation and Artificial Intelligence: Party Positions+. The DigAiPo+ Dataset’, v. 1.0. DOI: https://doi.org/10.5281/zenodo.17345008 Aknowledgements: We gratefully acknowledge the valuable contribution of the student assistants from the Department of Law, Political and Administrative Sciences for their support with data coding and processing activities: Ida Halm Mario Navarro Ana María Sánchez Blasco Antonio Sorni Data Sources We code textual data from parties’ official documents and statements. These are primarily party manifestos at statewide general elections, if not indicated otherwise. Data Overview Unit Manifesto sentence Countries (Published in red) Sweden, Norway, Denmark, Finland, Belgium, Netherlands, Luxembourg, France, Italy, Spain, Greece, Portugal, Germany, Austria, Switzerland, United Kingdom, Ireland, Cyprus, United States, Canada, Australia, New Zealand, Turkey, Bulgaria, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Moldova, Poland, Romania, Russia, Slovakia, Slovenia, Ukraine, Argentina, Bolivia, Colombia, Costa Rica, Ecuador, Chile, Panama, Uruguay, Dominican Republic, Mexico, Peru, Brazil, South Africa Elections 1639 Manifestos Coded Quasi Sentences 57937 Data Sources Comparative Manifesto Project (Lehmann et al. 2024) Language English or translations to English by Lehmann et al. (2024) Documents’/Statements’ Variables To secure data interoperability, DigAiPo+ adopts the same coding by the Comparative Manifesto Project. Party name and code (CMP) (11110) Green Ecology Party (11220) Left Party (11320) Social Democratic Labour Party (11420) Liberal People’s Party … N manifesto_id Election date + party date 11110_200609 (Green Ecology Party) 11220_200609 (Left Party) 11320_200609 (Social Democratic Labour Party) 11420_200609 (Liberal People’s Party) cmp_code … N Coding provided by CMP text pre post Quasi sentences including reference to digital technologies Previous paragraph Subsequent paragraph Coding Procedures and methodological notes DigAiPo+ results from a mixed human/machinic coding process. We extract data from the Comparative Manifesto Project corpus (Lehman et al. 2024). Through a dictionary query, we filter the corpus. We employ the following words: 'digital', 'cyber', 'internet', 'online', 'platform', 'information communication technolog', 'social media', 'ICT', 'e-govern', 'egovern', 'e-democr', 'electronic democr', 'open govern', 'opengover', 'broadband', '2.0', '3.0', '4.0', 'industry 4.0', 'robot', 'cloud', 'software', 'hardware', 'data', 'AI', 'artificial intelligen', 'big-tech', 'bigtech', 'GAFAM', 'tech', 'techno', 'net', 'network', 'green-tech', 'greentech', 'fin-tech', 'fintech', 'uber-', 'rider', 'privacy', 'facial recognition', 'electronic', 'Digital', 'Cyber', 'Internet', 'Online', 'Platform', 'Information Communication Technolog', 'Social Media', 'e-Govern', 'Egovern', 'e-Democr', 'Electronic Democr', 'Open Govern', 'Opengover', 'Broadband', '2.0', '3.0', '4.0', 'Industry 4.0', 'Robot', 'Cloud', 'Software', 'Hardware', 'Data', 'Artificial Intelligen', 'Big-Tech', 'Bigtech', 'GAFAM', 'Tech', 'Techno', 'Net', 'Network', 'Green-Tech', 'Greentech', 'Fin-Tech', 'Fintech', 'Uber-', 'Rider', 'Privacy', 'Facial Recognition', 'Electronic'. Coding has unfolded through the following steps and procedures: 1Manual coding of 4000 sentences by five different coders. Each coder with a superposition of 500 sentences with other ones. 2Non-matching criteria was qualitatively discussed to provide a ‘gold’ standard sample dataset. 3Text embeddings were generated using the ‘text-embedding-3-large’ model to represent all samples in the dataset that included gold standard labels. A retriever was built using the Nearest Neighbors algorithm to identify examples with semantically similar text among the labeled data. 4For each new text to be classified, the most similar gold standard examples were retrieved and incorporated as few-shot examples in the model prompt. The GPT-5o-mini model (OpenAI) was used for text classification. The prompt included: - a description of the available labels (label book); - rules for disambiguating closely related categories; - and the retrieved few-shot examples. A self-consistency strategy was applied (three shots), in which the model performed the classification three times using a moderate temperature (0.4) to allow for controlled variability. The final label was determined by majority vote across the three outputs." Human Coders 5 coders LLM Coding OpenAI model: “text-embedding-3-large” and “gpt-5-mini-2025-0807” Reliability The algorithm achieves an F1-score above 0.85 across all codes, which is particularly remarkable given the complexity of the variable 1_Digi_Topic, encompassing 38 possible values. Content Analysis DigAiPo+ results from the coding of quasi-sentences to multiple dimensions. The coding categories are the result of a four-step process: 1. The research team has inferred from literature the widest range of classifications of parties’ and/or other political groups statements on digitalisation, digital transition, platform society, artificial intelligence, etc. 2. These categories have been refined based on researchers’ interests and expertise. 3. The list of code went through a first round of refinement based on piloting a sample of the dataset (N= 1000). 1_ Digi_Topic Definition: The primary substantive issue or policy area addressed in the statement. Specific Coding Instructions: - Assign one value per quasi-sentence - Select the most specific applicable topic. 1_00 Automation and Robotisation Mentions of socioeconomic and political changes due to automation and robotisation Examples - ‘Our Industry 4.0 plan will support growth through robotisation’. - ‘Automation and robotisation put future employment at risk’. 1_01 Childhood Digital Primary focus on the effects/impacts/risks of digital platforms on children. Examples - ‘We propose new regulations for children to access social media’. - ‘Our party will protect children from online harm and violence’. 1_02 Cyber Warfare Mentions of hi-tech weapons and or defence mechanisms. Examples - ‘Cyber warfare poses new challenges to our national defence systems’. - ‘We plan to raise the defence budget for our army to have the most advanced drones’. 1_03 Cybersecurity Focus on individual online security and/or institutional means to guarantee safety. Examples - ‘We must protect our nation from foreign intrusions in our digital systems’. - ‘Our party has a balanced approach on cybersecurity to balance freedom and safety’. 1_04 Cyberviolence The primary focus is on the sources, targets, and modes of spread of online violence. Examples - ‘We need new regulations and get tougher on cyberviolence, especially when targeting the most vulnerable’. - ‘The party stands against the spread of online hate speech that may promote violent crimes in real life’. 1_05 Data Governance Mentions of how institutions address issues surrounding data regulation, storage, property etc. Examples - ‘Our proposal is to invest in national data centres to protect our citizens’. - Data is the new gold of the 21st century and should not be let in the hands of few companies. 1_06 Digital Access Statements on issues surrounding access, lack thereof, to varied digital systems. Examples - We need to guarantee that those without digital skills can access essential services. - We need to reduce the digital divide between rural and urban areas. 1_07 Digital Border Control Digital systems may improve border protection or harm human rights. Examples - Our party firmly stands against attacks on liberty, such as biometric controls at the border. - With new technologies such as blockchain, we can prevent irregular immigration. 1_08 Digital Commons Mentions of digital platforms as non-exclusive, shared, democratically governed spaces. Examples - Platforms as Wikipedia demonstrate how digitalisation can prompt the sharing of knowledge and common goods. - Digitalisation breaks the barriers between public and private, fostering a new culture of the commons. 1_09 Digital Copyright Primary focus on how digitalisation modifies copyright and intellectual property. Examples - Peer-to-peer platforms pose new challenges for protecting copyright over cultural products. - Digitalisation opens new opportunities for sharing knowledge, overcoming outdated copyright laws. 1_10 Digital Culture Statements concerned with cultural production/diffusion/consumption in online spaces. Examples - We will create a public platform connecting our national cultural sites. - We must protect our cultural workers against the greed of the digital cultural industry. 1_11 Digital Education Education is mentioned as a site of digitalisation and/or digital platforms foster new learning processes. 2_06 Supranational/International Arenas Reference and/or implications for supra/international institutions and arenas. Examples “The European Union must advance with protecting citizens' online privacy.” 3_ Techno_Domain Definition: The primary technological system that the statement refers to. Specific Coding Instructions: - Code the domain most directly referenced or affected. - Use ‘Other’ with parsimony only when there is no primary technological system. 3_00 Algorithms and Big Data Reference to algorithms, big data and other systems presiding over the functioning of platform societies. Examples “Artificial Intelligence brings new risks and opportunities.” 3_01 Artificial Intelligence Reference to Artificial Intelligence systems and their political or public policy implications. Examples “We propose to connect on Internet all school libraries across the country.” 3_02 Commercial Internet Reference to the Internet and their political or public policy implications. Examples “We need more regulations to make social media companies accountable for the spread of fake news.” 3_03 Digital Media and Telecommunications Reference to digital media, including the so-called 'social media' stage of digitalisation and/or telecommunication infrastructure. Examples “Our Industry 4.0 plan will support growth through robotisation.” 3_04 Robot and Internet of Things Reference to algorithmically-driven machines, including but not limited to applications to industry, public spaces and policing operations. Examples “We plan to raise defense budget for our army to have the most advanced drones.” 3_05 Digital Military Equipment Reference to digitalised, robotic and automatic military equipment. Examples “The digital revolution is dramatically changing how we work, consume, communicate.” 3_06 Other (Multiple/Unspecified) Generic reference to digital platforms, digital revolution etc. without proper technological target. Examples 4_ Digi_Statement Definition: The primary communication/political goal that the statement implicates. Specific Coding Instructions: - Code whether the party is expressing its own position on an issue/matter or affirming its capacity to manage that issue/matter effectively. - In case of ambiguity, look at the context of the statement. What is the party prioritising? 4_00 Position The statement expresses party's positioning within its broader belief system. Examples “Our democracy can renew representation thanks to digital communication between citizens and politicians.” “Hate speech is particularly harmful for women and sexual minorities and this is why we propose new regulations.” 4_01 Valence The statement expresses party's valence, that is stating the capacity to manage digitalisation effectively. Examples “We have incentivised the creation of new companies in the tech sector to be more competitive.” “Thanks to our smart city plans, we have reduced pollution and improved mobility.” 5_ Digi_Orientation Definition: The primary direction where the statement points to. Specific Coding Instructions: - Is the party proposing a reflection on an existing state of things or proposing something? - In case of ambiguity, or co-existence in the same statement, look at the context to infer priority. 5_00 Proposal The statement is oriented towards proposal formulation, ranging from general to specific. Examples “Our digital and green plans will create hundreds of thousands of new jobs.” “We propose tax breaks for our innovative start-ups in digital and green technologies.” 5_01 Status Quo The statement is a reflection over a status quo, from past or present conditons. Examples “We are now living in the age of disinformation which affects the quality of information and journalism.” “Big platforms force citizens to give consent without transparent information.” 6_ Digi_Agreement Definition: The primary positioning of the statement in relation to a topic, domain, actor etc. Specific Coding Instructions: - With this code, we interpret the relations of party statements with given topics, actors and/or domains. - In case of a statement that could be both an agreement on one digi_topic code and disagreement on the other, refer to context. What is the primary objective of the statement? 6_00 Strongly Agree Explicit statement or implicit logical assumption of strong agreement with a given situation or others' positions. Examples “Our digitalisation plan will allow to de-bureaucratise the state for citizens and companies.” “Our position is that critical data infrastructure must be nationalised.” 6_01 Agree Explicit statement or implicit logical assumption of agreement with a given situation or others' positions. Examples “Digital economy allows citizens to become entrepreneurs with less frictions and risks.” “We propose to regulate e-waste and the recycling of lithium batteries as they can harm our environment.” 6_02 Neutral The statement neither expresses agreement or disagreement on given topics, domains, positions etc. Examples “Our party supports the digital economy and competitiveness, while seeking to protect citizens' data privacy.” “We stand on the side of innovation, for instance supporting digitalisation.” 6_03 Disagree Explicit statement or implicit logical assumption of disagreement with a given situation or others' positions. Examples “Our party opposes the adoption of means for facial recognition by police.” “According to a recent report, digitalisation may cause the loss of 1 million jobs in the next five years.” 6_04 Strongly Disagree Explicit statement or implicit logical assumption of strong disagreement with a given situation or others' positions. Examples “Our party opposes the sale of medical records to private tech companies.” “Our party firmly stands against attacks to liberty such as biometric controls at the border.” 7_ Digi_Centrality Definition: The relative relevance and centrality of digitalisation in relation to other issues and or positions, Specific Coding Instructions: - The goal is to ascertain whether digitalisation acquire the status of a central issue. - In case of ambiguity, think counter-intuitively: what will happen to the statement if there is no reference to digitalisation? 7_00 Adjacent Digitalisation and another issue/topic mutually shape each other meaning standing side-by-side. Examples “Thanks to new technologies, our national police can collect information and prevent crime.” 7_01 Core The reference to digitalisation has meaning indipendently of other issues/topics. Examples “Data is the new gold of the 21st century and should not be let in the hands of few companies.” 7_02 Marginal Digitalisation is a marginal issue/topic to provide additional signification to a different core issue/topic. Examples “We support creativity and the invention of our entrepreneurs who compete in digital markets.” 8_ Digi_Context_Pre Definition: The relative relevance and centrality of digitalisation in the textual context (previous paragraph). Specific Coding Instructions: - Skimming through context is vital to understand what the statement refer to, in terms of logical implications, sequences etc. - Coding focuses on interpreting the presence/absence and centrality/marginality of digitalisation in any of the parts of the previous paragraph. 8_00 Adjacent Digitalisation and other issues/topics mutually shape each other, meaning they stand side-by-side. 8_01 Core The reference to digitalisation has meaning independently of other issues/topics. 8_02 Marginal Digitalisation is a marginal issue/topic to provide additional signification to different core issues/topics. 8_03 None There is no mention to digitalisation. 9_ Digi_Context_Post Definition: The relative relevance and centrality of digitalisation in the textual context (previous paragraph). Specific Coding Instructions: - Skimming through context is vital to understand what the statement refer to, in terms of logical implications, sequences etc. - Coding focuses on interpreting the presence/absence and centrality/marginality of digitalisation in any of the parts of the following paragraph. 9_00 Adjacent Digitalisation and other issues/topics mutually shape each other, meaning they stand side-by-side. 9_01 Core The reference to digitalisation has meaning independently of other issues/topics. 9_02 Marginal Digitalisation is a marginal issue/topic to provide additional signification to different core issues/topics. 9_03 None There is no mention to digitalisation.