Environmental activism is crucial for raising public awareness and support toward addressing the climate crisis. However, using climate change mitigation as the cause for blockage, vandalism, and harassment activities might be counterproductive and risk causing negative repercussions and declining public support. The paper describes a dataset of metadata of 89 blockage, vandalism, and harassment events happening in recent years. The dataset comprises three main categories: 1) Events, 2) Activists, and 3) Consequences. For researchers interested in environmental activism, climate (...) change, and sustainability, the dataset is helpful in studying the effectiveness and appropriateness of strategies to raise public awareness and support. For researchers in the field of security studies and green criminology, the dataset offers resources to study features and impacts of blockage, vandalism, and harassment events. The Bayesian Mindsponge Framework (BMF) analytics was employed to validate the dataset. Consequently, the estimated result aligns with the Mindsponge Theory’s theoretical reasoning. (shrink)

All datasets emerge from and are enmeshed in power-laden semiotic systems. While emerging data ethics curriculum is supporting data science students in identifying data biases and their consequences, critical attention to the cultural histories and vested interests animating data semantics is needed to elucidate the assumptions and political commitments on which data rest, along with the externalities they produce. In this article, I introduce three modes of reading that can be engaged when studying datasets—a denotative reading, a connotative reading, and (...) a deconstructive reading. I then outline how I have taught students to engage these methods when analyzing three datasets in Data and Society—a course designed to cultivate student competency in politically aware data analysis and interpretation. I show how combined, the reading strategies prompt students to grapple with the double binds of perceiving contemporary problems through systems of representation that are always situated, incomplete, and inflected with diverse politics. While I introduce these methods in the context of teaching, I argue that the methods are integral to any data practice in the conclusion. (shrink)

AGGA (Academic Guidelines for Generative AIs) is a dataset of 80 academic guidelines for the usage of generative AIs and large language models in academia, selected systematically and collected from official university websites across six continents. Comprising 181,225 words, the dataset supports natural language processing tasks such as language modeling, sentiment and semantic analysis, model synthesis, classification, and topic labeling. It can also serve as a benchmark for ambiguity detection and requirements categorization. This resource aims to facilitate research (...) on AI governance in educational contexts, promoting a deeper understanding of the integration of AI technologies in academia. (shrink)

The Sense Complexity Dataset (SeCoDa) provides a corpus that is annotated jointly for complexity and word senses. It thus provides a valuable resource for both word sense disambiguation and the task of complex word identification. The intention is that this dataset will be used to identify complexity at the level of word senses rather than word tokens. For word sense annotation SeCoDa uses a hierarchical scheme that is based on information available in the Cambridge Advanced Learner’s Dictionary. This (...) way we can offer more coarse-grained senses than directly available in WordNet. (shrink)

Bayesian methods are ubiquitous in contemporary observational cosmology. They enter into three main tasks: cross-checking datasets for consistency; fixing constraints on cosmological parameters; and model selection. This article explores some epistemic limits of using Bayesian methods. The first limit concerns the degree of informativeness of the Bayesian priors and an ensuing methodological tension between task and task. The second limit concerns the choice of wide flat priors and related tension between parameter estimation and model selection. The Dark Energy Survey and (...) its recent Year 1 results illustrate both these limits concerning the use of Bayesianism. (shrink)

IGGA (Industrial Guidelines/policy statements for Generative AIs) is a comprehensive dataset comprising 160 guidelines and policy statements pertaining to the use of generative AIs and large language models across 14 industry sectors. These guidelines were systematically selected and gathered from official company websites and reliable sources spanning six continents. The dataset, containing 295,692 words, is designed to support various natural language processing tasks, including language modeling, sentiment analysis, semantic analysis, model synthesis, classification, and topic labeling. Additionally, it serves (...) as a benchmark for tasks such as ambiguity detection and requirements categorization. This resource aims to advance research in AI governance within industrial settings, fostering a deeper understanding of how AI technologies are integrated into industry practices. (2024-11-17). (shrink)

This paper introduces a streamlined taxonomy for categorizing offensive language in Hebrew, addressing a gap in the literature that has, until now, largely focused on Indo-European languages. Our taxonomy divides offensive language into seven levels (six explicit and one implicit level). We based our work on the simplified offensive language (SOL) taxonomy introduced in (Lewandowska-Tomaszczyk et al. 2021a) hoping that our adjustment of SOL to the Hebrew language will be capable of reflecting the unique linguistic and cultural nuances of Hebrew. (...) The study involves both linguistic and cultural analysis beyond Natural Language Processing (NLP). We employed manual linguistic analysis to understand the nuances of offensive language in Hebrew. An accompanying dataset, gathered on Twitter and manually curated by human annotators, is described in detail. This dataset was constructed to both validate the taxonomy and serve as a foundation for future research on offensive language detection and analysis in Hebrew. Preliminary analysis of the dataset reveals intriguing patterns and distributions, underscoring the complexity and specificity of offensive expressions in the Hebrew language. The aim of our work is to capture the complexity and specificity of offensive expressions in Hebrew beyond what automated NLP methods alone can provide. Our findings highlight the significance of considering linguistic and cultural variations when researching and correcting abusive language online. We believe that our streamlined taxonomy and associated dataset will be crucial in improving research in Hebrew language sociocultural studies, natural language processing, and offensive language detection. Our study also makes a substantial contribution to the study of low-resource languages and can be used as a model for future research on other languages. (shrink)

Generative AI models are shaping our future. In this work, we discover and expose the bias against physically challenged people in generative models. Generative models (Stable Diffusion XL and DALL·E 3) are unable to generate content related to the physically challenged, e.g., inclusive washroom, even with very detailed prompts. Our analysis reveals that this disability bias emanates from biased AI datasets. We achieve this using a novel strategy to automatically discover bias against underrepresented groups like the physically challenged. Finally, we (...) track the root of this disability bias in search engines (Google, Bing, Yandex, and DuckDuckGo). Search engines suffer from disability bias for neutral prompts. The standard strategy of using synonyms to retrieve diverse results does not automatically include the physically challenged. Search engines require specific mention of the underrepresented group to retrieve relevant results. Therefore, conscious effort is required to include underrepresented groups while scraping datasets from search engines. A future built by generative AI having little understanding of physically challenged people will have serious implications. We hope that our effort lays the groundwork for future datasets and algorithms that include this underrepresented group. (shrink)

Temporal dynamics have been increasingly recognized as an important component of facial expressions. With the need for appropriate stimuli in research and application, a range of databases of dynamic facial stimuli has been developed. The present article reviews the existing corpora and describes the key dimensions and properties of the available sets. This includes a discussion of conceptual features in terms of thematic issues in dataset construction as well as practical features which are of applied interest to stimulus usage. (...) To identify the most influential sets, we further examine their citation rates and usage frequencies in existing studies. General limitations and implications for emotion research are noted and future directions for stimulus generation are outlined. (shrink)

In the area of machine learning, different techniques are used to train machines and perform different tasks like computer vision, data analysis, natural language processing, and speech recognition. Computer vision is one of the main branches where machine learning and deep learning techniques are being applied. Optical character recognition is the ability of a machine to recognize the character of a language. Pashto is one of the most ancient and historical languages of the world, spoken in Afghanistan and Pakistan. OCR (...) application has been developed for various cursive languages like Urdu, Chinese, and Japanese, but very little work is done for the recognition of the Pashto language. When it comes to handwritten character recognition, it becomes more difficult for OCR to recognize the characters as every handwritten character’s shape is influenced by the writer’s hand motion dynamics. The reason for the lack of research in Pashto handwritten character data as compared to other languages is because there is no benchmark dataset available for experimental purposes. This study focuses on the creation of such a dataset, and then for the evaluation purpose, a machine is trained to correctly recognize unseen Pashto handwritten characters. To achieve this objective, a dataset of 43000 images was created. Three Feed Forward Neural Network models with backpropagation algorithm using different Rectified Linear Unit layer configurations were trained and tested with this dataset. The simulation shows that Model 1 achieved accuracy up to 87.6% on unseen data while Model 2 achieved an accuracy of 81.60% and 3% accuracy, respectively. Similarly, loss was the lowest for Model 1 with 0.15 and 3.17 for training and testing, followed by Model 2 with 0.7 and 4.2 for training and testing, while Model 3 was the last with loss values of 6.4 and 3.69. The precision, recall, and f-measure values of Model 1 were better than those of both Model 2 and Model 3. Based on results, Model 1 is found to be the most efficient as compared to the other two models in terms of accuracy to recognize Pashto handwritten characters. (shrink)

This paper presents the first dataset for Japanese Legal Judgment Prediction (LJP), the Japanese Tort-case Dataset (JTD), which features two tasks: tort prediction and its rationale extraction. The rationale extraction task identifies the court’s accepting arguments from alleged arguments by plaintiffs and defendants, which is a novel task in the field. JTD is constructed based on annotated 3477 Japanese Civil Code judgments by 41 legal experts, resulting in 7978 instances with 59,697 of their alleged arguments from the involved (...) parties. Our baseline experiments show the feasibility of the proposed two tasks, and our error analysis by legal experts identifies sources of errors and suggests future directions of the LJP research. (shrink)

Background: Academic productivity has been studied by scholars all round the world for many years. However, in Vietnam, this topic has scarcely been addressed. This research therefore aims at better understanding the correlations between gender, age, research experience, the leading role of corresponding authors, and the total number of their publications in the specific realm of social sciences and humanities. Methods: The study employed a Scopus dataset with publication profiles of 410 Vietnamese researchers between 2008 and 2017. Results: Men (...) did not differ from women in academic publications (P=0.827). The proficiency of corresponding authors positively correlated with the number of published papers (rs=0.61, P<0.001). Lastly, the age of lead authors strongly correlated with scientific output (rs=0.74, P<0.001 for authors between 40 and 50 years of age). Conclusion: While scientific output correlated with the author ages and number of articles in which they led, it was not correlated with their gender in Vietnamese social science and humanities authors. (shrink)

Big Data has been variously defined in the literature. In the main, definitions suggest that Big Data possess a suite of key traits: volume, velocity and variety, but also exhaustivity, resolution, indexicality, relationality, extensionality and scalability. However, these definitions lack ontological clarity, with the term acting as an amorphous, catch-all label for a wide selection of data. In this paper, we consider the question ‘what makes Big Data, Big Data?’, applying Kitchin’s taxonomy of seven Big Data traits to 26 datasets (...) drawn from seven domains, each of which is considered in the literature to constitute Big Data. The results demonstrate that only a handful of datasets possess all seven traits, and some do not possess either volume and/or variety. Instead, there are multiple forms of Big Data. Our analysis reveals that the key definitional boundary markers are the traits of velocity and exhaustivity. We contend that Big Data as an analytical category needs to be unpacked, with the genus of Big Data further delineated and its various species identified. It is only through such ontological work that we will gain conceptual clarity about what constitutes Big Data, formulate how best to make sense of it, and identify how it might be best used to make sense of the world. (shrink)

This paper presents a streamlined taxonomy for categorizing offensive language in Arabic, specifically Modern Standard Arabic (MSA) and the Levantine dialect. Addressing a gap in the existing literature, which has mainly focused on Indo-European languages, our taxonomy divides offensive language into seven levels (six explicit and one implicit). We adapted our framework from the simplified offensive language (SOL) taxonomy by (Lewandowska-Tomaszczyk, Barbara, Slavko Žitnik, Anna Bączkowska, Chaya Liebeskind, Jelena Mitrovic & Giedre Valunaite Oleškeviciente. 2021a. Lod-connected offensive language ontology and tagset (...) enrichment. In Shubert R. Carvalho & Renato R. Souza (eds.), Proceedings of the workshops and tutorials held at ldk 2021 co-located with the 3rd language, data and knowledge conference, Vol. 3064, 135–150. CEUR Workshop Proceedings), customizing it to reflect the unique linguistic and cultural nuances of Arabic. To validate this taxonomy, we created a new dataset from various social media platforms, primarily focusing on Twitter. This dataset was manually curated by human annotators and is described in detail within the paper, serving as both a validation tool for our taxonomy and a foundation for future research on offensive language detection in Arabic. Initial analysis of the dataset reveals complex patterns of offensive expressions in MSA and Levantine Arabic, underscoring the need to account for linguistic and cultural variations in studying online abuse. Our taxonomy and dataset are vital for advancing research in Arabic sociocultural studies, natural language processing, and linguistic analysis, and contribute to the study of low-resource languages. (shrink)

In climate science, observational gridded climate datasets that are based on in situ measurements serve as evidence for scientific claims and they are used to both calibrate and evaluate models. However, datasets only represent selected aspects of the real world, so when they are used for a specific purpose they can be a source of uncertainty. Here, we present a framework for understanding this uncertainty of observational datasets which distinguishes three general sources of uncertainty: (1) uncertainty that arises during the (...) generation of the dataset; (2) uncertainty due to biased samples; and (3) uncertainty that arises due to the choice of abstract properties, such as resolution and metric. Based on this framework, we identify four different types of dataset ensembles — parametric, structural, resampling, and property ensembles—as tools to understand and assess uncertainties arising from the use of datasets for a specific purpose. We advocate for a more systematic generation of dataset ensembles by using these sorts of tools. Finally, we discuss the use of dataset ensembles in climate model evaluation. We argue that a more systematic understanding and assessment of dataset uncertainty is needed to allow for a more reliable uncertainty assessment in the context of model evaluation. The more systematic use of such a framework would be beneficial for both scientific reasoning and scientific policy advice based on climate datasets. (shrink)

In response to growing concerns of bias, discrimination, and unfairness perpetuated by algorithmic systems, the datasets used to train and evaluate machine learning models have come under increased scrutiny. Many of these examinations have focused on the contents of machine learning datasets, finding glaring underrepresentation of minoritized groups. In contrast, relatively little work has been done to examine the norms, values, and assumptions embedded in these datasets. In this work, we conceptualize machine learning datasets as a type of informational infrastructure, (...) and motivate a genealogy as method in examining the histories and modes of constitution at play in their creation. We present a critical history of ImageNet as an exemplar, utilizing critical discourse analysis of major texts around ImageNet’s creation and impact. We find that assumptions around ImageNet and other large computer vision datasets more generally rely on three themes: the aggregation and accumulation of more data, the computational construction of meaning, and making certain types of data labor invisible. By tracing the discourses that surround this influential benchmark, we contribute to the ongoing development of the standards and norms around data development in machine learning and artificial intelligence research. (shrink)

Language research has come to rely heavily on large‐scale, web‐based datasets. These datasets can present significant methodological challenges, requiring researchers to make a number of decisions about how they are collected, represented, and analyzed. These decisions often concern long‐standing challenges in corpus‐based language research, including determining what counts as a word, deciding which words should be analyzed, and matching sets of words across languages. We illustrate these challenges by revisiting “Word lengths are optimized for efficient communication” (Piantadosi, Tily, & Gibson, (...) 2011), which found that word lengths in 11 languages are more strongly correlated with their average predictability (or average information content) than their frequency. Using what we argue to be best practices for large‐scale corpus analyses, we find significantly attenuated support for this result and demonstrate that a stronger relationship obtains between word frequency and length for a majority of the languages in the sample. We consider the implications of the results for language research more broadly and provide several recommendations to researchers regarding best practices. (shrink)

Artificial intelligence (AI) and machine learning (ML) are becoming increasingly significant areas of research for scholars in science and technology studies (STS) and media studies. In March 2020, Waymo, Google/Alphabet’s autonomous vehicle project, introduced the ‘Open Dataset Virtual Challenge’, an annual competition leveraging their Waymo Open Dataset. This freely accessible dataset comprises annotated autonomous vehicle data from their own Waymo vehicles. Yearly, Waymo has continued to host iterations of this challenge, inviting teams of computer scientists to tackle (...) evolving machine learning and vision problems using Google's data and tools. This article analyses these challenges, situating them within the context of the ‘Grand Challenges’ of artificial intelligence (AI), which aimed to foster accountable and commercially viable advancements in the late 1980s. Through two exploratory workshops, we adopted a ‘technographic’ approach to examine the pivotal role of challenges in the development and political economy of AI. Serving as an organising principle for the AI innovation ecosystem, the challenge connects companies and external collaborators, driving advancements in specific machine vision domains. By exploring six key themes—interface methods, incrementalism, metrics, AI vernacular, applied domains, and competitive advantages—the article illustrates the role of these challenges in shaping AI research and development. By unpacking the dynamic interaction between data, computation, and labour, these challenges serve as catalysts propelling advancements towards self-driving technologies. The study reveals how challenges have historically and presently shaped the evolving landscape of self-driving and AI technologies. (shrink)

Consumer-based datasets are the products of data brokerage firms that agglomerate millions of personal records on the adult US population. This big data commodity is purchased by both companies and individual clients for purposes such as marketing, risk prevention, and identity searches. The sheer magnitude and population coverage of available consumer-based datasets and the opacity of the business practices that create these datasets pose emergent ethical challenges within the computational social sciences that have begun to incorporate consumer-based datasets into empirical (...) research. To directly engage with the core ethical debates around the use of consumer-based datasets within social science research, I first consider two case study applications of consumer-based dataset-based scholarship. I then focus on three primary ethical dilemmas within consumer-based datasets regarding human subject research, participant privacy, and informed consent in conversation with the principles of the seminal Belmont Report. (shrink)

This paper addresses a data integration problem: given several mutually consistent datasets each of which measures a subset of the variables of interest, how can one construct a probabilistic model that fits the data and gives reasonable answers to questions which are under-determined by the data? Here we show how to obtain a Bayesian network model which represents the unique probability function that agrees with the probability distributions measured by the datasets and otherwise has maximum entropy. We provide a general (...) algorithm, OBN-cDS, which offers substantial efficiency savings over the standard brute-force approach to determining the maximum entropy probability function. Furthermore, we develop modifications to the general algorithm which enable further efficiency savings but which are only applicable in particular situations. We show that there are circumstances in which one can obtain the model (i) directly from the data; (ii) by solving algebraic problems; and (iii) by solving relatively simple independent optimisation problems. (shrink)

Transformer-based Pre-Trained Language Models currently dominate the field of Natural Language Inference (NLI). We first survey existing NLI datasets, and we systematize them according to the different kinds of logical inferences that are being distinguished. This shows two gaps in the current dataset landscape, which we propose to address with one dataset that has been developed in argumentative writing research as well as a new one building on syllogistic logic. Throughout, we also explore the promises of ChatGPT. Our (...) results show that our new datasets do pose a challenge to existing methods and models, including ChatGPT, and that tackling this challenge via fine-tuning yields only partly satisfactory results. (shrink)

This paper addresses the problem of finding a Bayesian net representation of the probability function that agrees with the distributions of multiple consistent datasets and otherwise has maximum entropy. We give a general algorithm which is significantly more efficient than the standard brute-force approach. Furthermore, we show that in a wide range of cases such a Bayesian net can be obtained without solving any optimisation problem.

This article investigates the work of processors who curate and “clean” the data sets that researchers submit to data archives for archiving and further dissemination. Based on ethnographic fieldwork conducted at the data processing unit of a major US social science data archive, I investigate how these data processors work, under which status, and how they contribute to data sharing. This article presents two main results. First, it contributes to the study of invisible technicians in science by showing that the (...) same procedures can make technical work invisible outside and visible inside the archive, to allow peer review and quality control. Second, this article contributes to the social study of scientific data sharing, by showing that the organization of data processing directly stems from the conception that the archive promotes of a valid data set—that is, a data set that must look “pristine” at the end of its processing. After critically interrogating this notion of pristineness, I show how it perpetuates a misleading conception of data as “raw” instead of acknowledging the important contribution of data processors to data sharing and social science. (shrink)

A behavioral dataset contains various preference orderings displayed by the same individual in different payoff-irrelevant circumstances. We introduce a framework for eliciting the individual’s underlying preferences in such cases, in which it is conjectured that the variation in the observed preference orderings is the outcome of some cognitive process that distorts the underlying preferences. We then demonstrate for two cognitive processes how to elicit the individual’s underlying preferences from behavioral datasets.

Research interest in cuteness perception and its effects on subsequent behavior and physiological responses has recently been increasing. The purpose of the present study was to produce a dataset of Japanese infant faces that are free of portrait rights and can be used for cuteness research. A total of 80 original facial images of 6-month-old infants were collected from their parents. The cuteness level of each picture was rated on a 7-point scale by 200 Japanese people. Prototypical high- and (...) low-cuteness faces were created by averaging the top 10 and bottom 10 faces according to the mean cuteness ratings. Then, 50 composite faces were made by mixing two faces randomly chosen from the 60 unused middle-cuteness faces. The normative cuteness ratings of these composite faces were obtained from 229 Japanese men and women in their 20s–60s. The shape of each composite face was transformed to be cuter or less cute along a continuum between the high- and low-cuteness prototypical faces. A two-alternative forced-choice task confirmed that cuteness discrimination was better than the chance level for all 50 face pairs. Moreover, the results showed that young men had poorer sensitivity to cuteness differences in infant faces than older men and women of any age. This Japanese Cute Infant Face dataset, including composite face images and normative rating scores, is publicly available online. (shrink)

The role of the major histocompatibility complex (MHC) in mate choice in humans is controversial. Nowadays, the availability of genetic variation data at genomic scales allows for a careful assessment of this question. In 2008, Chaix et al. reported evidence for MHC‐dependent mate choice among European American spouses from the HapMap 2 dataset. Recently, Derti et al. suggested that this observation was not robust. Furthermore, when Derti et al. applied similar analyses to the HapMap 3 European American samples, they (...) did not see a significant effect. Although some of the points raised by Derti et al. are relevant, we disagree with the reported absence of evidence for MHC‐dependent mate choice within the HapMap samples. More precisely, we show here that the MHC dissimilarity among HapMap 3 European American spouses is still extreme in comparison to the rest of the genome, even after multiple testing correction. This finding supports the hypothesis of MHC‐dependent mate choice in some human populations. (shrink)

Recently, infrared human action recognition has attracted increasing attention for it has many advantages over visible light, that is, being robust to illumination change and shadows. However, the infrared action data is limited until now, which degrades the performance of infrared action recognition. Motivated by the idea of transfer learning, an infrared human action recognition framework using auxiliary data from visible light is proposed to solve the problem of limited infrared action data. In the proposed framework, we first construct a (...) novel Cross-Dataset Feature Alignment and Generalization framework to map the infrared data and visible light data into a common feature space, where Kernel Manifold Alignment and a dual aligned-to-generalized encoders model are employed to represent the feature. Then, a support vector machine is trained, using both the infrared data and visible light data, and can classify the features derived from infrared data. The proposed method is evaluated on InfAR, which is a publicly available infrared human action dataset. To build up auxiliary data, we set up a novel visible light action dataset XD145. Experimental results show that the proposed method can achieve state-of-the-art performance compared with several transfer learning and domain adaptation methods. (shrink)

Large-scale public datasets are vital for driving the progress of abstractive summarization, especially in law, where documents have highly specialized jargon. However, the available resources are English-centered, limiting research advancements in other languages. This paper introducesLAWSUIT, a collection of 14K Italian legal verdicts with expert-authored abstractive maxims drawn from the Constitutional Court of the Italian Republic.LAWSUITpresents an arduous task with lengthy source texts and evenly distributed salient content. We offer extensive experiments with sequence-to-sequence and segmentation-based approaches, revealing that the latter (...) achieve better results in full and few-shot settings. We openly releaseLAWSUITto foster the development and automation of real-world legal applications. (shrink)

Syllogism is a type of everyday reasoning. For instance, given that ‘Avicenna wrote the famous book the Canon of Medicine’ and ‘The Canon of Medicine has influenced modern medicine,’ it can be conc...

In the natural sciences higher order structures often occur. There seems to be a need for good methods of describing what we mean by higher order structures in various contexts. This is what hyperstructures are intended to do. We motivate and introduce this new concept. Next we illustrate how it can be applied in various types of genomic analysis—particular the correlations between single nucleotide polymorphisms and diseases. The suggested structure is quite general and may be applied to a variety of (...) situations. Finally we discuss how data sets (f. ex. genomic) may lead to topological spaces, giving new invariants and lead to the prediction of hyperstructures. (shrink)

Integrating RDF datasets has become a relevant problem for both researchers and practitioners. In the literature, there are many genetic proposals that learn rules that allow to link the resources that refer to the same real-world entities, which is paramount to integrating the datasets. Unfortunately, they are context-unaware because they focus on the resources and their attributes but forget about their neighbours. This implies that they fall short in cases in which different resources have similar attributes but refer to different (...) real-world entities or cases in which they have dissimilar attributes but refer to the same real-world entities. In this article, we present a proposal that learns context-aware rules that take into account both the attributes of the resources and their neighbours. We have conducted an extensive experimentation that proves that it outperforms the most advanced genetic proposal. Our conclusions were checked using statistically sound methods. (shrink)

The use of language models in Web applications and other areas of computing and business have grown significantly over the last five years. One reason for this growth is the improvement in performance of language models on a number of benchmarks — but a side effect of these advances has been the adoption of a “bigger is always better” paradigm when it comes to the size of training, testing, and challenge datasets. Drawing on previous criticisms of this paradigm as applied (...) to large training datasets crawled from pre-existing text on the Web, we extend the critique to challenge datasets custom-created by crowdworkers. We present several sets of criticisms, where ethical and scientific issues in language model research reinforce each other: labour injustices in crowdwork, dataset quality and inscrutability, inequities in the research community, and centralized corporate control of the technology. We also present a new type of tool for researchers to use in examining large datasets when evaluating them for quality. (shrink)

Public sector adoption of AI techniques in welfare systems recasts historic national data as resource for machine learning. In this paper, we examine how the use of register data for development of predictive models produces new ‘afterlives’ for citizen data. First, we document a Danish research project’s practical efforts to develop an algorithmic decision-support model for social workers to classify children’s risk of maltreatment. Second, we outline the tensions emerging from project members’ negotiations about which datasets to include. Third, we (...) identify three types of afterlives for citizen data in machine learning projects: (1) data afterlives for training and testing the algorithm, acting as ‘ground truth’ for inferring futures, (2) data afterlives for validating the algorithmic model, acting as markers of robustness, and (3) data afterlives for improving the model’s fairness, valuated for reasons of data ethics. We conclude by discussing how, on one hand, these afterlives engender new ethical relations between state and citizens; and how they, on the other hand, also articulate an alternative view on the value of datasets, posing interesting contrasts between machine learning projects developed within the context of the Danish welfare state and mainstream corporate AI discourses of the bigger, the better. (shrink)

In the paper, several data reduction techniques for machine learning from big datasets are discussed and evaluated. The discussed approach focuses on combining several techniques including stacking, rotation, and data reduction aimed at improving the performance of the machine classification. Stacking is seen as the technique allowing to take advantage of the multiple classification models. The rotation-based techniques are used to increase the heterogeneity of the stacking ensembles. Data reduction makes it possible to classify instances belonging to big datasets. We (...) propose to use an agent-based population learning algorithm for data reduction in the feature and instance dimensions. For diversification of the classifier ensembles within the rotation also, alternatively, principal component analysis and independent component analysis are used. The research question addressed in the paper is formulated as follows: does the performance of a classifier using the reduced dataset be improved by integrating the data reduction mechanism with the rotation-based technique and the stacking? (shrink)

We present a novel annotated dataset of legal cases pertaining to Article 6 – the right to a fair trial – of the European Convention on Human Rights (ECHR). This dataset will serve as a useful resource to the research community, to assist in the training and evaluation of AI systems designed to embody the legal reasoning involved in determining the appropriate legal outcome from a description of the case material. The annotations were applied to provide finer-grain classifications (...) of legal cases at the document level. Each classification label was sourced from a domain knowledge model, derived with legal expert guidance and known as an Angelic Domain Model (ADM), such that the classifications correspond to the actual legal rationales used by the Court when determining the outcome of a given case. We discuss our annotation pipeline, including annotator training, inter-annotator reliability evaluation, and the dissemination of the annotation outputs and associated metadata. (shrink)

Legal inference is fundamental for building and verifying hypotheses in police investigations. In this study, we build a Natural Language Inference dataset in Korean for the legal domain, focusing on criminal court verdicts. We developed an adversarial hypothesis collection tool that can challenge the annotators and give us a deep understanding of the data, and a hypothesis network construction tool with visualized graphs to show a use case scenario of the developed model. The data is augmented using a combination (...) of Easy Data Augmentation approaches and round-trip translation, as crowd-sourcing might not be an option for datasets with sensible data. We extensively discuss challenges we have encountered, such as the annotator’s limited domain knowledge, issues in the data augmentation process, problems with handling long contexts and suggest possible solutions to the issues. Our work shows that creating legal inference datasets with limited resources is feasible and proposes further research in this area. (shrink)

Natural language processing techniques contribute more and more in analyzing legal documents recently, which supports the implementation of laws and rules using computers. Previous approaches in representing a legal sentence often based on logical patterns that illustrate the relations between concepts in the sentence, often consist of multiple words. Those representations cause the lack of semantic information at the word level. In our work, we aim to tackle such shortcomings by representing legal texts in the form of abstract meaning representation, (...) a graph-based semantic representation that gains lots of polarity in NLP community recently. We present our study in AMR Parsing and AMR-to-text Generation specifically for legal domain. We also introduce JCivilCode, a human-annotated legal AMR dataset which was created and verified by a group of linguistic and legal experts. We conduct an empirical evaluation of various approaches in parsing and generating AMR on our own dataset and show the current challenges. Based on our observation, we propose our domain adaptation method applying in the training phase and decoding phase of a neural AMR-to-text generation model. Our method improves the quality of text generated from AMR graph compared to the baseline model. 2018; and “Legal Text Generation fromMeaning Representation”, published in the 32nd International Conference on Legal Knowledge and Information Systems 2019.). (shrink)

Ozone is one of the pollutants with most negative effects on human health and in general on the biosphere. Many data-acquisition networks collect data about ozone values in both urban and background areas. Usually, these data are incomplete or corrupt and the imputation of the missing values is a priority in order to obtain complete datasets, solving the uncertainty and vagueness of existing problems to manage complexity. In the present paper, multiple-regression techniques and Artificial Neural Network models are applied to (...) approximate the absent ozone values from five explanatory variables containing air-quality information. To compare the different imputation methods, real-life data from six data-acquisition stations from the region of Castilla y León are gathered in different ways and then analyzed. The results obtained in the estimation of the missing values by applying these techniques and models are compared, analyzing the possible causes of the given response. (shrink)

An important topic in evolutionary art is the development of systems that can mimic the aesthetics decisions made by human begins, e.g., fitness evaluations made by humans using interactive evolution in generative art. This paper focuses on the analysis of several datasets used for aesthetic prediction based on ratings from photography websites and psychological experiments. Since these datasets present problems, we proposed a new dataset that is a subset of DPChallenge.com. Subsequently, three different evaluation methods were considered, one derived (...) from the ratings available at DPChallenge.com and two obtained under experimental conditions related to the aesthetics and quality of images. We observed different criteria in the DPChallenge.com ratings, which had more to do with the photographic quality than with the aesthetic value. Finally, we explored learning systems other than state-of-the-art ones, in order to predict these three values. The obtained results were similar to those using state-of-the-art procedures. (shrink)

Understanding the processing of sexual stimuli has become a significant part of research on human sexuality. In addition to individual characteristics, empirical studies have shown that cultural factors play an important role in sexual stimuli processing. The attitudes toward sex have been reported to be more conservative in East Asian societies as compared to western countries, and significantly more sexual difficulties are observed among East Asian people. However, stimulus materials, which potentially facilitate human sexuality research on native East Asian people, (...) are relatively not satisfactory. Erotic stimuli depicting East Asian figures are limited in the existing picture datasets. To address this issue, we present a collection of 237 erotic and 108 control pictures, accompanied by self-reported ratings of sexual arousal, pleasantness, and sexual attractiveness for opposite-sex erotic stimuli by heterosexual males and females. This collection is divided into six categories, depending on their contents: dressed males, semi-nude males, nude males, dressed females, semi-nude females, and nude females. We showed gender differences in sexual arousal, pleasantness, and sexual attractiveness ratings in response to opposite-sex erotic pictures. Males reported the highest levels of sexual arousal, pleasantness, and sexual attractiveness for nude female pictures, whereas females reported the highest levels of sexual arousal, pleasantness, and sexual attractiveness for semi-nude male pictures. The erotic picture dataset may provide a useful resource of erotic stimuli that can be used as stimulus materials in experimental research on sexual function in East Asians. (shrink)

The Computational History of Philosophy of Science Dataset aims to be a comprehensive set of article and book chapter metadata for philosophy of science. The dataset covers the full run of over 40 journals and 3 major book series in the field. An automated author disambiguation script is used to construct canonical names for each author, and a combination of gender attribution methods is used to attribute the gender of each author. The full code used to generate the (...) dataset is available at [URL not allowed]. See the file data_dictionary.txt for data dictionary and additional information. (shrink)