Pasar al contenido principal

FAIRTRANSNLP-LANGUAGE: ANALYSING TOXICITY AND STEREOTYPES IN LANGUAGE FOR UNBIASED, FAIR AND TRANSPARENT SYSTEMS

PID2021-124361OB-C33

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal para Impulsar la Investigación Científico-Técnica y su Transferencia
Subprograma Subprograma Estatal de Generación de Conocimiento
Convocatoria Proyectos de I+D+i (Generación de Conocimiento y Retos Investigación)
Año convocatoria 2021
Plan Estatal 2021-2023 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario UNIVERSIDAD DE BARCELONA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 10
Encontrada(s) 1 página(s)

Repositorio Digital de la UPF
Publicación Capítulo o parte de un libro (BookPart)

Platón y las sombras chinescas de la interlengua en el aula de ELE/L2. El uso de corpus de aprendices sinófonos, más allá de la lingüística contrastiva

  • Díaz, Lourdes
En un foro destinado a profesores de español para estudiantes sinófonos, con experiencia en el aula y conocimientos directos, cuando no nativos, sobre la lengua de los alumnos (conocimientos de los que carezco), me resulta un reto inasumible pensar en aportar algo que pueda ser categóricamente útil desde el punto de vista de la práctica. Así que propongo a cambio desgranar algunas evidencias de primera mano acerca de cómo, desde una aproximación teórica no solo a la lengua, sino a la comunicación, algunos investigadores en adquisición de lenguas segundas, terceras u otras, no pocos investigadores-docentes de ELE, entre los que me encuentro, hemos incurrido en errores de bulto (cavernazos, permítanme el neologismo) sobre cómo abordar la interlengua de estos aprendices en la investigación y la didáctica del ELE y cómo se ha ido avanzando y corrigiendo la percepción errada estudio tras estudio y proponiendo una explicación para el aula. En particular, propongo espigar tres ejemplos o ámbitos, sobre (i) pronombres, (ii) vocabulario y (iii) aspectos socioculturales; casos en que, para abordar la explicación de los datos bien sea de la producción de los aprendices o de la comparación entre lenguas, nos hemos excedido en nuestra confianza en las teorías o enfoques, tropezando porque nos han faltado datos (cantidad de evidencia necesaria), precisión metodológica para extraerlos, o bien porque los hemos leído o interpretado., El presente trabajo se inserta en el marco de la investigación del proyecto FairTransNLP-Language: Analysing toxicity and stereotypes in language for unbiased, fair and transparent systems (PID2021-124361OB-C33) financiado por MCIN/AEI/10.13039/ 501100011033/FEDER,UE, de la UB; y también cuenta con el apoyo de la Fundación Hernández-Artieda.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Overview of DETESTS-Dis at IberLEF 2024:DETEction and classification of racial STereotypes inSpanish - Learning with Disagreement

  • Wolfgang S. Schmeisser-Nieto
  • Pastells, Pol
  • Frenda, Simona
  • Alejandro Ariza-Casabona
  • Farrús, Mireia
  • Taulé, Mariona
  • Rosso, Paolo|||0000-0002-8922-1242
[ES] Este artículo presenta un resumen de la tarea DETESTS como parte del workshop IberLEF 2024, dentro del congreso SEPLN 2024. Propusimos dos subtareas jerárquicas: En la subtarea 1, los participantes tuvieron que determinar la presencia de estereotipos raciales en oraciones. En la subtarea 2, de las oraciones etiquetadas con estereotipo, los participantes tuvieron que decidir si este era implícito. El dataset DETESTS-Dis contiene 12.111 oraciones de comentarios y tuits respondiendo a artículos de periódicos y bulos raciales sobre inmigración en español. 15 equipos se registraron para participar, de los cuales 6 enviaron predicciones de sistemas y 3 de ellos enviaron artículos. En este artículo presentamos información sobre los datasets de entrenamiento y de prueba, los sistemas utilizados por los participantes, las métricas de evaluación y sus resultados., [EN] This paper presents an overview of the DETESTS-Dis shared task aspart of the IberLEF 2024 Workshop on Iberian Languages Evaluation Forum, withinthe framework of the SEPLN 2024 conference. We proposed two hierarchical sub-tasks: In subtask 1, participants had to determine the presence of stereotypes in thetexts. For subtask 2, participants had to decide which texts labeled with stereotypeswere implicit stereotypes. The DETESTS-Dis dataset contains 12,111 comment sen-tences and tweets in response to newspaper articles and verified racial hoaxes involv-ing immigration in Spanish. 15 teams signed up to participate, 6 of which sent runs,and 3 of them sent their working notes. In this paper, we provide information aboutthe training and test datasets, the systems used by the participants, the evaluationmetrics of the systems and their results, This work was supported by the interna-tional project STERHEOTYPES: STudyingEuropean Racial Hoaxes and sterEOTYPESfunded by the Compagnia di San Paolo andVolksWagen Stiftung under the Challengesfor Europe call (CUP: B99C20000640007) ;the SGR CLiC project (2021 SGR 00313) funded by the Generalitat de Catalunya,and the FairTransNLP-Language project (PID2021-124361OB-C33) funded by MI-CIU/AEI/10.13039/501100011033/and byFEDER, UE.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Aggregating Value Systems for Decision Support

  • Lera-Leri, Roger X.
  • Liscio, Enrico
  • Bistaffa, Filippo
  • Jonker, Catholijn M.
  • Lopez-Sanchez, Maite
  • Murukannaiah, Pradeep K.
  • Rodriguez-Aguilar, Juan A.
  • Salas-Molina, Francisco|||0000-0002-1168-7931
[EN] We adopt an emerging and prominent vision of human -centred Artificial Intelligence that requires building trustworthy intelligent systems. Such systems should be capable of dealing with the challenges of an interconnected, globalised world by handling plurality and by abiding by human values. Within this vision, pluralistic value alignment is a core problem for AI- that is, the challenge of creating AI systems that align with a set of diverse individual value systems. So far, most literature on value alignment has considered alignment to a single value system. To address this research gap, we propose a novel method for estimating and aggregating multiple individual value systems. We rely on recent results in the social choice literature and formalise the value system aggregation problem as an optimisation problem. We then cast this problem as an l(p)-regression problem. Doing so provides a principled and general theoretical framework to model and solve the aggregation problem. Our aggregation method allows us to consider a range of ethical principles, from utilitarian (maximum utility) to egalitarian (maximum fairness). We illustrate the aggregation of value systems by considering real -world data from two case studies: the Participatory Value Evaluation process and the European Values Study. Our experimental evaluation shows how different consensus value systems can be obtained depending on the ethical principle of choice, leading to practical insights for a decision -maker on how to perform value system aggregation., The authors were supported by the research projects ACISUD (PID2022-136787NB-I00) , VAE (TED2021-131295B-C31) , VALAWAI (HE-101070930) , Fairtrans (PID2021-124361OB-C33) , Yoma OR (OPE02570) , TAILOR (952215) , and the Hybrid Intelligence Center, a 10-year programme funded by the Dutch Ministry of Education, Culture and Science through the Netherlands Organisation for Scientific Research. Maite Lopez-Sanchez belongs to the WAI research group (University of Barcelona) , an associated unit to CSIC through the IIIA.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Papel de la tecnología e innovación en la formación y mejora continua del profesorado de Lengua Española de Educación Primaria

  • Díaz Rodríguez, Lourdes
  • del-Olmo-Ibáñez, María-Teresa
  • del-Olmo-Ibáñez, María-Teresa|||/items/37947e27-0d7b-484e-8592-6226c7e9f1f7
In this paper we present the results of the assessment of the implementation of a linguistic subject –“Didáctica de la Lengua para la enseñanza primaria. Hablar y leer para escribir”- in the framework of a Teacher Training-Master program organized by the IUOGM and IGlobal under the Dominican Republic INAFOCAM’s umbrella. Data gathered from three questionnaires administered before, during and after course teaching to the target group of on service teachers, allowed us, teacher-researchers, to assess and improve the quality and adequacy of materials, procedures and assessing implemented in the four weeks hybrid-program. Since the experience assessed was a pilot course, their implications have been relevant for refining the final edition, as described in the paper.
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Aggregating value systems for decision support

  • Lera-Leri, Roger X.
  • Liscio, Enrico
  • Bistaffa, Filippo
  • Jonker, Catholijn M.
  • López-Sánchez, Maite
  • Murukannaiah, Pradeep K.
  • Rodríguez-Aguilar, Juan Antonio
  • Salas-Molina, Francisco
We adopt an emerging and prominent vision of human-centred Artificial Intelligence that requires building trustworthy intelligent systems. Such systems should be capable of dealing with the challenges of an interconnected, globalised world by handling plurality and by abiding by human values. Within this vision, pluralistic value alignment is a core problem for AI– that is, the challenge of creating AI systems that align with a set of diverse individual value systems. So far, most literature on value alignment has considered alignment to a single value system. To address this research gap, we propose a novel method for estimating and aggregating multiple individual value systems. We rely on recent results in the social choice literature and formalise the value system aggregation problem as an optimisation problem. We then cast this problem as an ℓp-regression problem. Doing so provides a principled and general theoretical framework to model and solve the aggregation problem. Our aggregation method allows us to consider a range of ethical principles, from utilitarian (maximum utility) to egalitarian (maximum fairness). We illustrate the aggregation of value systems by considering real-world data from two case studies: the Participatory Value Evaluation process and the European Values Study. Our experimental evaluation shows how different consensus value systems can be obtained depending on the ethical principle of choice, leading to practical insights for a decision-maker on how to perform value system aggregation., The authors were supported by the research projects ACISUD (PID2022-136787NB-I00), VAE (TED2021-131295B-C31), VALAWAI (HE-101070930), Fairtrans (PID2021-124361OB-C33), Yoma OR (OPE02570), TAILOR (952215), and the Hybrid Intelligence Center, a 10-year programme funded by the Dutch Ministry of Education, Culture and Science through the Netherlands Organisation for Scientific Research ., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Building rankings encompassing multiple criteria to support qualitative decision-making

  • Serramia, Marc
  • López-Sánchez, Maite
  • Moretti, Stefano
  • Rodríguez-Aguilar, Juan Antonio
Decision makers are commonly challenged with comparing, and ultimately ranking, elements with regards to the degree to which they satisfy multiple criteria and in terms of their own preferences. This calls for a new decision making framework, which we formally present here. Within such a framework, we present multi-criteria lex-cel: a new method for ranking single elements. Furthermore, we formally establish that our contributions generalise recent results in the social choice literature. We also illustrate our contributions through a case study that poses an ethical decision-making problem., Research supported by projects: COREDEM (H2020-785907); Crowd4SDG (H2020-872944); CI-SUSTAIN (PID2019-104156GB-I00); TAILOR (H2020-952215); COMRDI18-1-0010-02; 2021 SGR 00754; 2021 SGR 000313; THEMIS (ANR-20-CE23-0018); “FairTransNLP-Language: Analysing toxicity and stereotypes in language for unbiased, fair and transparent systems” (PID2021-124361OB-C33) funded by MCIN/AEI/10.13039/501100011033/FEDER,UE; “AUTODEMO: The stealth side of participatory democracy: process preferences towards automated decision-making” code SR21-00329; “VAE: Value-Awareness Engineering” TED2021-131295B-C31 funded by MCIN/AEI/10.13039/501100011033 and by the European Union NextGenerationEU/PRTR; and VALAWAI project, funded by the European Commission under Grant #101070930., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Encoding Ethics to Compute Value-Aligned Norms

  • Serramia, Marc
  • Rodríguez-Soto, Manel
  • López-Sánchez, Maite
  • Rodríguez-Aguilar, Juan Antonio
  • Bistaffa, Filippo
  • Boddington, Paula
  • Wooldridge, Michael
  • Ansotegui, Carlos
Norms have been widely enacted in human and agent societies to regulate individuals’ actions. However, although legislators may have ethics in mind when establishing norms, moral values are only sometimes explicitly considered. This paper advances the state of the art by providing a method for selecting the norms to enact within a society that best aligns with the moral values of such a society. Our approach to aligning norms and values is grounded in the ethics literature. Specifically, from the literature’s study of the relations between norms, actions, and values, we formally define how actions and values relate through the so-called value judgment function and how norms and values relate through the so-called norm promotion function. We show that both functions provide the means to compute value alignment for a set of norms. Moreover, we detail how to cast our decision-making problem as an optimisation problem: finding the norms that maximise value alignment. We also show how to solve our problem using off-the-shelf optimisation tools. Finally, we illustrate our approach with a specific case study on the European Value Study., This work has been funded by projects VAE (TED2021-131295B-C31), VALAWAI (HE-101070930), LOGISTAR (H2020-769142), AI4EU (H2020-825619), Crowd4SDG (H2020-872944), ACISUD (PID2022-136787NB-I00), COREDEM (H2020-785907), Fairtrans (PID2021-124361OB-C33), AUTODEMO (SR21-00329), TAILOR (H2020-952215), assist-Decidim (23S02974-001), 2021 SGR 00313, and 2021 SGR 00754. F. Bistaffa was supported by the H2020-MSCAIF- 2016 HPA4CF project. The authors have no competing interests to declare that are relevant to the content of this article. Maite Lopez-Sanchez belongs to the WAI research group (University of Barcelona) an associated unit to CSIC through the IIIA., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

A Conversational Agent That Learns to be Aligned With the Moral Value of Respect

  • Roselló-Marín, Eric
  • Rodríguez, Inmaculada
  • López-Sánchez, Maite
  • Rodríguez-Soto, Manel
  • Rodríguez-Aguilar, Juan Antonio
Human-Aware Artificial Intelligence Special Issue., Videogame developers typically conduct user experience surveys to gather feedback from users once they have played. Nevertheless, as users may not recall all the details once finished, we propose an ethical conversational agent that respectfully conducts the survey during gameplay. To achieve this without hindering user’s engagement, we resort to reinforcement learning and an ethical embedding algorithm. Specifically, we transform the learning environment so that it guarantees that the agent learns to be respectful (i.e. aligned with the moral value of respect) while pursuing its individual objective of eliciting as much feedback information as possible. When applying this approach to a simple videogame, our comparative tests between the two agents (ethical and unethical) empirically demonstrate that endowing a survey-oriented conversational agent with this moral value of respect avoids disturbing user’s engagement while still pursuing its individual objective, which is to gather as much information as possible., The author(s) received no financial support for the research, authorship, and/or publication of this article., This research was partially supported by projects VAE TED2021-131295B-C31, funded by MCIN/AEI/10.13 039/501100011033 and NextGenerationEU/PRTR, VALAWAI (Horizon Europe #101070930), ACISUD (PID2022-136787NB-I00 funded by MICIU/AEI/10.13039/501100011033), AUTODEMO (SR21-00329) by Fundación La Caixa, Crowd4SDG (H2020-872944), COREDEM (H2020-785907) and FairTransNLP-Language (PID2021-124361OB-C33). Maite Lopez-Sanchez and Inmaculada Rodriguez belong to the WAI research group (University of Barcelona) associated unit to CSIC by IIIA. Rodriguez-Aguilar is also supported by the ‘YOMA Operational Research’ project funded by the Botnar Foundation., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Multi-objective reinforcement learning for provably incentivising alignment with value systems

  • Rodríguez-Soto, Manel
  • Rădulescu, Roxana
  • Bistaffa, Filippo
  • Ricart, Oriol
  • Mayoral-Macau, Arnau
  • López-Sánchez, Maite
  • Rodríguez-Aguilar, Juan Antonio
  • Nowé, Ann
This paper addresses the problem of ensuring that autonomous learning agents align with multiple moral values. Specifically, we present the theoretical principles and algorithmic tools necessary for creating an environment where we ensure that the agent learns a behaviour aligned with multiple moral values while striving to achieve its individual objective. To address this value alignment problem, we adopt the Multi-Objective Reinforcement Learning framework and propose a novel algorithm that combines techniques from Multi-Objective Reinforcement Learning and Linear Programming. In addition, we illustrate our value alignment process with an example involving an autonomous vehicle. Here, we demonstrate that the agent learns to behave in alignment with the ethical values of safety, achievement, and comfort, with achievement representing the agent’s individual objective. Such ethical behaviour differs depending on the ordering between values. We also use a synthetic multi-objective environment to evaluate the computational costs of guaranteeing ethical learning as the number of values increases., This research project was made possible through the access granted by the Galician Supercomputing Center (CESGA) to its supercomputing infrastructure. The supercomputer FinisTerrae III and its permanent data storage system have been funded by the Spanish Ministry of Science and Innovation, the Galician Government, and the European Regional Development Fund (ERDF). Funding sources The research presented in this paper was supported by the EU-funded VALAWAI (# 101070930) project, the Spanish-funded EVASAI (# PID2024-158227NB-C31), ANNOTATE-SenSIT (# PID2024-156022OB-C33), VAE (# TED2021-131295B-C31), EMOROBCARE (# IASOMMA2024) and Rhymas (# PID2020-113594RB-100) projects. This work was supported by grant PID2022-136787NB-I00 funded by MCIN/AEI/10.13039/501100011033. It was also funded by GUARDEN (101060693), Fairtrans (PID2021-124361OB-C33), AUTODEMO (SR21-00329), TAILOR (H2020-952215), PrepParticip2.0 ( 24S03545-001), grants 2021 SGR 00313 and 2021 SGR 00754. Maite Lopez-Sanchez belongs to the WAI research group (University of Barcelona) associated unit to CSIC by IIIA. Roxana Rădulescu was partly supported by the Research Foundation Flanders (FWO), grant number 1286223N. This research was partially supported by funding from the Flemish Government under the “Onderzoeksprogramma Artificiële Intelligentie (AI) Vlaanderen” program and by the FWO, grant number G062819N., Peer reviewed
Recercat. Dipósit de la Recerca de Catalunya
Publicación Capítulo o parte de un libro (BookPart)

Platón y las sombras chinescas de la interlengua en el aula de ELE/L2. El uso de corpus de aprendices sinófonos, más allá de la lingüística contrastiva

  • Díaz, Lourdes
En un foro destinado a profesores de español para estudiantes sinófonos, con experiencia en el aula y conocimientos directos, cuando no nativos, sobre la lengua de los alumnos (conocimientos de los que carezco), me resulta un reto inasumible pensar en aportar algo que pueda ser categóricamente útil desde el punto de vista de la práctica. Así que propongo a cambio desgranar algunas evidencias de primera mano acerca de cómo, desde una aproximación teórica no solo a la lengua, sino a la comunicación, algunos investigadores en adquisición de lenguas segundas, terceras u otras, no pocos investigadores-docentes de ELE, entre los que me encuentro, hemos incurrido en errores de bulto (cavernazos, permítanme el neologismo) sobre cómo abordar la interlengua de estos aprendices en la investigación y la didáctica del ELE y cómo se ha ido avanzando y corrigiendo la percepción errada estudio tras estudio y proponiendo una explicación para el aula. En particular, propongo espigar tres ejemplos o ámbitos, sobre (i) pronombres, (ii) vocabulario y (iii) aspectos socioculturales; casos en que, para abordar la explicación de los datos bien sea de la producción de los aprendices o de la comparación entre lenguas, nos hemos excedido en nuestra confianza en las teorías o enfoques, tropezando porque nos han faltado datos (cantidad de evidencia necesaria), precisión metodológica para extraerlos, o bien porque los hemos leído o interpretado., El presente trabajo se inserta en el marco de la investigación del proyecto FairTransNLP-Language: Analysing toxicity and stereotypes in language for unbiased, fair and transparent systems (PID2021-124361OB-C33) financiado por MCIN/AEI/10.13039/ 501100011033/FEDER,UE, de la UB; y también cuenta con el apoyo de la Fundación Hernández-Artieda.