Uncovering Values: Detecting Latent Moral Content from Natural Language with Explainable and Non-Trained Methods

Asprino, Luigi; Bulla, Luana; Giorgis, Stefano De; Gangemi, Aldo; Marinucci, Ludovica; Mongiovì, Misael

doi:10.18653/v1/2022.deelio-1.4

Cited by 10 publications

(9 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We consider the user's annotations of emotions and moral values (answers to questions Q3 and Q4, respectively) as the ground truth and compare the system's predictions with them to measure performances. For the emotion classification task we compare the unsupervised NLI-based model [4], and the supervised E-BERTeet [24] and E-DistilRoBERTa [15] models, introduced in Sect. 2.2.…”

Section: Results and Evaluationmentioning

confidence: 99%

“…For the purposes of our study, we selected three different methods based on natural language models to classify the emotional and moral perspective of users based on their natural language descriptions of artworks 5 . The first is based on Asprino et al [4], which employs a zero-shot classification in an unsupervised way, drawing inspiration from Yin et al [36]. This approach is based on the use of a model trained for a natural language inference (NLI) task, where a premise is determined whether or not it entails a hypothesis.…”

Section: Reference Modelsmentioning

confidence: 99%

“…Following Asprino et all. [4], we apply a pre-trained Natural Language Inference system as ready-made zeroshot sequence classifier (namely NLI-based). We use the framework to categorize emotions and moral values, developing two different system configurations based respectively on the taxonomies of Ekman and Haidt.…”

Section: Comparing Perspectives By Emotion and Moral Value Classifica...mentioning

confidence: 99%

“…Today VR is applied in many domains related to cognitive sciences, including social technology, which facilitates interactions by promoting the psychological well-being of people in social contexts. As a countermeasure, as a part of a wider project 4 , which aims to ensure social cohesion, participation and inclusion through cultural engagement, we use the virtual reality technology to build an innovative system, in which users are subjected to the Interpretation-Reflection Loop (IRL) [7] in an immersive environment that makes them share their personal interpretations with others, listen to others' interpretations given by different people, and take another point of view through virtual embodiment.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Comparing User Perspectives in a Virtual Reality Cultural Heritage Environment

Bulla

Giorgis

Gangemi

et al. 2023

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Virtual reality enables the creation of personalized users experiences that put together people of different cultures and ethnicity. We consider a novel concept of virtual reality innovation in museums, which is cognitively grounded, and based on data and their semantics, to enable users to share their experiences, as well as to assume the perspective of other users, with the ultimate goal of increasing social cohesion. The implementation of this scenario requires an autonomous artificial system capable to detect emotions and values from a dialogue involving museum visitors who express their personal point of view, listen to the point of view of other visitors, and assume others' perspectives. An important feature of this system is the ability of detecting similarity and dissimilarity between users' perspectives expressed by speech when exposed to artworks. This aids in defining an effective strategy for sharing diverse user perspectives for increasing social cohesion. Moreover, it enables an unbiased quantification of the success of the interaction in terms of change in the user perspective. Based on results from previous work, we employ the Ekman emotion model and Haidt moral value model to detect emotional and moral value profiles from user descriptions of artworks. We propose a novel method for measuring the similarity between user perspectives by comparing emotional and moral value profiles. Our results show that the employment of unsupervised text classification models is a promising research direction for this task.

show abstract

Section: Results and Evaluationmentioning

confidence: 99%

Section: Reference Modelsmentioning

confidence: 99%

Section: Comparing Perspectives By Emotion and Moral Value Classifica...mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Comparing User Perspectives in a Virtual Reality Cultural Heritage Environment

Bulla

Giorgis

Gangemi

et al. 2023

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

show abstract

“…et al [38], [39]. This corpus has been used to design novel methods for moral sentiment classification [40] used in models to investigate the impacts of moral framing in other domains (e.g., misinformation and polarization Mutlu et al , Ruch et al [41], [42] and has been applied to train models that produce morally salient text (e.g., arguments and jokes, Alshomary et al Yamane et al [43], [44])…”

Section: Moral Foundation Using Sentiment Analysismentioning

confidence: 99%

Understanding the human psychology towards relationships using the Reddit dataset

Dubey¹

View full text Add to dashboard Cite

There is an increasing interest in exploiting social media data to address problems in various domains, including social science and psychology. Reddit is a well-known social media platform that promotes user interactions on various topics and accumulates massive amounts of data daily. Reddit users participate in discussions by posting reddits, which could be text posts, links, images, and videos, or by reacting to reddits via up- or down-votes. These reddits can be analyzed to investigate mass opinion on particular issues. In this thesis, we analyze users' behavioral trends and morals of expiry dating using reddits. A relationship can start from casual dating and may not come with future bonding; such a relationship starts with an expiration date, which is known at the initial stage. People tend to seek advice on this type of relationship on social media, including Reddit. In this study, we extract reddits on relationships and apply machine-learning-based methods to infer trends on expiry dating. We adapt anchor-based topic models to infer topics from reddit corpus and estimate the trends over these topics. In addition, we infer the moral intentions--care/harm, fairness/cheating, purity/degradation, authority/subversion, loyalty/betrayed--of individuals while getting into expiry relationships based on the updated moral foundation theory framework. We collected ~665K reddits under various subreddits from 2019 to 2022. Our analysis shows an increasing trend in the topic 'dating with end date' where the moral is 'fairness' over these years.

show abstract

Basic Human Values and Moral Foundations Theory in ValueNet Ontology

Giorgis

Gangemi

Damiano

2022

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Values, as intended in ethics, determine the shape and validity of moral and social norms, grounding our everyday individual and community behavior on commonsense knowledge. The attempt to untangle human moral and social value-oriented structure of relations requires investigating both the dimension of subjective human perception of the world, and socio-cultural dynamics and multi-agent social interactions. Formalising latent moral content in human interaction is an appealing perspective that would enable a deeper understanding of both social dynamics and individual cognitive and behavioral dimension. To formalize this broad knowledge area, in the context of ValueNet, a modular ontology representing and operationalising moral and social values, we present two modules aiming at representing two main informal theories in literature: (i) the Basic Human Values theory by Shalom Schwartz and (ii) the Moral Foundations Theory by Graham and Haidt. ValueNet is based on reusable Ontology Design Patterns, is aligned to the DOLCE foundational ontology, and is a component of the Framester factual-linguistic knowledge graph.

show abstract

Uncovering Values: Detecting Latent Moral Content from Natural Language with Explainable and Non-Trained Methods

Cited by 10 publications

References 0 publications

Comparing User Perspectives in a Virtual Reality Cultural Heritage Environment

Comparing User Perspectives in a Virtual Reality Cultural Heritage Environment

Understanding the human psychology towards relationships using the Reddit dataset

Basic Human Values and Moral Foundations Theory in ValueNet Ontology

Contact Info

Product

Resources

About