A commonsense reasoning framework for substitution in cooking

This chapter proposes EmoSenticSpace, a new framework for affective common-sense reasoning that extends WordNet-Affect and SenticNet by providing both emotion labels and polarity scores for a large set of natural language concepts. The framework is built by means of fuzzy c-means clustering and support-vector-machine classification, and takes into account a number of similarity measures, including point-wise mutual information and emotional affinity. EmoSenticSpace was tested on three emotion-related natural language processing tasks, namely sentiment analysis, emotion recognition, and personality detection. In all cases, the proposed framework outperforms the state-of-the-art. In particular, the direct evaluation of EmoSenticSpace against psychological features provided in the benchmark ISEAR dataset shows a 92.15

A fuzzy syllogism in fuzzy logic is defined as an inference schema in which the major premise, the minor premise, and the conclusion are propositions containing fuzzy quantifiers. A basic fuzzy syllogism in fuzzy logic is the intersection/product syllogism. Several other basic syllogisms are developed that may be used as rules of combination of evidence in expert systems. Among these is the consequent conjunction syllogism. Furthermore, it is shown that syllogistic reasoning in fuzzy logic provides a basis for reasoning with dispositions, that is, with propositions that are preponderantly but not necessarily always true. It is also shown that the concept of dispositionality is closely related to the notion of usuality and serves as a gateway to what might be called a theory of usuality, a theory that may eventually provide a computational framework for commonsense reasoning.

Detecting false captioning using common-sense reasoning

In this paper, we consider dynamic DNA devices called DNA tweezers whose operation is based on the mechanism of DNA strand displacement. We show how the systematic use of this simple but robust mechanism makes it possible to produce a DNA-fuelled molecular machine for reasoning with dispositions which are essential ingredients of commonsense of an individual. A biochemical reaction on DNA strands is used to activate DNA tweezers for reasoning with dispositions taken from the commonsense-based knowledge base. The dispositions are basically propositions that are preponderantly but not necessarily always true. The concept of dispositionality is closely related to the notion of usuality which provides a computational framework for commonsense reasoning. As human perception is usually represented in a vague qualitative fashion, we consider fuzzy set as one tool of engineering and try to mimic human intelligence of reasoning based on perception. Since childhood, an individual perceives the world around him/her and accordingly makes several commonsense-based judgements which are essentially reasoning with dispositions.

Framework for Human-Like Reasoning for Agents Operating in the Internet of Things

While large language models (LLMs) demonstrate emerging reasoning capabilities, current inference-time expansion methods incur prohibitive computational costs through exhaustive sampling. Through analyzing decoding trajectories, we observe that most next-token predictions align well with the golden output, except for a few critical tokens that lead to deviations. Inspired by this phenomenon, we propose a novel Hint-Practice Reasoning (HPR) framework that operationalizes this insight through two synergistic components: 1) a hinter (powerful LLM) that provides probabilistic guidance at critical decision points, and 2) a practitioner (efficient smaller model) that executes major reasoning steps. The framework's core innovation lies in Distributional Inconsistency Reduction (DIR), a theoretically-grounded metric that dynamically identifies intervention points by quantifying the divergence between practitioner's reasoning trajectory and hinter's expected distribution in a tree-structured probabilistic space. Through iterative tree updates guided by DIR, HPR reweights promising reasoning paths while deprioritizing low-probability branches. Experiments across arithmetic and commonsense reasoning benchmarks demonstrate HPR's state-of-the-art efficiency-accuracy tradeoffs: it achieves comparable performance to self-consistency and MCTS baselines while decoding only 1/5 tokens, and outperforms existing methods by at most 5.1% absolute accuracy while maintaining similar or lower FLOPs.

Any theory aimed at understanding commonsense reasoning, the process that humans use to cope with the mundane but complex aspects of the world in evaluating everyday situations, should account for its flexibility, its adaptability, and the speed with which it is performed. Current theories of reasoning, however, do not satisfy these requirements, a fact we attribute, at least partly, to their separation from learning.While the central role of learning in cognition is widely acknowledged, most lines of research nevertheless study the phenomenon of “learning” separately from that of “reasoning”. The work presented here is motivated by the belief that learning is at the core of any attempt at understanding high level cognitive tasks. A formal model for the study of reasoning is developed in which a learning component has a principal role, and its advantages over traditional formalisms for the study of reasoning are shown.This paper presents an integrated theory of learning, knowledge representation and reasoning within a unified framework called Learning to Reason. The Learning to Reason framework combines the interfaces to the world used by known learning models with a reasoning task and a performance criterion suitable for it. It is shown that the framework efficiently supports “more reasoning” than traditional approaches and at the same time matches our expectations of plausible patterns of reasoning. Several results are presented to substantiate this claim, presenting cases where learning to reason about the world is feasible but either reasoning from a given representation of the world or learning representations of the world do not have efficient solutions.Overall, this framework suggests an “operational” approach to reasoning, that is nevertheless rigorous and amenable to analysis. As such, it may be a step toward a rigorous large-scale empirical study of learning and reasoning.KeywordsKnowledge RepresentationDeductive ReasoningReasoning TaskDefault ReasoningReasoning ProblemThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

We present DEGARI 2.0 (Dynamic Emotion Generator And ReclassIfier): an explainable, affective-based, art recommender relying on the commonsense reasoning framework TCL and exploiting an ontological model formalizing the Plutchik’s theory of emotions. The main novelty of this system relies on the development of diversity-seeking affective recommendations obtained by exploiting the spatial structure of the Plutchik’s ‘wheel of emotion’. In particular, such development allows to classify and to suggest, to museum users, cultural items able to evoke not only the very same emotions of already experienced or preferred objects (e.g. within a museum exhibition), but also novel items sharing different emotional stances. The system’s goal, therefore, is to break the filter bubble effect and open the users’ view towards more inclusive and empathy-based interpretations of cultural content. The system has been tested, in the context of the EU H2020 SPICE project, on the community of deaf people and on the collection of the GAM Museum of Turin. We report the results and the lessons learnt concerning both the acceptability and the perceived explainability of the received diversity-seeking recommendations.

A General Theory of Emotions and Social Life