Abstract:International audienceDrawn from the lessons learned in an application for the subway company in Paris, we pointed out that operators used practices instead of the procedures developed by the company, practices appearing like contextualization of the procedures taking into account specificity of the task at hand and the current situation. This leads us to propose, first, a working definition of context at a theoretical level, and, second, its implementation in a software called Contextual Graphs. In this paper… Show more
“…Contextual: Context typically refers to domain information about objects other than the ones explicitly participating as inputs to or outputs generated by the system, such as information about users, situations, and broader environment affecting the computation [7,12,25]. In [13], the author defines context as a "collection of relevant conditions and surrounding influences that make a situation unique and comprehensible". Thus, context-aware explanations often include extra information that is not contained in the current described situation but is part of the users' context.…”
The role of explanations in intelligent systems has in the last few years entered the spotlight as AI-based solutions appear in an ever-growing set of applications. Though data-driven (or machine learning) techniques are often used as examples of how opaque (also called black box) approaches can lead to problems such as bias and general lack of explainability and interpretability, in reality these features are difficult to tame in general, even for approaches that are based on tools typically considered to be more amenable, like knowledge-based formalisms. In this paper, we continue a line of research and development towards building tools that facilitate the implementation of explainable and interpretable hybrid intelligent socio-technical systems, focusing on features that users can leverage to build explanations to their queries. In particular, we present the implementation of a recently-proposed application framework (and make available its source code) for developing such systems, and explore user-centered mechanisms for building explanations based both on the kinds of explanations required (such as counterfactual, contextual, etc.) and the inputs used for building them (coming from various sources, such as the knowledge base and lower-level data-driven modules). In order to validate our approach, we develop two use cases, one as a running example for detecting hate speech in social platforms and the other as an extension that also contemplates cyberbullying scenarios.
“…Contextual: Context typically refers to domain information about objects other than the ones explicitly participating as inputs to or outputs generated by the system, such as information about users, situations, and broader environment affecting the computation [7,12,25]. In [13], the author defines context as a "collection of relevant conditions and surrounding influences that make a situation unique and comprehensible". Thus, context-aware explanations often include extra information that is not contained in the current described situation but is part of the users' context.…”
The role of explanations in intelligent systems has in the last few years entered the spotlight as AI-based solutions appear in an ever-growing set of applications. Though data-driven (or machine learning) techniques are often used as examples of how opaque (also called black box) approaches can lead to problems such as bias and general lack of explainability and interpretability, in reality these features are difficult to tame in general, even for approaches that are based on tools typically considered to be more amenable, like knowledge-based formalisms. In this paper, we continue a line of research and development towards building tools that facilitate the implementation of explainable and interpretable hybrid intelligent socio-technical systems, focusing on features that users can leverage to build explanations to their queries. In particular, we present the implementation of a recently-proposed application framework (and make available its source code) for developing such systems, and explore user-centered mechanisms for building explanations based both on the kinds of explanations required (such as counterfactual, contextual, etc.) and the inputs used for building them (coming from various sources, such as the knowledge base and lower-level data-driven modules). In order to validate our approach, we develop two use cases, one as a running example for detecting hate speech in social platforms and the other as an extension that also contemplates cyberbullying scenarios.
“…Enterprises develop procedures to address a focus in any situation, but generally procedures result in sub-optimal solutions for any specific focus. As a consequence, actors develop practices that contextualize the procedure for addressing the specific contexts where is the focus (Brézillon, 2005b(Brézillon, , 2006. In some way, they elaborate a contextualized task model in the spirit of what is shown Figure 2.7.…”
Section: Context and Focus Processingmentioning
confidence: 99%
“…As said previously, this supposes too the management of the shared context with the addition of particular contextual elements for modeling the management of group members' interaction (turn, acceptance, etc.) during group-activity development (Brézillon., 2006). The shared context becomes a crucial place for interaction management as well as task management in each actor's activity.…”
Modeling and Using Context: 25 Years ofLessons Learned 1This chapter presents the results of 25 years of research on modeling and using context in realworld applications. The first milestone was an operational definition of context relating it to a focus (problem-solving, decision-making, task realization). The focus provides context with links between context and knowledge, a division between contextual knowledge and external knowledge with respect to a focus, and a dynamic linked to reasoning evolution. Conversely, context makes focus more explicit. The second milestone was a context-based representation of decision trees as contextual graphs with contextual elements (expressing the focus) instead of chance or event nodes. The operational definition of context opens up a conceptual framework that can be associated with an implementation framework. The third milestone was the development of a context-based representation formalism that led to the software Contextual Graphs (CxG) that has been used in about 20 real-world applications. The fourth milestone concerns an extension of the CxG representation formalism for representing group activity with a cyclic use of a CxG for simulation purposes. The formalism now allows a modeling of "nonlinear" behaviors of a group (e.g. interaction, negotiation, etc.) as well as of a user (e.g. checking alternatives). The fifth milestone opens the way to Context-based Intelligent Assistant Systems.
“…1. the Monitor uses a parser to uncover its context [15], using a domain ontology. Initially, the domain ontology employed deals with these subjects: Intelligent Agents, First-Order Logic, and Knowledge…”
Collaborative learning can be motivated via environments that provide communication, and areas for discussion. In such environments, both students and instructors need online support in order to produce useful interactions. In this paper we present SmartChat, an intelligent environment for collaborative discussions. It uses an argumentation model to organise users interactions and it classifies users in pre-defined stereotypes, based on their participation in the discussion. That classification helps the SmartChat's agent society to interfere in the discussion, in order to motivate users participation, and also recommend references or incentivate pair collaboration. SmartChat was developed in Java, using a knowledge base accessed through JEOPS. The initial tests have indicated promising results.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.