Explainable Artificial Intelligence (XAI) is an emerging field within Artificial Intelligence (AI) that has provided many methods that enable humans to understand and interpret the outcomes of AI systems. However, deciding on the best explanation approach for a given AI problem is currently a challenging decision-making task. This paper presents the iSee project, which aims to address some of the XAI challenges by providing a unifying platform where personalized explanation experiences are generated using Case-Based Reasoning. An explanation experience includes the proposed solution to a particular explainability problem and its corresponding evaluation, provided by the end user. The ultimate goal is to provide an open catalog of explanation experiences that can be transferred to other scenarios where trustworthy AI is required.

Deep learning is especially remarkable in terms of image classification. However, the outcomes of models are not explainable to users due to their complex nature, having an impact on the users’ trust in the provided classifications. To solve this problem, several explanation techniques have been proposed, but they greatly depend on the nature of the images being classified and the users’ perception of the explanations. In this work, we present Case-Based Reasoning as a learning-based solution to the problem of selecting the best explanation method for the image classifications obtained by models. We propose the elicitation of a case base that reflects the human perception of the quality of the explanations and how to reuse this knowledge to select the best explanation approach for a given image classification.

Abstract Traditionally, recommender systems use collaborative filtering or content-based approaches based on ratings and item descriptions. However, this information is unavailable in many domains and applications, and recommender systems can only tackle the problem using information about interactions or implicit knowledge. Within this scenario, this work proposes a novel approach based on link prediction techniques over graph structures that exclusively considers interactions between users and items to provide recommendations. We present and evaluate two alternative recommendation methods: one item-based and one user-based that apply the edge weight, common neighbours, Jaccard neighbours, Adar/Adamic, and Preferential Attachment link prediction techniques. This approach has two significant advantages, which are the novelty of our proposal. First, it is suitable for minimal knowledge scenarios where explicit data such as ratings or preferences are not available. However, as our evaluation demonstrates, this approach outperforms state-of-the-art techniques using a similar level of interaction knowledge. Second, our approach has another relevant feature regarding one of the most significant concerns in current artificial intelligence research: the recommendation methods presented in this paper are easily interpretable for the users, improving their trust in the recommendations.

The Becalm project is an open and low-cost solution for the remote monitoring of respiratory support therapies like the ones used in COVID-19 patients. Becalm combines a decision-making system based on Case-Based Reasoning with a low-cost, non-invasive mask that enables the remote monitoring, detection, and explanation of risk situations for respiratory patients. This paper first describes the mask and the sensors that allow remote monitoring. Then, it describes the intelligent decision-making system that detects anomalies and raises early warnings. This detection is based on the comparison of cases that represent patients using a set of static variables plus the dynamic vector of the patient time series from sensors. Finally, personalized visual reports are created to explain the causes of the warning, data patterns, and patient context to the healthcare professional. To evaluate the case-based early-warning system, we use a synthetic data generator that simulates patients' clinical evolution from the physiological features and factors described in healthcare literature. This generation process has been verified with a real dataset and allows the validation of the reasoning system with noisy and incomplete data, threshold values, and life/death situations. The evaluation demonstrates promising results and good accuracy (0.91) for the proposed low-cost solution to monitor respiratory patients.

With the increasing importance of e-commerce and the immense variety of products, users need help to decide which ones are the most interesting to them. This is one of the main goals of recommender systems. However, users’ trust may be compromised if they do not understand how or why the recommendation was achieved. Here, explanations are essential to improve user confidence in recommender systems and to make the recommendation useful. Providing explanation capabilities into recommender systems is not an easy task as their success depends on several aspects such as the explanation’s goal, the user’s expectation, the knowledge available, or the presentation method. Therefore, this work proposes a conceptual model to alleviate this problem by defining the requirements of explanations for recommender systems. Our goal is to provide a model that guides the development of effective explanations for recommender systems as they are correctly designed and suited to the user’s needs. Although earlier explanation taxonomies sustain this work, our model includes new concepts not considered in previous works. Moreover, we make a novel contribution regarding the formalization of this model as an ontology that can be integrated into the development of proper explanations for recommender systems.

Lower back Pain (LBP) is pathological and occurs in about 80% of the population at least once in their life. Physiotherapists personalise manual treatments to heal or relieve pain according to the patient characteristics. The contribution of this methodological paper is the description and evaluation of the configuration software associated to a therapy machine that executes back segment mobilisations. The configuration software uses Case-Based Reasoning (CBR), a successful Machine Learning technique, based on mimicking the human decision making process by reusing previously applied configuration episodes on similar individuals. This paper demonstrates its feasibility and cost-effectiveness for the configuration of treatments as it reuses expert knowledge and maximises effectiveness by taking into account the patient’s personal medical record and similar patterns among different patients. Having a baseline of 31% success rate using a standard solution based on interpolation, the CBR engine can achieve, on average, up to 70% success rate when proposing a machine configuration to the physiotherapist. Regarding clinical results, we run a longitudinal observational study that achieves an average improvement of 31.63% using the pain Visual Analogue Scale (VAS), a 7% according to the Oswestry Disability Index (ODI), and 13% in the 36-Item Short Form Health Survey (SF-36).

Social systems by their definition encourage interaction between users and both on-line content and other users thus generating new sources of knowledge that is valuable for recommender systems. In this paper we deal with the situation of having a recommender system where, even if a social structure implicitly exist, its users are not explicitly connected through a social network. We describe SocialFan, a domain independent tool that allows defining and integrating the social network infrastructure to capture and use the social knowledge into an existing recommender system.

Classical recommender systems focus on recom- mending the most relevant items to users. An active area of research proposes to complete the recommendation process by considering additional contextual information, such as time, location, budget, weather or social position. Researchers and practitioners in different domains have already recognized the great impact of contextual information in decision-making pro- cesses. In this paper, we focus on recommenders for tourism and leisure activities where contextual information plays a central role to modify the initial user preferences. We present Madrid Live, a context-aware recommender system (CARS) to recommend leisure activities in Madrid. In Madrid Live, users state their own restrictions and preferences to their plans. The system recommends the set of activities that satisfies these preferences together with the contextual knowledge. The main contributions of our approach are the contextual recommendation and the system explanation interface that allows the user to understand the recommendation process.

Nowadays, recommender systems are useful tools to filter items and information for users. There is a huge diversity of approaches to create customized recommendations. Because of this, a developer needs to know the features of these approaches to select which one is the best approach in a specific domain. In this paper, we explain the first step in the design of our intelligent framework to create recommender systems. This first step is called RecOnto: an ontology to model recommender system as a collection of components related between them. This ontology defines and classifies all components that compound a recommender system. Moreover, depending on the information used by the recommender system, it can filter the components used by the system. In addition, this ontology can be extended to add more components or apply this model in other domains. Finally, we explain an example about how to apply RecOnto to model CoCARE, a real context-aware recommender system in the health domain.

Social recommender systems exploit the social knowledge available in social networks to provide accurate recommendations. However, their instantiation is not straightforward due to its complexity. To alleviate this development complexity, we propose a methodology based on templates that conceptualize the behavior of such applications and can be reused to create several social recommender applications in social networks. This development methodology comprises not only templates but also a generic architecture named ARISE and a collection of software components that provide the required functionality. We prove that our social templates speed up and facilitate the development process, and demonstrate the viability of our generic architecture in two different case studies.

In this article we review the existing techniques in group recommender systems and we propose some improvement based on the study of the different individual behaviors when carrying out a decision-making process. Our method includes an analysis of group personality composition and trust between each group member to improve the accuracy of group recommenders. This way we simulate the argumentation process followed by groups of people when agreeing on a common activity in a more realistic way. Moreover, we reflect how they expect the system to behave in a long term recommendation process. This is achieved by including a memory of past recommendations that increases the satisfaction of users whose preferences have not been taken into account in previous recommendations.

The goal of this paper is to show a movie recommender system for groups of people, integrated in the social network Face book through an application called Happy movie. This application tries to mitigate certain limitations in existing group recommender systems, like obtaining the users profile or offering trading methods for users in order to reach a final agreement. The method used to make the group recommendation is based on three important features: personality, social trust and memory of past recommendations. This way we simulate in a more realistic way the argumentation process followed by groups of people when deciding a joint activity.

Textual reuse is an integral part of textual case-based reasoning (TCBR) which deals with solving new problems by reusing previous similar problem-solving experiences documented as text. We investigate the role of text reuse for text authoring applications that involve feedback or review generation. Generally providing feedback in the form of assigning a rating from a likert scale is far easier compared to articulating explanatory feedback as text. When previous feedback generated about the same or similar objects are maintained as cases, there is opportunity for knowledge reuse. In this paper, we show how compositional and transformational adaptation techniques can be applied once sentences in a given case are aligned to relevant structured attribute values. Three text reuse algorithms are introduced and evaluated on a dataset gathered from online Hotel reviews from TripAdvisor. Here cases consists of both structured sub-rating attributes together with textual feedback. Generally, aligned sentences linked to similar sub-rating values are clustered together and prototypical sentences are then extracted to enable reuse across similar authors. Experiments show a close similarity between our proposed texts and actual human edited review text. We also found that problems with variability in vocabulary are best addressed when prototypes are formulated from larger sets of similar sentences in contrast to smaller sets from local neighbourhoods.

In this paper we describe some new ideas to improve recommendations to groups of people. Our approach maximizes the global satisfaction for the group taking into account people personality and the social relationships among people in the group. We present some results with two cases of study based on the movie recommendation domain with heterogeneous groups. The first case study uses synthetically generated groups of people to test how the group composition affects the accuracy of the recommendation. Our second case study uses real users and groups where the topology of the groups is based on a social network. This second case of study with real users confirms the wide conclusions of the preliminary experiment with synthetic data, which allows us to conclude that it is possible to realize trustworthy experiments with synthetic data.

jCOLIBRI is currently a reference platform in the CBR community for building CBR systems that includes facilities to design different types of CBR applications \cite{ICCBR05CBRT,jscp07BuildingCBRsystems,AI06OntBasedCBR}. In this paper we focus in some recently included tools that allow the improvement of performance of previously designed applications. These optimization tools mainly facilitate to adjust features on large case bases like clustering and noise reduction techniques, and to adjust processes like refine similarity metrics through case base visualization, parallelization of retrieval or distribution of the case base and reasoning thought different agents. We present the tools and exemplify how to use them in a real scenario. We have developed an experiment for the automatic classification of a textual case base made of 1500 academic journals belonging to 20 different areas.

In this article we introduce a novel method of making recommendations to groups based on existing techniques of collaborative filtering and taking into account the group personality composition. We have tested our method in the movie recommendation domain and we have experimentally evaluated its behavior under heterogeneous groups according to the group personality composition.

The goals, technologies and problems related to the Semantic Web are well known for research purposes. Due to its extent, the inclusion of the concepts that concern to the Semantic Web in Computer Science courses is not easy. In this paper we detail our experience on a project-oriented approach to learn and to put into practice the main problems, concepts and technologies related to the Semantic Web. The project domain focuses on semantic mark up and retrieval of pictures, and the comparison between syntactical and semantic retrieval methods.

Our goal is to support system developers in rapid prototyping recommender systems using Case-Based Reasoning (CBR) techniques. In this paper we describe how jCOLIBRI can serve to that goal. jCOLIBRI is an object-oriented framework in Java for building CBR systems that greatly benefits from the reuse of previously developed CBR systems.

Case Based Reasoning (CBR) systems reason by similarity between current unsolved problems and past solved problems. In such systems interaction between users and the reasoning module is very important to identify the specific aspects of the query problem. In this paper we describe a textual interface module included into JCOLIBRI, a Java framework to design CBR systems. To be able to "understand" the query, the module processes the text using information extraction, analysis and reasoning techniques, based on external resources like domain ontologies and other linguistic resources such as Wordnet.

Case-based reasoning (CBR) is a paradigm for combining problem solving and learning that has become one of the most successful applied subfields of AI in recent years. Now that CBR has become a mature and established technology two necessities have become critical: the availability of tools to build CBR systems, and the accumulated practical experience of applying CBR techniques to real-world problems. In this paper we are presenting jcolibri, an object-oriented framework in Java for building CBR systems, that greatly benefits from the reuse of previously developed CBR systems.