8th International Workshop on Personalization Approaches in Learning Environments (PALE 2018) Preface Milos Kravcik Olga C. Santos Jesus G. Boticario Educational Technology Lab, DFKI aDeNu Research Group, UNED, Spain aDeNu Research Group, UNED, Spain GmbH, Berlin, Germany ocsantos@dia.uned.es jgb@dia.uned.es Milos.Kravcik@dfki.de Maria Bielikova Tomas Horvath Ilaria Torre Slovak University of Technology in Eötvös Loránd University Hungary TEL&DH Research Group, DIBRIS, Bratislava, Slovakia tomas.horvath@inf.elte.hu Genoa University, Italy maria.bielikova@stuba.sk Ilaria.Torre@unige.it ABSTRACT Personalization is a well-established topic in education and there 1 INTRODUCTION have been over 30 years of experience in adaptation and The 8th edition of the International Workshop on personalization approaches that use artificial intelligence. Bringing Personalization Approaches in Learning Environments (PALE) together methods, techniques and experiences from these research took place on June 30th, 2018 and was held in conjunction with the areas is the motivation of PALE this year at AIED. Its aim is to 19th International Conference on Artificial Intelligence in share and discuss the new trends in current research, with specific Education (AIED 2018). This workshop is a follow-up of seven focus on how current research on artificial intelligence combined previous editions. The focus of this workshop series is put on the with data science and other disciplines can support designers and different and complementary perspectives in which personalization developers to improve learning in its different stages. The purpose can be addressed in learning environments (e.g., informal, is to give and share promising ideas on approaches that cater for the workplace, lifelong, mobile, contextualized, and self-regulated increasing amount of information available from immediate (e.g., learning). Previous editions addressed several important topics in in terms of wearable devices) to broader contexts in order to this field, such as behavior and embodiment of pedagogic agents, provide personalized learning assistance bridging the behavioral suitable support of self-regulated learning, appropriate balance and the computational. In particular, this eighth edition of PALE between learner control and expert guidance, design of personal workshop includes 6 papers dealing with detecting reading learning environments, contextual recommendations at various strategies, providing personalized scaffolding to support student levels of the learning process, tracking of and reaction to affective learning of written argumentation, using digital avatars who states of learners, harmonization of educational and technological resemble learners to investigate their impact on learning, evaluating standards, big data processing for learning purposes, predicting the learning effectiveness of a recommender system, comparing the student outcomes, adaptive learning assessment, and evaluation of performance of a proposed eye-gaze feature classification method, personalized learning systems. and providing instructors with visualized information on sentiment and affective state of their students. The first five editions of the PALE workshop led to the Special issue on User Modelling to Support Personalization in Enhanced CCS CONCEPTS Educational Settings in the International Journal of Artificial •Applied computing → Education → Interactive learning Intelligence in Education (IJAIED) [1]. environments; • Information Systems → World Wide Web ➝ Personalization • Users and interactive retrieval ➝ From the past experience we have identified new areas of Personalization interest in this research scope to complement the previous ones. Thus, in this workshop edition we aim at sharing and discussing the new trends in current research on how user modeling and associated KEYWORDS artificial intelligent techniques are able to contextualize and Personalization, Adaptive Learning Environments, Engagement, manage the increasing amount of information coming from the task Context Awareness at hand and its surrounding environment in order to provide the PALE 2018 at AIED, London M. Kravcik et al. personalization support in a wide range of learning environments,  Cognitive and meta-cognitive scaffolding which are increasingly more sensitive to the learners and their  Cognitive and meta-cognitive scaffolding context. This covers many interrelated fields such as: intelligent  Adaptive mobile learning tutoring systems, learning management systems, personal learning  Wearable devices for sensing and acting in ubiquitous environments, serious games, agent-based learning environments, learning scenarios and others. Furthermore, we aim to cover the demanding need of  Tracking technologies for accessible learning for all personalized learning in wider contexts ranging from daily life activities to massive open online courses (MOOCs). Thus, PALE 3 CONTRIBUTIONS offers an opportunity to present and discuss a wide spectrum of issues and solutions. A peer-reviewed process was carried out to select the workshop papers. At least three members of the Program Committee with Following the experience from previous editions of this and expertise in the area reviewed each paper. As a result, 6 related workshops, PALE combines the classic 'mini-conferences' submissions were accepted (out of 7), which discuss ideas and approach with working group meetings around a specific problem. progress on several interesting topics, such as detecting reading It follows the Learning Cafe methodology to promote discussions strategies, providing personalized scaffolding to support student on some of the open issues coming from the presented papers. Each learning of written argumentation, using digital avatars who Learning Cafe consists of brief presentations of the key questions resemble learners to investigate their impact on learning, evaluating posed and small group discussions with participants randomly the learning effectiveness of a recommender system, comparing the grouped at tables. Each table is moderated by one expert in the topic performance of a proposed eye-gaze feature classification method, under discussion (mostly the presenter of the paper who has and providing instructors with visualized information on sentiment addressed the issue) and participants change tables during the and affective state of their students. All these works specially focus discussion with the aim to share ideas among the groups. on the enhanced sensitivity towards the management of educational data coming from multimodal learners' interactions and What follows is an introduction of PALE 2018 motivation and technological deployment, and how can this wide range of themes as well as an overview of the contributions accepted and situations and features impact on modeling the learner interaction discussed in the workshop. and context. In particular, there are four contributions that deal with multimodal input data. 2 MOTIVATION Kachergis et al. [2] describe a supervised machine learning PALE 2018 is focused on enhancing sensitivity towards the based system aimed at detecting reading strategies in task-oriented management of big educational data coming from learners' readings. Three relevant reading strategies features were detected interactions (e.g., multimodal sensor detection of attention and (the ratio of sentences that readers skimmed too quickly, the affect) and technological deployment (including web, mobiles, number of unique sentences read, and the variance of time spent tablets, tabletops), and how can this wide range of situations and reading each sentence). These features are easy to automatically features impact on modeling the learner interaction and context. In extract in tablet based reading and differ from typical process the current state of the art it is not clear how the new information variables used to study task oriented reading. The study involved sources are to be managed and combined in order to enhance 44 fourth-year vocational secondary training students and 1091 interaction in a way that positively impacts on the learning process graphs of students’ behavior recorded on tablets, which were whose nature is essentially adaptive. Thus, this edition of PALE at classified by human coders. These ratings were used to train a AIED 2018 aims to give and share promising ideas to the research classifier (eXtreme Gradient Boosting) on 13 features extracted question: "Which approaches can be followed to cater for the from the students’ reading behavior. The overall accuracy for increasing amount of information available from immediate (e.g., classifying reading strategies was 0.74, significantly greater than in terms of wearable devices) to broader contexts in order to chance. Searched reading strategies were the easiest to identify, provide effective and personalized assistance in learning situations with a balanced accuracy of 0.84, followed by intensive (0.81) and bridging the behavioral and the computational?" Thus, it captures targeted reading strategies (0.69). However, both human coders and current trends of the research fields of AIED, learning sciences, the classifier had difficulty identifying targeted reading, suggesting learning analytics as well as multimodal interaction research in a need for further research. HCI. Elouazizi et al. [3] report a pilot study on the use of the The AIED session of PALE includes (but is not limited to) the MindWare software, which offers personalized scaffolding to following topics: support student learning of written argumentation. This system is equipped with Natural Language Processing and Machine Learning  User engagement in learning processes modules that analyze and weigh the usage of the components of an  Data processing within and across learning situations argumentation voice, viz., the balanced use of stancing, hedging,  Ambient intelligence logical connections, and coherence. MindWare is used to scaffold  Learner and context awareness the metacognitive processes that underlie learning aspects of 2 Workshop Preface PALE 2018 at AIED, London written argumentation in the context of science education. To this Simulation: RACAS) and the student may impact on their learning. it provides, in terms of dashboards, scaffolding and formative The analysis and results from a preliminary study involving 41 feedback (in visual and numerical form) to the learner and the students focused on their perceptions while interacting with both a performance of a particular student, and/or those of groups of doppelganger avatar and a virtual human (with photorealistic students to the instructors. Preliminary results from a small-scale appearance, not based on any resemblance to the participant), offer pilot show that meta-cognitive scaffolding strategies have some clues into the possibilities and limitations of the application contributed to increasing the levels of the learners’ confidence in of this technology to build pedagogical agents. The paper did not appreciating and using the components of the argumentation voice find any significantly statistical result over different hypothesis but in their written essays. From the study follows that it is required found some evidence on a possible trend that personalizing a further analysis on both (1) how the components of the pedagogical agent’s appearance to be similar to the student’s argumentation voice have evolved or devolved across the drafts of physical appearance may play a role in the efficacy of pedagogical the essays the students have submitted to MindWare, and (2) the agents. Still, this issue needs to be further investigated. significance, if any, of the changes in the grades of the students. Parikh and Kalva [6] present a paper that focuses on comparing Dang and Ghergulescu [5] focus on evaluating the learning the performance of an eye-gaze feature classification method effectiveness of a recommender system (powered by Adaptemy's proposed by the authors (FWLC, a non-probabilistic statistical AI Engine) in terms of average lesson success rate and feature weighted linguistic classifier) with five popular classifiers. improvement per lesson. The data used in this analysis comes from The ultimate purpose is to detect learning difficulty during a 4257 students and 80266 learning lessons in a Math course. Three learning exercise and adapt content. More specifically, learning main cases with different levels of teachers’ guidance are studied. difficulty is defined here in terms of the novelty of words in written In the first case the system makes recommendations with no input text. This is reflected in the classification process, which “classifies from the teacher, in the second recommendations are loosely- into two level of learning: a novel (positive class) or a familiar guided by teacher input through assignment in a topic, and in the (negative class)". From a preliminary small-scale case study third there are no system recommendations but lessons specified by involving eight students, results show that the given method (in its teachers. The centre of the recommendation is the specific concept three versions) provides better True Positive Rates (TPR) for novel to work with. In each case the results are compared between the words than the five machine learning classifiers. However, the lessons done on system-recommended concepts and the lessons mean prediction accuracy of the best classifier is 6,6% higher than done on other concepts. The results indicated that both the learning the best version of FWLC. From this study, follows that both the success-rate and the improvement per lesson are higher if the method and its usage needs further research. system-based recommendations are followed, in all the three cases. The work described in Schubert et al [7] provides instructors According to this study, choosing the right difficulty levels of with visualized information on the sentiment and affective state of concepts to be worked on is part of the reason why working on the their students and allows them to examine how the students' concepts recommended by the engine would gain higher sentiment and emotional state change over the duration of a course. improvement per lesson. The approach is aimed at showing to the instructor both the sentiment of the overall group of students and the emotional state and personality features of an individual student. The ultimate 4 CONCLUSIONS purpose here is to leverage the combination of these approaches in The current edition of the PALE workshop deals with several order to enable instructors to know how a very large body of interesting issues: detecting reading strategies in task-oriented students are perceiving the work to be performed as well as readings, personalized scaffolding to support student learning of personalize intervention techniques based on the situation an written argumentation, investigating how the increasing similarity individual is facing. However, the approach, which uses Microsoft of the physical appearance between the agent (a digital Text Analytics API (for sentiment extraction) and IBM Watson doppelganger) and the student may impact on a learning-by- Tone Analyzer (for detection of emotional state and personality explaining paradigm, evaluating the learning effectiveness of a profile) was not used for actual interventions yet. The paper recommender system in terms of average lesson success rate and concludes pointing to further research on scoring text based on improvement per lesson, comparing the performance of a proposed several factors including the subject domain, weighting and eye-gaze feature classification method with five popular classifiers, managing posts and tracking interventions to trigger and refining and providing instructors with visualized information relating to the appropriate actions. sentiment and affective state of their students and allow them to The other two papers propose a technological deployment. examine how the students' sentiment and emotional state changes Wang et al. [4] discuss the design and evaluation of a digital over the duration of a course. doppelganger as a virtual human listener in a learning-by- explaining paradigm. Digital doppelgangers are virtual humans that highly resemble the real self but behave independently. The paper 5 ACKNOWLEDGEMENTS investigates how the increasing similarity of the physical PALE chairs would like to thank the authors for their submissions appearance between the agent (built with Rapid Avatar Capture and and the AIED workshop chairs for their advice and guidance during the PALE workshop organization. Moreover, we also would like to 3 PALE 2018 at AIED, London M. Kravcik et al. thank the following members of the Program Committee for their reviews (in alphabetical order): Mihaela Cocea, Paul De Bra, Mirjana Ivanovic, Jelena Jovanovic, Noboru Matsuda, Alexander Nussbaumer, Alexandros Paramythis, Elvira Popescu, Sergio Salmeron-Majadas, Natalia Stqash, Jozef Tvarozek, Carsten Ullrich, and Stephan Weibelzahl. The organization of the PALE 2018 workshop has been partially supported by the following projects: BIG-AFF – Fusing multimodal Big Data to provide low-intrusive AFFective and cognitive support in learning contexts (TIN2014-59641-C2-2-P) and Supervised Educational Recommender System (VEGA 1/0475/14), HIBER: Human Information Behavior in the Digital Space (APVV-15-0508) and ADAPTION: Migration zum Cyber- physischen Produktionssystem (BMBF 02P14B023). REFERENCES [1] O.C. Santos, M. Kravcik, and J.G. Boticario, (2016) Preface to Special Issue on User Modelling to Support Personalization in Enhanced Educational Settings. International Journal of Artificial Intelligence in Education, 26, 3. [2] G. Kachergis, J. Kielstra, L. Bokkers, B. Persad and I. Molenaar. Detecting reading strategies during task-oriented reading: Building an automated classifier. In Proceedings of the 8th Workshop on Personalization Approaches in Learning Environments (PALE 2018). Kravcik, M., Santos, O.C., Boticario, J.G., Bielikova, M., Horvath, T. and Torre I. (Eds.). 19th International Conference on Artificial Intelligence in Education (AIED 2018), CEUR workshop proceedings, this volume. [3] N. Elouazizi. G. Birol and G. Oberg. Learning technology-enabled (meta)- cognitive scaffolding for enabling students to learn aspects of written argumentation. In Proceedings of the 8th Workshop on Personalization Approaches in Learning Environments (PALE 2018). Kravcik, M., Santos, O.C., Boticario, J.G., Bielikova, M., Horvath, T. and Torre I. (Eds.). 19th International Conference on Artificial Intelligence in Education (AIED 2018), CEUR workshop proceedings, this volume. [4] N. Wang, A. Shapiro, A. Feng, C. Zhuang, D. Schwartz and S. Goldberg. An Analysis of Student Belief and Behavior in Learning by Explaining to a Digital Doppelganger. In Proceedings of the 8th Workshop on Personalization Approaches in Learning Environments (PALE 2018). Kravcik, M., Santos, O.C., Boticario, J.G., Bielikova, M., Horvath, T. and Torre I. (Eds.). 19th International Conference on Artificial Intelligence in Education (AIED 2018), CEUR workshop proceedings, this volume. [5] X. Dang and I. Ghergulescu. Effective Learning Recommendations powered by AI Engine. In Proceedings of the 8th Workshop on Personalization Approaches in Learning Environments (PALE 2018). Kravcik, M., Santos, O.C., Boticario, J.G., Bielikova, M., Horvath, T. and Torre I. (Eds.). 19th International Conference on Artificial Intelligence in Education (AIED 2018), CEUR workshop proceedings, this volume. [6] S. Parikh and H. Kalva.Eye Gaze Feature Classification for Predicting Levels of Learning. In Proceedings of the 8th Workshop on Personalization Approaches in Learning Environments (PALE 2018). Kravcik, M., Santos, O.C., Boticario, J.G., Bielikova, M., Horvath, T. and Torre I. (Eds.). 19th International Conference on Artificial Intelligence in Education (AIED 2018), CEUR workshop proceedings, this volume. [7] M. Schubert, D. Durruty and D.Joyner.Measuring Learner Tone and Sentiment at Scale Via Text Analysis of Forum Posts. In Proceedings of the 8th Workshop on Personalization Approaches in Learning Environments (PALE 2018). Kravcik, M., Santos, O.C., Boticario, J.G., Bielikova, M., Horvath, T. and Torre I. (Eds.). 19th International Conference on Artificial Intelligence in Education (AIED 2018), CEUR workshop proceedings, this volume. 4