Dr. Erich Teppan; AAU, Austria
Date: MO, Dec 4th, 2017 (10:00-11:00, seminar room of the Institute for Software Technology, Inffeldgasse 16b, 2nd floor)
Title: Heuristic Constraint Answer Set Programming
Details: can be found here
Dr. Charalampos Doukas; FBK/CREATE-NET, Italy
Date: WED, March 29th, 2017 (13:00-13:45, seminar room of the Institute for Software Technology, Inffeldgasse 16b, 2nd floor)
Title: AGILE: Adoptive Gateways for dIverse MuLtiple Environments in the Internet of Things
Abstract: AGILE is a modular hardware and software gateway for the Internet of Things (IoT) with support for protocol interoperability, device and data management, IoT apps execution, and external Cloud communication, featuring diverse pilot activities, Open Calls & Community building. The focus of the talk will be to provide an overview of the AGILE Horizon 2020 project and related research results.
Charalampos Doukas (https://create-net.fbk.eu/openiot/people/profile/cdoukas) is a technology researcher, maker and open source hardware advocate. His main interests include the Internet of Things, wireless sensors and embedded systems, cloud systems, medical data processing and classification, medical sensors and data transmission over heterogeneous networks. He has used open hardware in several projects and his hacks mostly involve automation, acquisition, transmission and processing of medical and self-tracking data. He also runs a small blog collecting resources about medical and health-related projects that utilize Arduino and a blog about enabling the Internet of Things using Open Hardware.
Charalampos has co-organised and presented in more than 30 workshops/keynotes, published more than 60 papers in international scientific conferences, 18 journal papers and 6 book chapters (cited more than 1300 times, h index = 18). He is currently working as a technical consultant in the IoT domain and senior researcher in OpenIoT Area of FBK/CREATE-NET, Italy. He is the Technical Manager and Project Coordinator of the ICT30 AGILE – Adoptive Gateways for Interoperable Environments H2020 Project.
Charalampos is the author of “Building Internet of Things with the Arduino“.
Dr. Martijn Willemsen; Einhoven University of Technology, Eindhoven, The Netherlands
Date: Nov. 3rd, 2015
Title: Improving user experience in recommender systems: How latent feature diversification can decrease choice difficulty and improve choice satisfaction (slides)
Abstract: Recommenders are often evaluated purely by means of objective measures. However, having an accurate algorithm is just a first step towards given relevant and satisfactory recommendations to a user. A recommender is typically part of a system that tries to support a user in making a decision. Therefore it is important to evaluate the recommender in a larger context and with several types of measures. In recent years we have developed a user-centric evaluation framework (Knijnenburg, Willemsen, Gartner, Soncu and Newell, UMUAI2012) to combine subjective and objective measures to better understand why particular recommender setups work or not. I will demonstrate in Detail an application in this area that investigates Latent feature diversification to decrease choice difficulty and improve user satisfaction (Willemsen, Graus and Knijnenburg, 2015, under review). According to the psychological and marketing literature, choice overload can often occur because large sets contain many attractive but similar items. To overcome this effect, the present paper proposes that increasing the diversity of item sets might make them more attractive and satisfactory, without making them much more difficult to choose from. To this purpose, we study diversification on the latent features of a matrix factorization recommender algorithm. Study 1 diversifies a set of recommended items while controlling for the overall attractiveness of the set, and tests it in two online user experiments with a movie recommender system. The studies show that that latent feature diversification increases the perceived diversity and attractiveness of the item set, while at the same time reducing the difficulty of choosing from the set. Diversification can increase users’ satisfaction with the chosen option, especially when they are choosing from small, diverse item sets. Study 2 extends these results by testing our diversification algorithm against traditional Top-N recommendations, and finds that diverse, small item sets are even more satisfying and less effortful to choose from than Top-N recommendations. Our results suggest that diverse small sets may be the best thing one could offer a user of a recommender system.
Martijn Willemsen (www.martijnwillemsen.nl) researches the cognitive aspects of Human -Technology Interaction, with a strong focus on judgment and decision making in online environments. He has a background in electrical engineering and a PhD. in the psychology of human decision making. His applied research focuses on how online decisions can be supported by recommender systems, and includes domains such as movies, health related decisions and energy-saving measures. He has published several papers in the ACM RecSys conference proceedings and recently co-authored two chapters in the upcoming 2nd dition of the Recommender Systems Handbook. He also has a strong interest in evaluation of user experience, and co-developed with Bart Knijnenburg a user-centric Evaluation framework (Knijnenburg et al., UMUAI 2012). From a more theoretical / psychological perspective, he has a special interest in process tracing technologies to capture and analyze in detail Information processing of decision makers. In this area he published papers in psychological journals such as Psychological Review, Journal of Experimental Psychology: General and Journal of Behavioral Decision Making.
Dr. Marko Tkalčič; Johannes Kepler University, Linz, Austria
Date: Jan. 26th, 2015
Title: Affect- and Personality-based Recommendations
Abstract: Early recommender systems relied mostly on the data that was available through logging the user actions on the server-side of various services. The goal of such systems was to predict the ratings of the unknown items. In the last few years, research has started to pay attention to the underlying mechanisms of the users’ behavior, i.e. various psychological aspects. Psychological mechanisms are thought to provide more information about why a user action (a rating, a click, a purchase etc.) has happened. In this talk I will present how the psychological constructs of emotions and personality have been used to address various issues in recommender systems. While personality is defined as a set of enduring user characteristics, mood and emotions change with time. Each of these constructs can be used in a variety of ways to solve issues, such as the cold-start problem, mood regulation, group recommendations, diversity, serendipity and others.
Marko Tkalčič is a post-doctoral researcher at the Department of Computational Perception of the Johannes Kepler University in Linz and holds the title of assistant professor at the University of Ljubljana. His research work lies at the crossroads between recommender systems and affective computing. He is investigating how to use psychological constructs (such as emotions and personality) to improve recommender systems. Currently he is working on the FP7 project Phenicx where he is investigating the relationship between personality and the preferences of listeners of classical music. He has published in high impact factor journals. He is the guest editor of two special issues in the UMUAI journal and the editor of a Springer book that will appear at the end of 2015. His work had impact both in academia and industry with a working news recommender system based on his research.
Dr. Luiz Augusto Pizzato; School of Information Technologies, Faculty of Engineering & IT, The University of Sydney
Date: 18. September 2012
Title: How to find love, A recommender system approach
Abstract: In today’s modern lifestyle, online dating plays an ever-increasing role in helping people to find their life partners. In this talk, I will highlight how online dating works and present the difficulties involved in romantically connecting people to people. I will discuss some of the approaches we have used to solve these problems and the results we obtained when applying our methods to one of Australia’s largest online dating websites. I will also discuss some aspect of reciprocal recommenders as recently published in our UMUAI article: “Recommending people to people: The nature of reciprocal recommenders with a case study in online dating.”
Dr. Luiz Augusto Pizzato. Luiz is a research fellow at the University of Sydney with interest in recommender systems, social network analysis, data mining and natural language processing. Luiz received his Bachelor of Computer Science in 2000 from the Pontifical Catholic University of Porto Alegre (PUCRS), Brazil. In the same year, he joined the Hewlett Packard/PUCRS Research Centre in High Performance Computing. In 2003, Luiz received a Master of Computer Science from PUCRS for his Thesis involving query expansion using thesauri information for information retrieval (IR). In 2003 at the University of Évora, Portugal, Luiz integrated his Masters research with the SINO search engine to enable the online search of legal decisions made by the Portuguese Attorney General. In 2009 at Macquarie University, Australia, Luiz was awarded his PhD for his work on the use the using linguistically motivated features in the document retrieval stages of the question answering task. In 2009, Luiz joined the CHAI research group at the University of Sydney and the Smart Services CRC on the personalisation project. Since then, Luiz has been working in the personalisation domain with strong focus on data mining and recommender systems on social networks. Luiz is currently applying his Research to people recommenders in social networks for task such as finding best matches in online dating and matching job candidates with employers.
Prof. Martin Robillard; McGill University, Montreal, Canada
Date: 8. August 2011
Title: Recommendation Systems for API Usage
Abstract: Most software projects reuse components exposed through Application Programming Interfaces (APIs). Large APIs can be difficult to use effectively. To increase the usability of large and complex APIs, we are currently experimenting with recommendation systems that attempt to fulfill the information needs faced by developers who must use these APIs. In this talk, I will present an overview of recommendation systems for API usability, and describe two such systems developed at McGill: API Explorer and SemDiff. API Explorer leverages the structural relationships between API elements to recommend methods or types which, although not directly reachable from the type a developer is currently working with, may be relevant to solving a programming task. SemDiff analyzes the change history of a framework to recommend how to adapt clients to new, backward-incompatible versions of the framework.
Martin Robillard is an Associate Professor of Computer Science at McGill University, where he heads the the Software Evolution Research Group (SWEVO). His current research focuses on the automated analysis of softwaredevelopment artifacts to support software evolution and maintenance. Prof. Robillard recently worked as a visiting researcher at Microsoft Research in Redmond, WA. He is the recipient of four ACM SIGSOFT Distinguished Paper Awards, and currently holds a Humboldt Fellowship. He is serving as the Program Co-Chair for the 20th ACM SIGSOFT International Symposium on the Foundations of Software Engineering, and previously served on the program committees of numerous software engineering conferences including the ACM/IEEE International Conference on Software Engineering. He received his Ph.D. and M.Sc. in Computer Science from the University of British Columbia and a B.Eng. from École Polytechnique de Montréal.
Dr. Walid Maalej; Technische Universität München, München, Deutschland
Date: 29. Juli 2010
Title: Potentials and Challenges of Recommendation Systems in Software Engineering
Abstract: By surveying recommendation systems in software engineering, we found that existing approaches have been focusing on “you might like what similar developers like” scenarios. However, structured artifacts and semantically well-defined engineering activities bear large potentials for further recommendation scenarios. We introduce a novel “landscape” of software engineering recommendation systems. We then line out several scenarios supporting developers’ personal productivity, collaboration and knowledge sharing, as well as the involvement of end users in the development process. The main research challenges are enabling context-awareness and addressing the role information providers.
Walid Maalej is a researcher at the research group of Applied Software Engineering at the Technische Universität München (TUM). He recently received a Ph.D. degree in Computer Science (on Intention-Based Integration of Software Engineering Tools) from the TUM. He also received a degree in Technology Management from the Center for Digital Technology and Management, an institution of the Bavarian Elite Network. Walid’s current research interests include human factors in software engineering, context-aware infrastructures for knowledge workers, agile methodologies, as well as “marrying” software engineering with the semantic web. Walid co-organized several international events in the field of Software engineering and knowledge management, such as the 17th IEEE Requirement Engineering Conference, the Social Software Engineering workshop series and the Managing Requirement Knowledge workshop series. He serves as consultant for several companies (including mediawave, Siemens Singapore, Deutsche Telecom, and Rohde & Schwarz) is a.o. the initiator of the TeamWeaver FLOSS platform, and the scientific coordinator of the EU-Funded project FastFix.
Prof. Robin Burke; DePaul University of Chicago, Chicago, IL, US
Date: 9. June 2010
Title: Hybrid Recommendation in Social Annotation Systems
Abstract: Social annotation systems allow users to annotate resources with personalized tags and to navigate large and complex information spaces. However, the size and heterogeneity of these systems means that users often require the assistance of recommender systems. In this talk, I will discuss the challenges in extending prior work in recommendation to
social annotation systems. I will introduce a class of weighted hybrid recommendation algorithms. These algorithms have performance comparable to state-of-the-art matrix factorization models, but are superior in scalability, flexibility, and updatability. I will demonstrate results on a number of real-world datasets. I will also examine the concept of “information channels”, underlying regularities in the social Annotation data that help explain the differences in recommenderperformance on the data sets.