Publications

Big data landscapes: improving the visualization of machine learning-based clustering algorithms (May 2018)

With the internet, massively heterogeneous data sources need to be understood and classified to provide suitable services to users such as content observation, data exploration, e-commerce, or adaptive learning environments. The key to providing these services is applying machine learning (ML) in order to generate structures via clustering and classification. Due to the intricate processes involved in ML, visual tools are needed to support designing and evaluating the ML pipelines. In this contribution, we propose a comprehensive tool that facilitates the analysis and design of ML-based clustering algorithms using multiple visualization features such as semantic zoom, glyphs, and histograms.

Dietrich Kammer, Mandy Keck, Thomas Gründer, and Rainer Groh. 2018. Big data landscapes: improving the visualization of machine learning-based clustering algorithms. In Proceedings of the 2018 International Conference on Advanced Visual Interfaces (AVI ’18). ACM, New York, NY, USA, Article 66, 3 pages. DOI: https://doi.org/10.1145/3206505.3206556.

Conversation Trainings – Towards a Multidimensional Visualization of Learning Flows (May 2018)

Conversation trainings are an e-learning method designed for the transfer of communication-related skills. With increasing conversation length and number of learners, creating and evaluating conversations becomes a complex task that needs appropriate visual guidance and data visualization. In this paper we show the inherent complexities of conversation trainings and we will guide through an approach that eases the conversation creation. We will explain data collection with Experience API and finally we present a Sankey Diagrams based approach to visualize large amounts of collected learning data.

Alexander Maasch, Romy Bürger. 2018. Conversation Trainings – Towards a Multidimensional Visualization of Learning Flows, Proc. of VisBIA 2018 – Workshop on Visual Interfaces for Big Data Environments in Industrial Applications, Castiglione della Pescaia, Grosseto (Italy), 29 May, 2018, 40-47, CEUR-WS.org, online CEUR-WS.org/Vol-1/lenzerini.pdf.

VisBIA 2018: workshop on Visual Interfaces for Big Data Environments in Industrial Applications (May 2018)

Industrial applications can benefit considerably from the overwhelming amount of still growing resources such as websites, images, texts, and videos that the internet offers today. The resulting Big Data Problem does not only consist of handling this immense volume of data. Moreover, data needs to be processed, cleaned, and presented in a user-friendly, intuitive, and interactive way. This workshop addresses visualization and user interaction challenges posed by the four V’s: Volume (huge data amounts in the range of tera and petabytes), Velocity (the speed in which data is created, processed, and analysed), Variety (the different heterogeneous data types, sources, and formats), and Veracity (authenticity and validity of data). Big Data driven interfaces combine suitable backend and frontend technologies as well as automatic and semi-automatic approaches in order to analyze data in various business contexts. An important aspect is human intervention in developing and training data-driven applications (human in the loop). Our focus is on Visual Big Data Interfaces in industrial contexts such as e-commerce, e-learning and business intelligence. We address interfaces for three important user groups: data scientists, data workers and end users.

Dietrich Kammer, Mandy Keck, Andreas Both, Giulio Jacucci, and Rainer Groh. 2018. VisBIA 2018: workshop on Visual Interfaces for Big Data Environments in Industrial Applications. In Proceedings of the 2018 International Conference on Advanced Visual Interfaces (AVI ’18). ACM, New York, NY, USA, Article 12, 3 pages. DOI: https://doi.org/10.1145/3206505.3206603.

Exploring Big Data Landscapes with Elastic Displays (September 2017)


In this paper, we propose a concept to help data analysts to quickly assess parameters and results of cluster algorithms. The presentation and interaction on a flexible display makes it possible to grasp the function- ing of algorithms and focus on the data itself. Two interaction concepts are presented, which demonstrate the strength of elastic displays: a layer concept that allows the recognition of differences between various parameter settings of cluster algorithms, and a Zoomable User Interface, which encourages the in-depth analysis of clusters.

Kammer, D., Keck, M., Müller, M., Gründer, T. & Groh, R. (2017). Exploring Big Data Landscapes with Elastic Displays. In: Burghardt, M., Wimmer, R., Wolff, C. & Womser-Hacker, C. (Hrsg.), Mensch und Computer 2017 – Workshopband. Regensburg: Gesellschaft für Informatik e.V., DOI: http://doi.org/10.18420/muc2017-ws08-0342.

A Framework for Training Hybrid Recommender Systems (August 2017)

Recommender Systems (RS) are widely used to provide users with personalized suggestions taken from an extended variety of items. One of the major challenges of RS is the accuracy in cold-start situations where little feedback is available for a user or an item. Exploiting available user and item metadata helps to cope with this problem. We propose a hybrid training framework consisting of two predictors, a collaborative filtering instance and a metadata-based instance relying on content and demographic data. Our framework supports a wide range of algorithms to be used as predictors. The cross-training mechanism we design minimizes the weaknesses of one instance by updating its training with predicted data from the other instance. A sophisticated sampling function selects ratings to be predicted for cross-training We evaluate our framework conducting multiple experiments on the MovieLens 100K dataset, simulating different scenarios including user and item cold-start. Our framework outperforms state-of-the-art algorithms and is able to provide accurate predictions across all tested scenarios.

Simon Bremer, Alan Schelten, Enrico Lohmann, Martin Kleinsteuber: A Framework for Training Hybrid Recommender Systems, In: Proceedings RecSysKTL 2017, 27 August 2017, Como, Italy

Towards Glyph-based Visualizations for Big Data Clustering (August 2017)

Data Analysts have to deal with an ever-growing amount of data resources. One way to make sense of this data is to extract features and use clustering algorithms to group items according to a similarity measure. Algorithm developers are challenged when evaluating the performance of the algorithm since it is hard to identify features that influence the clustering. Moreover, many algorithms can be trained using a semi-supervised approach, where human users provide ground truth samples by manually grouping single items. Hence, visualization techniques are needed that help data analysts achieve their goal in evaluating Big data clustering algorithms. In this context, Multidimensional Scaling (MDS) has become a prominent visualization tool. In this paper, we propose a combination with glyphs that can provide a detailed view of specific features involved in MDS. In consequence, human users can understand, adjust, and ultimately improve clustering algorithms. We present a thorough glyph design, which is founded in a comprehensive survey of related work and report the results of a controlled experiments, where participants solved data analysis tasks with both glyphs and a traditional textual display of data values.

Mandy Keck, Dietrich Kammer, Thomas Gründer, Thomas Thom, Martin Kleinsteuber, Alexander Maasch, and Rainer Groh. 2017. Towards Glyph-based visualizations for big data clustering. In Proceedings of the 10th International Symposium on Visual Information Communication and Interaction (VINCI ’17). ACM,  DOI: https://doi.org/10.1145/3105971.3105979. New York, NY, USA, 129-136.

A Zoomable Product Browser for Elastic Displays
(July 2017)


In this paper, we present an interaction and visualization concept for elastic displays. The interaction concept was inspired by the search process of a rummage table to explore a large set of product data. The basic approach uses a similarity-based search pattern – based on a small set of items, the user refines the search result by examining similar items and exchanging them with items from the current result. A physically-based approach is used to interact with the data by deforming the surface of the elastic display. The presented visualization concept uses glyphs to directly compare items at a glance. Zoomable UI techniques controlled by the deformation of the elastic surface allow to display different levels of detail for each item.

Mathias Müller, Mandy Keck, Thomas Gründer, Natalie Hube and Rainer Groh: A Zoomable Product Browser for Elastic Displays, In: Proceedings xCoAx 2017, 5-7 July 2017, Lisbon, Portugal