Existing scholarly information extraction (SIE) datasets focus on scientific papers and overlook implementation-level details in code repositories. README files describe datasets, source code, and other implementation-level artifacts, however, their free-form Markdown offers little semantic structure, making automatic information extraction difficult. To address this gap, NERdME is introduced: 200 manually annotated README files with over νm10000 labeled spans and 10 entity types. Baseline results using large language models and fine-tuned transformers show clear differences between paper-level and implementation-level entities, indicating the value of extending SIE benchmarks with entity types available in README files. A downstream entity-linking experiment was conducted to demonstrate that entities derived from READMEs can support artifact discovery and metadata integration.

Dataset distillation is an increasingly prevalent technique for condensing a large-scale dataset into more compact versions while preserving their intrinsic utility. However, very few studies have investigated the trustworthiness of data distillation, i.e., privacy, robustness, and fairness. The deficiency is particularly striking given the existing research that underscores the vulnerabilities in current AI models, including privacy breaches, biased predictions against underrepresented subgroups, and susceptibility to imperceptible attacks. To bridge the gap, we propose a trustworthy benchmark for assessing representative dataset distillation solutions across the benchmark CIFAR10 with comprehensive evaluation metrics. Through extensive experiments, we uncover vulnerabilities inherent in the application of dataset distillation, offering valuable insights for practitioners. Our work aims to drive the development of more transparent, reliable, and responsible machine learning models, fostering AI systems that align with trustworthy principles.

Dataset distillation is attracting more attention in machine learning as training sets continue to grow and the cost of training state-of-the-art models becomes increasingly high. By synthesizing datasets with high information density, dataset distillation offers a range of potential applications, including support for continual learning, neural architecture search, and privacy protection. Despite recent advances, we lack a holistic understanding of the approaches and applications. Our survey aims to bridge this gap by first proposing a taxonomy of dataset distillation, characterizing existing approaches, and then systematically reviewing the data modalities, and related applications. In addition, we summarize the challenges and discuss future directions for this field of research.

Visual programming languages are a common way for enabling end users to create simple, customized programs in their smart homes. Although visual programming aims at lowering the syntactic and cognitive barriers existing languages lack a simple method for accessing services of smart spaces surrounding the home and therefore restrain the innovative capabilities of users. For example, accessing information about energy consumption or smart parking is often cumbersome and limited to users with advanced technical skills. Based on expert interviews, we extended the existing visual programming language Node-RED with functionalities that enable users to easily access data and services within and beyond their smart homes. Using this prototype, we performed a first evaluation of our solution with some inhabitants of smart homes.

The DChanges series of workshops focuses on changes in all their aspects and applications: algorithms to detect changes, models to describe them and techniques to present them to the users are only some of the topics that are investigated. This year, we would like to focus on collaboration tools for non-textual documents.

Nowadays, touch-input devices are widely available. The use of such touch input devices, e.g., large wall-mounted touchscreens in (team) meeting rooms appear appropiate and desirable. Thus, team meetings can change from one-man-presentation-shows towards collaborative and interactive developing of plans and processes via designing and creating its corresponding diagrams. In this paper, we present an approach for interacting with diagrams using uni-stroke touch gestures. We focus on large wall-mounted touchscreens and present an editor design for such environments. In order to validate the usability of our approach we report on the results of a user study with a diagram editor for Business Process Modeling Networks.

The DChanges series of workshops focuses on changes in all their aspects and applications: algorithms to detect changes, models to describe them and techniques to present them to the final users are only some of the topics we investigate. The workshop is open to researchers and practitioners from industry and academia. This year, we would like to focus on the application of change-tracking and diff algorithms to documents that are collaboratively edited via the web. We will also follow up on the discussion of DChanges 2015 about algorithms and interfaces to better understand and exploit detected changes, and about standards for modeling and transmitting changes. Particular attention will also be given to the use of diff, change tracking, and versioning techniques in the field of digital humanities.

In software development processes, diagram editors play an important role. For instance, editors for creating UML diagrams are often used in early stages of software development. Three input methodologies can be distinguished: traditional mode-based editors, sketching editors, and gesture-based editors. Most of today's diagram editors are mode-based and use the mouse as an input device. (Multi-)touch inputs have not been integrated very well, yet, although a variety of touch input devices, such as (multi-)touch monitors and tablet computers, are quite common today. In this paper, we present an integration of an off-the-shelf unistroke recognizer into an existing diagram editor framework. This allows to reduce the number of different interaction modes in diagram editors as users perform editing tasks, such as creating, modifying, and deleting, by unistroke gestures. In order to examine the effect on the usability of such editors, we conducted a pilot study using a Statechart editor and present some preliminary findings here.

In software development processes, diagram editors play an important role. For instance, editors for creating UML diagrams are often used in early stages of software development. Three input methodologies can be distinguished: traditional mode-based editing, sketch-based editing, and gesture-based editing. Most of today's diagram editors are mode-based and use the mouse as an input device. (Multi-)touch inputs have not been integrated very well, although a variety of touch input devices, such as (multi-)touch monitors and tablet computers, are quite common today. In this paper, we provide some insights into the integration of an off-the-shelf unistroke gesture recognizer into an existing diagram editor framework, and give some details about three example diagram editors that incorporate a new interaction mode, the so-called gesture mode.

The DChanges series of workshops focuses on changes in all their aspects and applications: algorithms to detect changes, models to describe them and techniques to present them to the final users are only some of the topics we investigate. The workshop is open to researchers and practitioners from industry and academia.

In this paper, we present methods and techniques to record, process, and visualize diagram changes. These methods help us understand diagram changes, and are the basis for more enhanced functionality, such as a more powerful undo functionality, or merging of different versions of a diagram. Our approach is based on the idea to record all diagram changes, including changes of the diagram layout. One of the main challenges is to provide a concise and comprehensible visualization of these changes. In order to produce such a visualization, the recordings need to be processed, comprising filtering and aggregation of changes. Our model-based approach is generic and makes use of the abstract and the concrete syntax (meaning and representation) of the specific diagram language.

With collaborative tools getting more and more widespread, users have started to become accustomized to features like automatic versioning of their documents or the visualization of changes made by other users. The research community, however, sees that the state of the current tools is seriously lack lusting. The second edition of the DChanges workshop focuses on these shortcomings, introducing new ways to produce version-aware documents and merge changes from multiple sources. Other aspects - in particular, the standardization of formats for tracking changes - are discussed, too.

Many people have approached the problem of investigating the evolution of documents and data from different perspectives, e.g. by tracking changes, versioning and diffing. The goal of this workshop is to share ideas, common issues and principles, and to foster research collaboration on these topics.

Graphs are known to be well-suited as an intermediate data structure in diagram editors. The syntax of a particular visual language can be defined by means of a graph grammar. In recent work we have proposed approaches to graph completion: given a possibly ldquoincompleterdquo graph, this graph is modified in such a way that the resulting graph is a member of the grammarpsilas language. In this paper we describe how graph completion can be used to realize diagram completion, an important requirement for the realization of content assist in diagram editors. With our approach, the advantages of free-hand and structured editing can be effectively combined: drawing of diagrams with maximal freedom and powerful guidance whenever needed.