Interactive, Orthogonal Hyperedge Routing in Schematic Diagrams Assisted by Layout Automatisms

Schematic diagrams are used in graph-based engineering systems. They focus mainly on the structure of the design object. Graph-based engineering systems help to solve a concrete design task. This is primarily realized by the application of domain-specific languages. The layout of schematic diagrams is of particular importance, and a neat representation is desirable. But automatically generated layouts cannot always fully match the intention of a modeler. To improve automatic layouts and enable a user-specific representation, an algorithm that allows interactive changes of the orthogonal hyperedge geometry was implemented. In this paper, we present this algorithm and give an overview of such interactions. Additionally, several reductions of the hyperedge geometry are shown. Furthermore, a local, automatic routing considering interactions on the hyperedge geometry is presented. The consideration of domain-specific semantics and the possibility of interactive changes is a new approach. All algorithms were implemented in a self-developed software framework.

Dynamic Language Binding in Relational Visual Reasoning

We present Language-binding Object Graph Network, the first neural reasoning method with dynamic relational structures across both visual and textual domains with applications in visual question answering. Relaxing the common assumption made by current models that the object predicates pre-exist and stay static, passive to the reasoning process, we propose that these dynamic predicates expand across the domain borders to include pair-wise visual-linguistic object binding. In our method, these contextualized object links are actively found within each recurrent reasoning step without relying on external predicative priors. These dynamic structures reflect the conditional dual-domain object dependency given the evolving context of the reasoning through co-attention. Such discovered dynamic graphs facilitate multi-step knowledge combination and refinements that iteratively deduce the compact representation of the final answer. The effectiveness of this model is demonstrated on image question answering demonstrating favorable performance on major VQA datasets. Our method outperforms other methods in sophisticated question-answering tasks wherein multiple object relations are involved. The graph structure effectively assists the progress of training, and therefore the network learns efficiently compared to other reasoning models.

Achieving Complete and Near-Lossless Conversion from IFC to CityGML

The Singapore Government has embarked on a project to establish a three-dimensional city model and collaborative data platform for Singapore. The research herein contributes to this endeavour by developing a methodology and algorithms to automate the conversion of Building Information Models (BIM), in the Industry Foundation Classes (IFC) data format, into CityGML building models, capturing both geometric and semantic information as available in the BIM models, and including exterior as well as interior structures. We adopt a Triple Graph Grammar (TGG) to formally relate IFC and CityGML, both semantically and geometrically, and to transform a building information model, expressed as an IFC object graph, into a city model expressed as a CityGML object graph. The work pipeline includes extending the CityGML data model with an Application Domain Extension (ADE), which allows capturing information from IFC that is relevant in the geospatial context but at the same time not supported by CityGML in its standard form. In this paper, we elaborate on the triple graph grammar approach and the motivation and roadmap for the development of the ADE. While a fully complete and lossless conversion may never be achieved, this paper suggests that both a TGG and an ADE are natural choices for supporting the conversion between IFC and CityGML.