The current limitations show that the translation approach is a bottleneck for evaluating HEX-programs. Overcoming them requires to intervene in the model building process, which can only be done by developing native evaluation algorithms for HEX-programs. This leads us to the following project goals, which are discussed in more detail in Section 3: 1. Scalability. The current use of HEX-programs is limited because of a lack of advanced model building and reasoning algorithms that scale to real-world instances of applications, roughly measured by the number of external atoms which occur in the program. The overall goal of the project is thus to substantially increase the number of external atoms which can be handled efficiently, relative to the intrinsic complexity of program evaluation. Depending on the program structure, nearly an exponential speed-up should be achieveable in some relevant cases, while a prediction on average increase is difficult to make. 2. Native Algorithms. Rather than to compute models of a programs from a compilation to ordinary ASP, we want to have native algorithms to evaluate programs with external access which compute models from first principles. While several native algorithms for ordinary ASP programs exist, no such algorithms exist for programs with external source access and will be completely novel. As they work on a lower level than compilation, more effective opportunities for optimization exist that can be exploited to obtain scalability. This will require the use and development of respective techniques, which therefore is a central goal of the project. 3. Evaluation Framework. The evaluation shall be managed by a newly developed evaluation framework. Its task is to analyze the input program, decomposes the overall evaluation into subtasks, for which it selects and runs the most efficient algorithms. Subsequently it combines their results into the result for the whole program. To this end, multiple concrete algorithms for different types of programs can be plugged in. 4. Enriched External Domains. The current evaluation strategy requires strong syntactic safety conditions on the input program. Since this reduces expressiveness, a better handling of external domains shall lead to more liberal conditions. 5. Prototype. As a proof of concept, one of the goals will be the provision of a prototypical implementation of the new algorithms and the evaluation framework, yielding a novel evaluation platform. The system prototypes shall be public and freely available open source software.