Test-driven Software Development Research Articles

Developing of NLP models by model based software developement

In the case of software development, we can talk about several methodologies, such as the waterfall model, test-driven software development, agile software development, but the question rightly arises whether software development methods in the traditional sense are applicable to the development of natural language processing language models, especially from the aspect of AI and dictionary-based NLP models. Learning AI models is substantially different from the classical software development process. Whereas in classical software development, developers explicitly describe the operation and behavior to the computer, in AI model learning, the models themselves learn from the input data. This paper presents a possible solution that applies the agile Scrum methodology used in classical software development to the development of dictionary-based NLP models, and identify the agile development opportunities in case of the machine learning-based NLP models.

A novel program synthesis approach in test driven software development

It is a viable alternative to automatically generate Java source code based on the specification provided by the associated unit tests. This possibility may seem far-fetched in the general case, but after considering the most common restrictions, which are applied nowadays as best practice, it turns out that a significant part of the production code can be generated automatically. The goal is to generate viable implementations, which fulfill the requirements imposed by unit tests. According to the presented vision the modern test frameworks, development guidelines and computational capacities make it possible to reach this goal.

Life Cycle VV&A Simulation Modeling Process Based on Validation

Simulation modeling VV&A is verification, validation and accreditation. In the field of petrochemical industry, focus of VV&A still lays on the model validation, and people always do appropriate model verification, validation work after model development. It does not yet formed a complete hierarchy or framework for effective implementation of life cycle VV&A. This paper referenced the research achievement of military simulation modeling and test-driven software development technology, introduced the modeling VV&A technology into the whole life of the model development. Simulation modeling process based on validation discussed in this paper is totally based on validation, and it is carried out closely around VV&A from appearance of the first idea to the final application. As a result, people can completely control the whole process of model development. At the same time of improving model accuracy, reliability and credibility, such process could greatly improve the efficiency of model development and effectively reduce the model development period. And this process can be also applied to system-level, complex object.