Browsing by Author "Schwarz, Tobias"

Filter results by typing the first few letters
Now showing 1 - 2 of 2
  • No Thumbnail Available
    Item
    Design and Assessment of an Engine for Embedded Feature Annotations
    (2021-03-03) Schwarz, Tobias; Göteborgs universitet/Institutionen för data- och informationsteknik; University of Gothenburg/Department of Computer Science and Engineering
    Features are an inherent unit of development of every software; and are defined as a set of implementation artifacts that constitute a functionality that adds value to the product, and is perceived useful by the customer. Locating features in source code is a typical software developer task, whether it before implementing a new feature, or maintaining and bug fixing of existing ones, as it is essential to know where to make changes. For tracing features to their implementation, two mechanisms can be used; external and internal documentation. As the names imply, external documentation refers to maintaining the traceability links externally, whereas internal documentation involves labeling assets inside the source code (aka embedded annotations). For internal documentation, two strategies are used namely eager and lazy approaches. The former involves annotating the code artifacts during development, whereas the latter requires extracting feature-related information from an un-annotated codebase based on heuristics. The former, although involves some added effort, result in significant returns in terms of accuracy and degree of reuse, also enabling a wider range of analyses. Also, the added effort can be minimal depending on the size of the project but soon begins to prove its worth in the short-run (when aiming to reuse) as well as the long run (when maintaining the code base). Embedded annotations (with eager strategy) allow a minimally invasive and almost cost-neutral way to document the product features inside source code. This brings some benefits, the significant ones being easier co-evolution of code and traceability links, elimination of feature location, and ease in tasks like feature and artifact reuse (cloning) and maintenance (propagation). Several approaches exist today on how to document features in source code. Different definitions lead to different implementations and therefore, reuse is not directly possible. This work tackles exactly this issue and provides a unified design of embedded annotations with a freeto-use reference library according to the presented specification. The functionality of this library, aka. engine, is shown on the use case of partial feature-based commits. Feature centric development, which is typical for agile projects get the possibility for isolated source code commits based on specific features aka. embedded annotations.
  • No Thumbnail Available
    Item
    An Empirical Survey of Bandits in an Industrial Recommender System Setting
    (2023-09-21) Schwarz, Tobias; Brandby, Johan; Göteborgs universitet/Institutionen för data- och informationsteknik; University of Gothenburg/Department of Computer Science and Engineering
    In this thesis, the effects of incorporating unstructured data—images in the wild—in contextual multi-armed bandits are investigated, when used within a recommender system setting, which focuses on picture-based content suggestion. The idea is to employ image features, extracted by a pre-trained convolutional neural network, and study the resulting bandit behaviors when including respective excluding this information in the typical context creation, which normally relies on structured data sources—such as metadata. The evaluation is made both online, through A/B-testing enabled by the industrial partner YouPic AB, and offline, effectuated by a simulation pipeline that models the online counterpart. The results are compiled as a survey, covering a selection of contextual bandit algorithms, highlighting the differences brought by the unstructured data. The offline result points towards that if the contextual bandit utilizes a joint or hybrid action-value function, with respect to the parameterization, the addition of the image vectors can significantly outperform the instances without it; however, if a disjoint model is instead employed, no noticeable change is observed. In comparison, those from the online trials can be interpreted as supporting the inclusion of convolutional features, but due to meager and unbalanced sample sizes, the outcomes are deemed inconclusive. To summarize, though there is support for incorporating unstructured data, given that the action-value function is joint or hybrid, the online experiments gave too little evidence for any trustworthy findings; in other words, the question is still partially open.