The use of object-oriented frameworks is often hampered by the complex interrelations among the individual components. Developers thus typically resort to examples they find in online articles, from which they identify the relevant components of the framework, retrieve the corresponding reference documentation, and generalize the examples to the problem at hand. This can get complicated because the examples are often based on outdated versions of the framework. This paper introduces SeaDoc, a self-adaptive document link provision system that facilitates this workflow.

The developer submits the uniform resource locator (URL) of an article with the example of interest to SeaDoc. From this, the system first identifies the framework by keywords. Based on the article’s publication date, SeaDoc then determines the current version of the framework at the time and the corresponding configuration and application programming interface (API) documents. SeaDoc then identifies the components relevant to the example and finally returns the URLs to the appropriate version of their documentation. An evaluation based on the Eclipse Corner Articles archive shows that the quality of the results is comparable to a manual selection of documents, but that such a selection would have to investigate 80 percent more documents.

While the qualification of SeaDoc as self-adaptable might be questioned (SeaDoc adapts its behavior just by considering the version of the framework and the context of the example), the paper nevertheless nicely demonstrates how, by clever processing of publicly available information, a specialized system may support a specific task much better than general-purpose web search engines.