Classifying stack overflow posts on API issues
Bibliographic record
Abstract
The design and maintenance of APIs are complex tasks due to the constantly changing requirements of its users. Despite the efforts of its designers, APIs may suffer from a number of issues (such as incomplete or erroneous documentation, poor performance, and backward incompatibility). To maintain a healthy client base, API designers must learn these issues to fix them. Question answering sites, such as Stack Overflow (SO), has become a popular place for discussing API issues. These posts about API issues are invaluable to API designers, not only because they can help to learn more about the problem but also because they can facilitate learning the requirements of API users. However, the unstructured nature of posts and the abundance of non-issue posts make the task of detecting SO posts concerning API issues difficult and challenging. In this paper, we first develop a supervised learning approach using a Conditional Random Field (CRF), a statistical modeling method, to identify API issue-related sentences. We use the above information together with different features of posts and experience of users to build a technique, called CAPS, that can classify SO posts concerning API issues. Evaluation of CAPS using carefully curated SO posts on three popular API types reveals that the technique outperforms all three baseline approaches we consider in this study. We also conduct studies to test the generalizability of CAPS results and to understand the effects of different sources of information on it.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.012 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.004 | 0.002 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.000 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".