Predicting Tags for Learner Questions on Stack Overflow
Bibliographic record
Abstract
Online question answering sites, such as Stack Overflow (SO), have become an important learning and support platform for computer-science learners and practitioners who are seeking help. Learners on SO are currently faced with the problem of unanswered questions, inhibiting their lifelong-learning efforts and contributing to delays in their software development process. The major reason for this problem is that most of the technical problems posted on SO are not seen by those who have the required expertise and knowledge to answer a specific question. This issue is often attributed to the use of inappropriate tags when posting questions. We developed a new method, BERT-CBA, to predict tags for answering user questions. BERT-CBA combines a convolutional network, BILSTM, and attention layers with BERT. In BERT-CBA, the convolutional layer extracts the local semantic features of an SO post, the BILSTM layer fuses the local semantic features and the word embeddings (contextual features) of an SO post, and the attention layer selects the important words from a post to identify the most appropriate tag labels. BERT-CBA outperformed four existing tag recommendation approaches by 2-73% as measured by F1@K=1-5. These findings suggest that BERT-CBA could be used to recommend appropriate tags to learners before they post their question which would increase their chances of getting answers.
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 | 0.002 | 0.009 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.004 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.002 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.002 |
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".