Characterizing Task-Relevant Information in Natural Language Software Artifacts
Bibliographic record
Abstract
Contains the supplementary material for the paper "Characterizing Task-Relevant Information in Natural Language Software Artifacts". All contents are explained in the file README.md. Abstract: To complete a software development task, a software developer often consults artifacts that contain largely natural language text, such as API documentation, bug reports, or Q&A forums. Not all information within these artifacts is relevant to a developer's current task forcing the developer to filter relevant information from large amounts of irrelevant information, a frustrating and time-consuming activity. Since failing to locate relevant information may lead to incorrect or incomplete solutions, many approaches mine potentially relevant text from such natural language artifacts. However, existing approaches identify text relevant for only certain categories of tasks (e.g., learning an API) and from a restricted set of artifact types. To explore how limitations on software development tasks and artifact types can be relaxed in future approaches, we conducted an experiment in which 20 participants identified which text appearing in 1874 sentences across 20 artifacts was relevant to six software development tasks. Participants created 2,463 distinct highlights in these sentences to indicate relevance. Although the results indicate variability in the text perceived as relevant, we observe consistency in the information considered key for task completion. The semantic meaning of relevant information, as identified through semantic frames, shows promise to automate the identification of relevant text. We discuss implications of our study for future research in the field.
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 | 0.003 | 0.047 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.028 | 0.006 |
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".