A hybrid deep learning model for user story effort estimation
Saadia Malik, Muhammad Hamid, Muhammad Saleem · 2026
Read A hybrid deep learning model for user story effort estimation by Saadia Malik, Muhammad Hamid, Muhammad Saleem (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0353348.txt.
Description
Saadia Malik, Muhammad Hamid, and Muhammad Saleem teach a computer a hard judgement call. How much work will a software task take? Their study appeared in PLOS ONE in 2026. Agile teams describe work as short 'user stories'. Guessing their effort is notoriously unreliable. The authors gathered thousands of real stories from many sources. They cleaned them down to about four thousand sound examples. Then they trained a hybrid model. One part reads language with a pre-trained transformer. The other tracks word order with a memory network.
On held-out stories the model beat conventional methods. Its error was clearly lower. Its scores were then mapped onto the point scale. That is the scale planning teams actually use. Reliable estimates mean fewer blown deadlines in software projects.
Catalog record
| CREATOR | Saadia Malik, Muhammad Hamid, Muhammad Saleem |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 55213 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |