Software that needs to fulfill many tasks requires a large number of components. Users of these software need a lot of time to find the desired functionality or follow a particular workflow. Recommendation systems can optimize a user’s working time by recommending the next features he/she needs. Given that, we evaluate the use of three algorithms (Markov Chain, IndRNN, and LSTM) commonly applied in sequence recommendation/classification in a dataset that reflects the use of the accounting software from Fortes Tecnologia. We analyze the results under two aspects: accuracy for top-5 recommendations and training time. The results show that the IndRNN achieved the highest accuracy, while the Markov Chain reached the lowest training time.
CITATION STYLE
Prata, V. R., Moreira, R. S., Cordeiro, L. S., Maia, Á. N., Martins, A. R., Leão, D. A., … Rocha Neto, A. R. (2019). A novel recommendation system for next feature in software. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 11871 LNCS, pp. 494–501). Springer. https://doi.org/10.1007/978-3-030-33607-3_53
Mendeley helps you to discover research relevant for your work.