AMAS: A DNN-Based Automatic Manhours Approval System
Yunfei Jia, Hao Ran Lu, Debin Hu · 2022
Manhours estimation and automatic approval plays an important role in project scheduling and reducing the development cost for factory. Traditional approaches usually have some limitations, such as technical text formatting limitations and industry-specific limitations. This paper presents a software system that estimates the manhours and approves the technology texts that have been submitted. Firstly, we collect a technology texts dataset and complement the dictionaries. Secondly, the word2vec is used to extract the features implicated in the technology text, and a DNN (deep neural network) model is proposed to estimate the manhours. Finally, an Odoo-based software system is developed, which provides the estimation manhours and approves automatically of the submitted technology texts. The advantages of approach in which is proposed in this paper is that, it has less limitations on the format of submitted technology texts. And, it can be used for various production technology.