Towards noise and error reduction on foundry data gathering processes

Microshrinkages are known as probably the most difficult defects to avoid in high-precision foundry. The presence of this failure renders the casting invalid, with the subsequent cost increment. Modelling the foundry process as an expert knowledge cloud allows properly-trained machine learning algorithms to foresee the value of a certain variable, in this case, the probability that a microshrinkage appears within a casting. Our previous research presented outstanding results with a machine-learning-based approach. Still, the data gathering phase for the training of these algorithms is performed in a manual way. Thereby, this learning process is subject to an accuracy reduction due to the noise introduced in such archaic data collection method. In this paper, we address the use of Singular Value Decomposition (SVD) and Latent Semantic Analysis (LSA) in order to reduce the number of ambiguities and noise in the dataset. Further, we have tested this approach comparing the results without this preprocessing step in order to show the effectiveness of the proposed method.