A Hybrid Method for Manufacturing Text Mining Based on Document Clustering and Topic Modeling Techniques

Abstract

As the volume of online manufacturing information grows steadily, the need for developing dedicated computational tools for information organization and mining becomes more pronounced. This paper proposes a novel approach for facilitating search and organization of textual documents and also extraction of thematic patterns in manufacturing corpora using document clustering and topic modeling techniques. The proposed method adopts K-means and Latent Dirichlet Allocation (LDA) algorithms for document clustering and topic modeling, respectively. Through experimental validation, it is shown that topic modeling, in conjunction with document clustering, facilitates automated annotation and classification of manufacturing webpages as well as extraction of useful patterns, thus improving the intelligence of supplier discovery and knowledge acquisition tools.