The natural language descriptions of the capabilities of manufacturing companies can be found in multiple locations including company websites, legacy system databases, and ad hoc documents and spreadsheets. To unlock the value of unstructured capability data and learn from it, there is a need for developing advanced quantitative methods supported by machine learning and natural language processing techniques. This research proposes a hybrid unsupervised learning methodology using K-means clustering and topic modeling techniques in order to build clusters of suppliers based on their capabilities, automatically infer topics from the created clusters, and discover nontrivial patterns in manufacturing capability corpora. The capability data is extracted either directly from the website of manufacturing firms or from their profiles in e-sourcing portals and directories. Feature extraction and dimensionality reduction process in this work are supported by N-gram extraction and latent semantic analysis (LSA) methods. The proposed clustering method is validated experimentally based on a dataset composed of 150 capability descriptions collected from web-based sourcing directories such as the Thomas Net directory for manufacturing companies. The results of the experiment show that the proposed method creates supplier cluster with high accuracy. Two example applications of the proposed framework, related to supplier similarity measurement and automated thesaurus creation, are introduced in this paper.

Supply Chain Management has been an important issue for competitiveness in today's market. Different decision-support models focus on specific time horizons and goals. The most common way of structuring supply chain planning process is dividing it in three different levels: operational, tactical and strategic. However, the planning on one level generally does not communicate with the others, limiting its efficiency and feasibility. The present work proposes a communication procedure between tactical and operation support-decision models in order to coordinate both, thus improving their overall performance. The procedure will be applied to a test case comprising a Spare Parts Supply Chain (SPSC) problem for maintenance in production facilities with the concept of integrating spare parts supply chain and intelligent maintenance systems implemented.

The descriptions of capabilities of manufacturing companies can be found in multiple locations including company websites, legacy system databases, and ad hoc documents and spreadsheets. The capability descriptions are often represented using natural language. To unlock the value of unstructured capability information and learn from it, there is a need for developing advanced quantitative methods supported by machine learning and natural language processing techniques. This research proposes a multi-step unsupervised learning methodology using K-means clustering and topic modeling techniques in order to build clusters of suppliers based on their capabilities, extract and organize the manufacturing capability terminology, and discover nontrivial patterns in manufacturing capability corpora. The capability data is extracted either directly from the website of manufacturing firms or from their profiles in e-sourcing portals and directories. Feature extraction and dimensionality reduction process in this work in supported by Ngram extraction and Latent Semantic Analysis (LSA) methods. The proposed clustering method is validated experimentally based a dataset composed of 150 capability descriptions collected from web-based sourcing directories such as the Thomas Net directory for manufacturing companies. The results of the experiment show that the proposed method creates supplier cluster with high accuracy.

Manufacturing capability (MC) analysis is a necessary step in the early stages of supply chain formation. In the contract manufacturing industry, companies often advertise their capabilities and services in an unstructured format on the company website. The unstructured capability data usually portray a realistic view of the services a supplier can offer. If parsed and analyzed properly, unstructured capability data can be used effectively for initial screening and characterization of manufacturing suppliers specially when dealing with a large pool of suppliers. This work proposes a novel framework for capability-based supplier classification that relies on the unstructured capability narratives available on the suppliers' websites. Four document classification algorithms, namely, support vector machine (SVM ), Naïve Bayes, random forest, and K-nearest neighbor (KNN) are used as the text classification techniques. One of the innovative aspects of this work is incorporating a thesaurus-guided method for feature selection and tokenization of capability data. The thesaurus contains the formal and informal vocabulary used in the contract machining industry for advertising manufacturing capabilities. A web-based tool is developed for the generation of the concept vector model associated with each capability narrative and extraction of features from the input documents. The proposed supplier classification framework is validated experimentally through forming two capability classes, namely, heavy component machining and difficult and complex machining, based on real capability data. It was concluded that thesaurus-guided method improves the precision of the classification process.

No abstract