藉由網路新聞探勘供應商關係

Abstract

隨著時代的演進，企業興衰不再只是單純倚賴公司本身的績效，而是需要靠著整體供應鏈的興盛，才能達到持續經營以及獲利的目標，每一間公司不再只是產業中的個體，而是商業生態系統中的一部份。一個商業生態系統要穩定成長，必須仰賴其中成員們能夠良好地運作，其中一個環節出了問題，都會影響整個商業生態系統。所以每個企業都必須充分掌握自己供應鏈的狀態，基本上企業對於本身的供應商和顧客都很熟悉，但是再更深一層的關係，就不是這麼清楚了，很可惜的是，大部分企業的供應鏈資訊都沒有公開，想要獲得此資訊的成本相當高。一般進行企業資訊蒐集的研究，其資訊來源都是網路新聞，因此本研究也是希望能從網路新聞中，探勘出企業的供應商關係。 本研究利用文字探勘以及資料探勘的技術為基礎，建立分析供應鏈關係探勘系統的模型。其中首先會利用文字探勘的技術，將網路新聞文件進行分類，將含有供應商關係資訊的句子，以及供應商關係方向（例如：甲公司為乙公司的供應商）分類出來，再利用這些關係建立各公司的供應商/顧客關係圖，最後再藉由本研究基於關係圖所提出的十一個變數，進行連結評估，得到修正的供應商/顧客關係圖。 本研究提出的供應鏈關係探勘系統能夠自動地從網路新聞文件中擷取供應商資訊，並且建立關切公司的供應商/顧客關係圖，經過修正後，將能夠提供十分準確的供應商資訊給使用者。

As the time past, the prosperity of a company is no longer just related to the performance of itself, or wins the competition with its competitor, but the success of its entire supply chain. The company must endeavor to consolidate the whole supply chain and make it thriving and prosperous. A company cannot just be viewed as a member of a single industry but as part of a business ecosystem that crosses a variety of industries. A company should have knowledge of and closely monitor its complete supply chain and ensure that all or most of the members of its supply chain well-functioning. It can be expected that a company must know about its customers and suppliers, but rarely know about the customers’ customers and the suppliers’ suppliers. In addition, the competition between two companies evolves to not just only companies go head-to-head in one industry, but supply chain versus supply chain or we can scale it up, business ecosystem versus business ecosystem. But as we know, the supplier information of a company is usually private. If a company wants to recognize the supply chains of other companies, it must be costly to investigate the business environment. When gathering business information, most of us rely on finance news. Business news documents often reveal various types of business events and relationships, including suppliers of companies. Thus, we want to mine the supplier information of a focal company from online news documents. We exploit text mining and data mining techniques to construct the model of supplier relationship mining (SRM) system. First, by using text mining approach, we can classify the news documents, and extract the sentences with supplier relationships, as well as the directions of supplier relationships (i.e., A company is a supplier of B company). With these results, the system can construct a supplier/customer graph. On the basis of supplier/customer graph, we develop 11 variables to support link assessment. Finally, we can get a refined supplier/customer graph. We propose a supplier relationship mining (SRM) system that automatically discovers the supplier relationships concerning a focal company from news documents and generate a supplier/customer graph of a focal company. Through the refinement, we can give the user more accurate supplier information of a focal company.