本研究植基於本體理論、人力資源理論與系統發展理論,旨在發展不同型式的工作媒合系統,並進行媒合效能的比較。首先藉由探討國內外有關本體理論的相關文獻,主要在瞭解本體建構的方法。接著進行了一個先導研究,由國內外人力資源網站上蒐集了當前企業界所需要的軟體開發技能,配合本體理論文獻探討的結果,建構了一個軟體開發者所需要的資訊能力本體。本研究繼續探討了工作媒合系統的媒合方式與當前人力資源網站的搜尋缺點,並使用開放源碼的LAMP (Linux、Apache、MySQL與PHP)環境,發展了七種型式的工作媒合系統:單純關鍵字工作媒合系統、結構化關鍵字工作媒合系統、補償式工作媒合系統、比例式工作媒合系統、加權補償式工作媒合系統、里程碑式工作媒合系統與綜合式工作媒合系統。再藉由實際資料媒合的結果,分析不同媒合方式其媒合結果的差異,以及不使用本體作單純的媒合與使用本體作語意相似度式的媒合,其媒合結果的差異。研究結果發現,在單純關鍵字工作媒合系統與結構化關鍵字工作媒合系統中,可以使用本研究提出的技能重要性指標值,來達成篩選的功能。同時在一個工作媒合系統中,使用本體作語意式的延伸查詢,可以提升工作媒合的回覆率與相似度總分。在此七種工作媒合系統中,以里程碑式工作媒合系統有最高的媒合回覆率與最高的相似度總分。
The aim of the study was to develop different types of job matching systems based on Ontology Theory, human resource theory and system development theory. First, the study explored the literature related to Ontology to understand methodologies of ontology development. Then a pilot study was conducted to gather the skill requirements for software developers needed by current enterprises from human resource web sites nationwide and abroad. Based on the ontology development methodology and the result of the pilot study, an information competency ontology was built. The study continued to review the different types of job matching methods and their deficiencies. Using open source development platform LAMP (Linux, Apache, MySQL and PHP), the study implemented seven different types of job matching systems: simple keyword matching, structured keyword matching, compensated matching, proportional matching, weighted compensated matching, milestone matching and comprehensive matching system. The study analyzed the differences of matching results among job matching systems and the differences of matching results in a job matching system by using and not using ontologies. The result shows that using ontology for semantic searching can upgrade the recall and the similarity score within a job matching system. Among seven systems, the milestone job matching system has the highest recall and similarity score.