知識流探勘與文件推薦之整合研究

Abstract

知識是企業組織最重要的資產，也是獲取競爭優勢的來源。在面對快速變動的環境 中，組織必須使用有效的方法來管理知識，以幫助知識工作者找尋工作之相關知識，促 進組織中的知識保存、分享與再利用。因此，如何根據知識工作者過去工作紀錄，發掘 個人或群體之工作知識流動過程，以瞭解其資訊需求與文件參考行為，進而主動提供適 切之知識支援，為重要的研究議題。本計畫提出知識流探勘與知識文件推薦整合機制之 研究，運用資料探勘等相關資訊技術，進行個人與群體知識流之探勘，並考慮知識資源 的參考時間與引用關係，進而以知識流為基礎，設計知識文件之推薦方法。此外，更進 一步整合知識流、社群概念與推薦技術，發展一整合性之知識文件推薦機制。 本計畫提出知識流探勘與知識文件推薦整合研究，研究內容主要包括：(1)分析知識 工作者資訊需求與參考知識文件的特性，設計以個人知識流與群體知識流為基礎之文件 推薦方法；(2)探討如何運用資料探勘與資訊擷取技術分析知識文件、知識工作者的資訊 需求與參考知識文件之行為，以發掘個人知識流；(3)設計知識流調整之方法，以調整知 識流中文件與知識主題之重要性順序；(4)探討如何利用個人知識流之間的相關程度，建 立知識工作者的知識同好群組; (5)探討如何設計知識流的操作機制，以整合知識同好群 組中知識工作者的知識流; (6) 研究知識流的特性，探討如何整合知識流與推薦方法，以 有效地推薦相關知識文件與相關知識主題順序; (7)探討如何透過社群機制以達到組織知 識管理的運作; (8)發展與實作以個人知識流與群體知識流為基礎之文件推薦方法，並進 行方法的評估;以及發展與實作以知識流為基礎之社群知識推薦方法，並進行方法的評 估。

Knowledge is a critical property in organizations for obtaining competitive advantages. In constantly changing environments, organizations have to exploit effective and efficient approaches to manage knowledge for assisting knowledge workers to find task-relevance knowledge and encouraging organizations to preserve, share and reuse knowledge. Hence, an important issue is how to discover knowledge flow (KF) from historical working records of individual workers or a group of workers for understanding their task-needs and referencing behavior of documents, and actively providing adaptive knowledge support. This project proposed an integrated research which combines KF mining with recommendation mechanisms to recommend codified knowledge. Applying data mining and other related information techniques, KF mining, which takes the reference time and citation relations of knowledge resources into account, identifies KF of individual and a group of knowledge workers. Additionally, the flow-based recommendation schemes are developed to recommend codified knowledge. Moreover, such schemes are further extended as an integrated recommending mechanism which consolidates knowledge flow, concepts of community of practice and recommendation methods. The research directions of this project mainly include: (1) analyzing the task-needs and referencing characteristics to develop flow-based recommendation methods based on individual or a group of KF; (2) employing data mining and information retrieval techniques to analyze codified knowledge, task-needs and referencing behavior of knowledge workers; (3) designing an adjusting mechanism of KF for adjusting documents and knowledge topics in the order of significance; (4) exploiting the similarity of KF to build a knowledge interest group; (5) designing the operations of KF to integrate similar workers』 KFs into a group-based KF; (6) combing KF with recommendation approaches to effectively recommend the referencing patterns of relevant documents and topics; (7) investigating how to manage knowledge effectively in organizations through communities of practice; (8) developing, implementing and evaluating the proposed document recommendation approaches based on KF and communities of practice.