基於結構相似度之惡意程式原始碼分類研究

Malware Classification Based on Structure Similarity

目前競爭智慧（competitive intelligence）系統仍以企業內部的資訊傳遞為主，對於提供決策支援的延伸應用明顯不足。本研究聚焦於競爭智慧在延伸應用的改良，並訂定以「參與決策」為目標，嘗試在企業的組織管理與對手的新聞中，找出事件類型、決策模式、法則經驗及部門參與等的關係，據此建立競爭智慧的知識模型及應用系統。本研究以開發知識庫方式解題，並以設計知識模型為核心，主要設計包括：(1) 領域本體，建立共通性的知識概念及實例，利用is-a 關係表達知識分類架構，提供溝通及標準術語；(2) 任務本體，建立目標導向的知識框架，利用has-a 關係表達問題的組合；及(3) 語意規則，建立推論隱性知識所需的規則。由實驗結果顯示：在新聞已傳遞正確之前提下，評估「決策模式正確率」及「部門權重正確率」等二項，均已接近100% 的正確率，顯見本研究提出的知識模型，確可使競爭智慧系統由資訊傳遞的角色，提升到參與決策的延伸應用。

Existing Competitive Intelligence (CI) systems are mostly designed for CI information delivery and lack the extending application for decision support. This study focuses on the improvement of CI’s application for decision and takes participatory decision support as research objective. This study seeks to build a CI knowledge model and to develop an application system by identifying concepts of and relations among the instance types, decision models, rules of thumb, and functional participation from the enterprise organization and CI news about competitors. This study uses knowledge base system development as a solution. This study uses knowledge-based system development as the approach of solving the research problem with the design of knowledge model as the core. Major research design components of this study includes: (1) a Domain Ontology for establishing common knowledge concepts and instances using is-a relations to express the knowledge categorization structure and to provide a standard terminology set for ontology communication; (2) a Task Ontology to establish an objective-oriented knowledge framework using has-a relations to express the combination of questions; and (3) Semantic Rules to develop the rules for inferring implicit knowledge. The case experiment results show that, under the condition of correct delivery of news entries, the individual news entries have achieved close to 100% correctness in Decision Model Correctness Rate and Departmental Weighting Correctness Rate. It is evidenced that the knowledge model proposed in this study has extended the application of CI from information delivery to the extended support for participatory decision.