2021-01-012024-05-17https://scholars.lib.ntu.edu.tw/handle/123456789/670846"在中文的知識庫發展上，自從中研院資訊所詞庫小組在 1990 年代開啟了中文詞庫工作以來，再從 Wordnet 和 Hownet 兩個詞彙語意脈絡下分別持續進行中文詞彙 (Chinese Wordnet) 與廣義知網 (e-Hownet) 的工作，打下了現代漢語紮實的詞彙語意基礎。然而在晚近也延伸了可能的問題：(1). 理論深度部分：較欠缺與新的語言／語意理論發現之間指引與討論，造成語意資源與標記的淺層與表面化，對於時間與事件表徵、語意組合與語用推理任務無法提供有力的知識資源。(2). 沒有長期挹注的經費能夠投入在專業標記的人力資源訓練， 造成無法永續經營與 。(3). 受制於相關著作法規不明而難以完全開源，並且對於跨語言資源整合欠缺視野與敏感度，也因此難以與國際研究社群對話與互動，形塑出創新領導地位。 本總計畫的目的，即在於以語意概念為本，開發出能夠蘊含語意組合推理內涵的中文知識圖譜。這樣的知識圖譜，旨在能夠對於「概念－事實－關係」統合與具備一致性的語意表徵，使得機器得以在符碼 (symbolic) 層次上即可完成概念運算，與知識檢索與發掘的相關應用。本子計劃立基於過去幾年的基礎，特別著重在中文詞彙化的語意概念 (lexicalized concept) 、以中文語言概念生成的上層概念知識本體 (upper-level ontology)，與文字蘊藏的常識 (world knowledge) 為本的資源建構，可以與知識圖譜較多在事實關係的取向做整合的互補。" "The solid foundation of lexical meaning in modern Chinese has been built since the CKIP had started Chinese Thesaurus work in 1990s and continued in 2 kinds of lexical meaning context, Wordnet and Hownet, which turn into Chinese Wordnet and e-Hownet. However, it extended to some possible problems during the past few years: (1). Theoretical depth: the lack of guidances and discussions between new language/semantics theories and findings, resulting in the shallow and superficial semantic resources and annotations, which the time and event reprecentation, semantic combination and pragmatic reasoning tasks cannot provide powerful knowledge resources. (2). There is no long-term funds invested in professionally annotation personnel training, resulting in the resources are unsustainable operations and not able to open for use. (3). It is difficult to be fully open source because of the unclear laws and regulations, and it is also difficult to have a dialogue and interact with the international research community to shape an innovative leadership because of the lack of vision and sensitivity in cross-language resource integration. The purpose of the main project is based on the semantic concepts, to develop a Chinese Knowledge Map contained semantic meaning and semantics combination. This Chinese Knowledge Map aims to integrate concept-fact-relation with consistent semantics representation to enable machines to complete the concept operation at the symbolic level and the related applications of knowledge retrieval and exploration. This sub-project is base on researches in the past few years, focus on the source construction with lexicalized concept in Chinese, the upper-level ontology generated by language concept in Chinese, and the world knowledge contained in characters, to integrate and complement with Knowledge Map on the fact-relation aspect."語意的知識表徵中文詞彙化的概念詞意的粒度知識本體semantic knowledge representationlexicalized conceptsemantic granularityontologies