人工知能学会論文誌
Online ISSN : 1346-8030
Print ISSN : 1346-0714
検索
OR
閲覧
検索
21 巻 , 5 号
選択された号の論文の3件中1~3を表示しています
    • |<
    • <
    • 1
    • >
    • >|
論文
  • 赤石 美奈
    21 巻 (2006) 5 号 p. 428-438
    公開日: 2006/06/27
    ジャーナル フリー
    This paper proposes a framework to access information based on a narrative structure of documents. This framework consists of two processes. The one is to decompose existing documents into smaller units. The other process is combining unit components into a new story taking on a new meaning based on a context. In this paper, a narrative structure for documents is modeled as follows. A story corresponding to a document is regarded as a sequence of scenes. A scene is a chunk of sentences. A sentence is mapped into a set of terms in the sentence. Decomposition process gives two mechanisms to decompose a story into scenes. Composition process shows four patterns to connect scenes. Both techniques to decompose/compose a story are based on the notions of term dependency and term attractiveness. This paper also showes visualization tools to express the narrative structure for documents. Word Colony overviews content of a story as a directed graph representing the relation among term dependency. Topic Sequence is also directed graph to show the sequence of scenes along a story plot. The basis of these visualization techniques is the notions of term dependency and term attractiveness. They show the variety of understandings of the same documents.
    抄録全体を表示
  • 光永 悠紀, 鷲尾 隆, 元田 浩
    21 巻 (2006) 5 号 p. 439-449
    公開日: 2006/06/27
    ジャーナル フリー
    A novel approach to subspace clustering is proposed to exhaustively and efficiently mine quantitative frequent itemsets (QFIs) from massive transaction data for quantitative association rule mining. The numeric part of a QFI is an axis-parallel and hyper-rectangular cluster of transactions in an attribute subspace formed by numeric items. For the computational tractability, our approach introduces adaptive density-based and Apriori-like subspace clustering. Its outstanding performance is demonstrated through the comparison with the past subspace clustering approaches and the application to practical and massive data.
    抄録全体を表示
  • 金盛 克俊, 延澤 志保, 太原 育夫
    21 巻 (2006) 5 号 p. 450-458
    公開日: 2006/07/05
    ジャーナル フリー
    Discovery learning, which acquires new concepts or knowledge, is one of the most advanced forms of machine learning. Few systems have been proposed for discovery learning in practical use, and most of them are based on various heuristics. Discovery learning is considered to consist of two processes: inductive acquisition of general structure(relational structure) from existing knowledge base, and application of the relational structure to a domain knowledge for acquiring new concepts in the domain. In this paper we mainly focus on the application process, and propose a method of generating new concepts, that is, new predicates which do not occur in the domain knowledge, by applying the relational structure. We prove that the new generated clauses including the new predicates are consistent with the domain knowledge, and propose an algorithm for approximate calculating the new clauses from the relational structure and the domain owledge in finite steps. We give proof of some useful theorems for this algorithm. In addition, we discuss the the method for the acquisition of relational structure from an existing knowledge base.
    抄録全体を表示
    • |<
    • <
    • 1
    • >
    • >|
feedback
Top