人工知能学会論文誌
検索
OR
閲覧
検索
20 巻 , 2 号
選択された号の論文の5件中1~5を表示しています
    • |<
    • <
    • 1
    • >
    • >|
一般論文
論文
  • 伊藤 孝行, 横尾 真, 松原 繁夫
    20 巻 (2005) 2 号 p. 84-93
    公開日: 2005/01/06
    ジャーナル フリー
    Auctions have become an integral part of electronic commerce and a promising field for applying multi-agent technologies. Correctly judging the quality of auctioned goods is often difficult for amateurs, in particular, in Internet auctions. However, experts can correctly judge the quality of goods. In this situation, it is difficult to make experts tell the truth and attain an efficient allocation, since experts have a clear advantage over amateurs and they would not reveal their valuable information without some reward. In our previous work, we have succeeded in developing such auction protocols under the following two cases: (1) the case of a single-unit auction among experts and amateurs, and (2) the case of a combinatorial auction among single-skilled experts and amateurs. In this paper, we focus on versatile experts. Versatile experts have an interest in, and expert knowledge on the qualities of several goods. In the case of versatile experts, there would be several problems, e.g., free riding problems, if we simply extended the previous VCG-style auction protocol. Thus, in this paper, we employ PORF (price-oriented, rationing-free) protocol for designing our new protocol to realize a strategy-proof auction protocol for experts. In the protocol, the dominant strategy for experts is truth-telling. Also, for amateurs, truth-telling is the best response when two or more experts select the dominant strategy. Furthermore, the protocol is false-name-proof.
    抄録全体を表示
  • 田中 貴紘, 安村 禎明, 片上 大輔, 新田 克己
    20 巻 (2005) 2 号 p. 94-104
    公開日: 2005/01/20
    ジャーナル フリー
    This paper describes an estimation of the support function of an online mediation support system for education, Similar Disputation Scene Search. This system refers to the old mediation cases. This function searches the scene that similar to the current disputation from the old mediation cases by similarity of the topic transition. The system recommends candidate list of reply to the user from the similar scene. This system consists of a mediation server, disputation interface, and case database. At first, a user connects the mediation server via a computer network, and enters a mediation room on this server. The user argues by exchanging arguments with facial expression of an animated agent each other using the disputation interface. The arguments are recorded as an XML document. We estimated the Similar Disputation Scene Search. We calculated recall and precision of the function and examined whether recommended reply is available or not. From this analysis, we discussed the availability of the mediation support system based on CBR.
    抄録全体を表示
  • 小山 聡, マニング クリストファー D.
    20 巻 (2005) 2 号 p. 105-116
    公開日: 2005/02/02
    ジャーナル フリー
    We propose a kernel method for using combinations of features across example pairs in learning pairwise classifiers. Pairwise classifiers, which identify whether two examples belong to the same class or not, are important components in duplicate detection, entity matching, and other clustering applications. Existing methods for learning pairwise classifiers from labeled training data are based on string edit distance or common features between two examples. However, if two examples from the same class have few common features, these methods have difficulties in finding these pairs and achieving high recall. One typical example is to check whether two abbreviated author names in different citations refer to the same person or not. Since similarities between examples from the same class become close to zero, classifiers fail to distinguish positive pairs from negative pairs. One approach to avoiding the problem of zero similarities is using conjunctions of different features across examples, but implementing this idea straightforwardly makes the computational cost prohibitive for practical problems. Using a kernel on pair instances, our method can use feature conjunctions across examples without actually doing feature mappings, which are computationally expensive. The kernel is a tensor product of two inner products on the original feature space. The corresponding feature mapping generates conjunctions of features only across the two different examples while that of the conventional polynomial kernel also generates conjunctions of features from the same example, which are irrelevant to pairwise classification and cause deterioration of accuracy. Our experiments on the author matching problem show that this method can give a precision 4 to 8 times higher than that of previous methods at medium recall levels.
    抄録全体を表示
論文特集:近未来チャレンジ
論文
    • |<
    • <
    • 1
    • >
    • >|
feedback
Top