Hiroaki Ohshima

In this research, we propose a method for estimating the degree of injury from text documents that describe accidents. It is assumed that a text document to be input consists of a few sentences. The proposed method is to estimate the degree of injury by solving a classification problem using machine learning techniques. The data used in this research is the accident data published in the Accident Information Data Bank System. The text in the “Summary of the accident” field is used as an input. In the proposed method, an input text is represented as a distributed representation using the generic language model called BERT. As a model for BERT, we use a pre-trained model trained using the Japanese Wikipedia. To improve the performance of the task of estimating the degree of injury, we introduce the following four ideas; (1) the class weights, (2) the ordinal classification, (3) the multitasking learning, and (4) the fine-tuning model with token label estimation. We examined the effects of using and not using these ideas on the accuracy, Macro F1, RMSE, and confusion matrices for the task of estimating the degree of injury. The results showed that Macro F1 and RMSE are improved when (1) the class weights and (2) the ordinal classification are introduced. In addition, the accuracy is improved when (3) the multitasking learning is introduced.

本研究では，提示型検索モデル（Ostensive Search Model）に基づくインタフェースによって鑑賞者個人の興味を反映したミュージアム体験を可能にする電子ガイドを提案し，そのログを分析することで実現可能になった事前学習，事後学習支援システムについても提案する．我々は，国立民族学博物館（みんぱく）の展示物のうち3,053点について，展示物の解説やビデオなどの詳細情報を検索し，メモなどのアノテーションを付与できるiPad用アプリケーションである「みんぱくガイド」を作成した．みんぱくガイドは，鑑賞者が新しい展示物に気付いたり，興味を明確化することができるように，一覧性の高い検索結果画面を中心に情報探索を繰り返せるインタフェースを持っている．このような電子ガイドの操作履歴や位置情報などのログを用いることで，個人のミュージアム体験を色濃く反映した事前・事後学習支援を可能にした．事前学習支援システムでは，ミュージアムに行く前にカードを整理しながら鑑賞計画を立てるウェブアプリケーションの利用を通じて，鑑賞者に自分が何を学びに行くかという鑑賞軸を自覚してもらう．また，事後学習を促す仕組みとして，ログから鑑賞者が興味を持った展示物を推定し，後から鑑賞体験を思い出しやすくするためのパーソナライズされたポストカードを自動生成するシステムも作成した． This paper proposes an electronic guide application that enables visitors to get a museum experience that reflects their individual interests through an interface based on an Ostensive Search Model. In addition, we propose a pre-learning and post-learning support system that can be connected to our electronic guide. The system is based on a search result screen with a high level of browsability. We propose a system that allows users to search for detailed information such as explanatory texts and videos of exhibits they are interested in and add annotations such as handwritten notes. We created the “Minpaku Guide,” an iPad application that allows users to search for detailed information such as explanatory text and videos on exhibits of interest and add annotations such as notes. We also developed pre-learning support systems connected to our guide application. As a pre-learning support system, we created a web application that allows users to manually organize cards before visiting the museum to clarify what the visitors will learn beforehand. As a post-learning support system, we implemented a system that summarizes the visitor's operation log of the guide application into a postcard that helps to recall the viewing experience later.

This study proposes a method for retrieving and ranking posts from social network services(SNSs) by specifying and providing feedback on the context of posts. Current search systems for SNS posts cannot handle user intent with regard to the context of posts to be retrieved, mainly owing to the incompleteness of SNS posts, i.e., they do not contain the users' contexts (e.g., situations or preferences) of users posting messages. Hence, we propose a search method that accepts two kinds of queries, namely, content queries and context queries, and that updates these queries based on the user feedback with special attention to the contexts of posts. Our search method considers the whole SNS dataset as a graph and the nodes surrounding each post as its context to find relevant posts in terms of content and context, our method propagates user feedback via this graph. Our experimental results based on a Twitter test collection revealed that our proposed method showed improved retrieval performance as compared with conventional SNS retrieval and relevance feedback. In addition, we could detect the optimal parameters for feedback propagating.

In this paper, we propose a method of ranking entities (e.g. products) based on pairwise preferences learned and inferred from user reviews. Our proposed method finds expressions from user reviews that indicate pairwise preferences of entities in terms of a certain attribute, and learns a function that determines the relative degree of the attribute to rank entities. Since there are a limited number of such expressions in reviews, we further propose a method of inferring pairwise preferences based on attribute dependencies obtained from reviews. As some pairwise preferences are less confident, we also propose a modified version of a learning to rank method, Fuzzy Ranking SVM, which can take into account the uncertainty of pairwise preferences. The experiment was carried out with three categories of products and several attributes specific to each category. The experimental results showed that our approach could learn more accurate pairwise preferences than baseline methods, and inference based on the attribute dependency could improve the performances.