人工知能学会第二種研究会資料
Online ISSN : 2436-5556
2021 巻, BI-018 号
第18回 ビジネス・インフォマテックス研究会
選択された号の論文の16件中1~16を表示しています
  • 小柳 佑介, 上村 健人, 浅井 達哉, 金児 純司, 大堀 耕太郎
    原稿種別: 研究会資料
    2021 年 2021 巻 BI-018 号 p. 01-
    発行日: 2021/03/05
    公開日: 2022/01/17
    研究報告書・技術報告書 フリー

    In this paper, we propose a framework for discovering characteristic causal relation-ships for each person or thing, such as each individual customer in marketing and each individualdefective product in manufacturing. Our method enumerates all correlations from a given set ofsamples and discovers characteristic causalities by applying a causal discovery method to all sub-sets of samples identified by conditions derived from the correlations. After that, we can obtain acharacteristic causality for each new sample by identifying a condition that the new sample satis-fies. We also report experimental results on real marketing data to show the effectiveness of ourmethod.

  • 二宮 一将, 倉橋 節也
    原稿種別: 研究会資料
    2021 年 2021 巻 BI-018 号 p. 02-
    発行日: 2021/03/05
    公開日: 2022/01/17
    研究報告書・技術報告書 フリー

    In recent years, Japanese organizations have faced issues such as individual isolation andorganizational division as the diversity of employment patterns, age groups, genders, nationalities, etc.progresses. Since the working population is expected to decrease, it is necessary to realize innovation anddigital transformation with the aim of improving the productivity of each person, but there are manyclaims that the division of the organization is hindering it. .. In this research, we focus on humanresources who bridge various organizations, and by clarifying new working styles to improve theperformance of organizations by social network analysis, conventional uniform evaluation of humanresources and human resources The purpose is to obtain suggestions for reviewing the evaluation index oftraining. For verification, use the transmission / reception log of the internal currency that evaluatesaltruistic behavior.

  • 西野 貴志, 章 進, 山成 侑香, 宮尾 勝, 劉 鵬達, 林 久志
    原稿種別: 研究会資料
    2021 年 2021 巻 BI-018 号 p. 03-
    発行日: 2021/03/05
    公開日: 2022/01/17
    研究報告書・技術報告書 フリー

    In online food delivery, which is expanding all over the world, the increasing number of traffic accidentsduring delivery is becoming problematic. To ensure the safety of delivery workers and the residents, it isnecessary to understand the incentives of workers for behavior choices. While many existing sharingplatforms are pulling workers into the online labor process by such incentives as on-peak/off-peaksurcharges, workers try to accomplish more than their goals within a limited time. In this paper, to preventover-speeding during delivery, we compare and evaluate some task assignment methods and incentiveschemes by multi-agent simulation. We model worker's rational choices of behaviors based onreinforcement learning considering the profit and over-speeding of delivery workers.

  • 菊地 剛正, 高橋 大志
    原稿種別: 研究会資料
    2021 年 2021 巻 BI-018 号 p. 04-
    発行日: 2021/03/05
    公開日: 2022/01/17
    研究報告書・技術報告書 フリー

    商品設計やサービス提供の領域では,顧客に関する具体的な議論を喚起し,関係者の合意形成を円滑にするため,顧客の仮想人格である「ペルソナ」を作成することが広く行われている.しかし,作成過程において,a)考慮すべき顧客の要素・特性や,b)ペルソナの正当性の担保や優先順位付けに関して,必ずしも明確な指針が存在するわけではない.そこで本研究では,社会シミュレーションのログ分析手法を援用し,ペルソナ作成過程における分析支援手法を検討する.想定される効用は以下の通り:社会シミュレーションの各種分析手法を援用することで,1)想定する結果に影響を与える要素・特性を明確化しうること,2)注目すべきペルソナを同定しうること.

  • 新井 崇夫, 倉橋 節也
    原稿種別: 研究会資料
    2021 年 2021 巻 BI-018 号 p. 05-
    発行日: 2021/03/05
    公開日: 2022/01/17
    研究報告書・技術報告書 フリー

    In this paper, we present our concept on the analysis of the coordination of crowd-workers by multi-agent systems. The results inferred from the current analysis and the future plan of the analysis are presented.

  • 平山 真康, 杉之内 将大, 水山 元
    原稿種別: 研究会資料
    2021 年 2021 巻 BI-018 号 p. 06-
    発行日: 2021/03/05
    公開日: 2022/01/17
    研究報告書・技術報告書 フリー

    The spread of COVID-19 has changed people's lifestyle, which caused great loss in variousindustries including the flower industry. Under the circumstance, new online shopping services forindividuals are attracting attention. However, it is not easy for those who have not purchased flowers oftenbefore to find "a right flower for them." Therefore, this study (1) defines some features of flower imagesand creates a hyperspace representing the similarity among flower images, and (2) proposes and evaluatesan interactiveevolutionary computation system for recommending a flower that reflects theuser's potentialtaste from a large number of candidates according only to a relatively small number of intuitive flowerevaluation madeby a user.

  • 加藤 大望, 矢田 昇平, 倉橋 節也
    原稿種別: 研究会資料
    2021 年 2021 巻 BI-018 号 p. 07-
    発行日: 2021/03/05
    公開日: 2022/01/17
    研究報告書・技術報告書 フリー

    近年、生産性の向上等を目的とし、工場の Internet of things (IoT)化、スマート化が検討されている。このような工場では、加工中の製品が一方向に流れていく、従来型のベルトコンベア等を用いた生産システムとは異なり、Automated guided vehicles (AGV)を用いて複数工程を行き来しながら生産を行うフレキシブルなジョブショップ型の生産方式となる。本生産方式の課題は、タスクの適切なスケジューリングであり、最適解を保証することが難しい Non-deterministicPolynomial time-hard (NP-hard)となることにある。加えて、本論文がモチーフとする半導体製造工程では、微細な加工を行うことから大気中のごみによる汚染や搬送時の振動が製造歩留まりに大きな影響を与えるため、工場内の AGV は決められた搬送路上を通行し、搬送車を追い越すことができない制約条件が存在する。この制約により、AGV の空間干渉が生じるため、搬送路上の AGV は渋滞が生じ、その結果、適切な時機での製造装置に加工製品の受渡しができず、納期遅れが発生する等の課題が生じる。このような課題に対し、本論文では、Multi-agent system (MAS)を用いることで、搬送車の空間干渉モデルを考慮した交通流量に関する解析モデルの検討を行った。MAS を用いることで、搬送車の空間干渉をモデリングすることが可能となる。加えて、主問題を分割し、副問題として処理を行う Contract Net Protocol (CNP)を解析モデルに適用し、加工装置をタスク管理者、AGV は契約者として扱い、動的に契約交渉を行うとした。CNP を用いることで、AGV の割当てエージェントとタスク管理者エージェントで協調し、さらに、工場内におけるドメイン知識である渋滞の時系列情報を利用することで、AGV の渋滞緩和が可能か検討を行った。検討の結果、搬送車の空間干渉モデルのみを用いた計算では、単位時間あたり n =0.4台/s で流入した場合に渋滞が生じていたが、CNP を用いること、および AGV の車間距離 Lh を適切に調節することで、渋滞を緩和させられることを見出した。

  • 野矢 淳, 高橋 大志
    原稿種別: 研究会資料
    2021 年 2021 巻 BI-018 号 p. 08-
    発行日: 2021/03/05
    公開日: 2022/01/17
    研究報告書・技術報告書 フリー

    情報技術の発達に伴い,株式市場分析においても情報の重要性が注目されてきている.本研究は,ニュース記事に着目し,情報技術を用いて投資家が判断や意思決定を支援するモデルの構築を目指す.その方法として,TOPIX Core30 の構成銘柄のニュース記事の要約文章を生成し,分散表現に変換して機械学習で分析を行った.

  • 浦 杏弥, 後藤 裕介
    原稿種別: 研究会資料
    2021 年 2021 巻 BI-018 号 p. 09-
    発行日: 2021/03/05
    公開日: 2022/01/17
    研究報告書・技術報告書 フリー

    In this study, we tried to predict the brand switching behavior (variety-seeking behavior) forstimulating customer's purchases at the store in real-time. We used customer purchase data from Japaneseretail stores for evaluating our proposed method. Our proposed method includes some additionalexplanatory variables from existing studies, which considers purchasing pattern in a time series. The resultsshowed that our proposed method increased the prediction accuracy of brand switching behavior. Ourproposed method includes some additional explanatory variables from existing studies, which considerspurchasing pattern in a time series. The results showed that our proposed method increased the predictionaccuracy of brand switching behavior. Furthermore, our proposed method can be used for various storesand products because experimental results showed that our proposed method's effectiveness is independentof stores and products.

  • 中村 淳海, 高橋 大志
    原稿種別: 研究会資料
    2021 年 2021 巻 BI-018 号 p. 10-
    発行日: 2021/03/05
    公開日: 2022/01/17
    研究報告書・技術報告書 フリー

    This study examines the effectiveness of pricing strategies in business hotels. Through theanalysis, we will attempt to identify the impact on the lease period and sales of business hotels, and how itchanges with the introduction of policy simulation. In this study, pricing strategy is considered as a policy.Through Agent-Based Simulation, we will match borrower and lender agents and analyze the results ofthe balance between supply and demand, and we will consider pricing strategy in our analysis. We will bemeasured the policy effects from individual strategies, using the framework created by previous studiesthat considered prices. A bottom-up policy evaluation will be conducted, and exploratory studies will beconducted to determine that it is an effective tool.

  • 矢田 昇平, 倉橋 節也
    原稿種別: 研究会資料
    2021 年 2021 巻 BI-018 号 p. 11-
    発行日: 2021/03/05
    公開日: 2022/01/17
    研究報告書・技術報告書 フリー

    In this research, for companies that have introduced the free address system, we extract featuresof various work styles from employee location data and explore how they relate to performance. By doingso, we considered models that increase freedom of work environment and employee performance. Themethod adopted to model the relationship between work styles and performance is a least-squaresregression model using satisfaction data synthesized by sparse principal component analysis. From theresults, we proved the initial hypothesis that the performance of employees who have face-to-faceconnections with many employees regardless of their departments is high in a quick meeting area. Besides,as a new finding, it is the communication that occurs in a quick meeting area, not the planned meeting heldin the meeting room, that affects employees performance. In the relationship of employees who have similarwork contents, it was found that the performance of employees who bridge between different groups ishigher than that of employees who have a lot of face-to-face communication among employees.

  • 刘 梦垚, 上瀧 弘晃, 高橋 大志
    原稿種別: 研究会資料
    2021 年 2021 巻 BI-018 号 p. 12-
    発行日: 2021/03/05
    公開日: 2022/01/17
    研究報告書・技術報告書 フリー

    Following the COVID-19 outbreak, unlike the plunge in traditional mainstream assets, crypto assets haveperformed very well. Indices that reflect changes in the crypto assets market have also grown in recent years. However,in terms of cryptocurrency index-related studies, there are not many, and the period analyzed is mostly three years. Inthis study, we analyze the impact of crypto assets on portfolio construction and attempt to clarify the risk-returncharacteristics of portfolios that include crypto assets as investment targets. We analyze historical data from July 2014to April 2020 based on the cryptocurrency index CRIX and six other traditional mainstream assets. By using theDCC-GARCH model, this study finds out the low dynamic correlation between the crypto assets and traditional ones.By using the mean-variance model, Cornish-fisher expansion, and T-copula CVaR approach to check the frontier lineand portfolio performance, we find out that crypto assets have the potential to improve the risk-return characteristics oftraditional portfolios.

  • 宮森 望, 堀田 大貴, 上田 賀一
    原稿種別: 研究会資料
    2021 年 2021 巻 BI-018 号 p. 13-
    発行日: 2021/03/05
    公開日: 2022/01/17
    研究報告書・技術報告書 フリー

    Companies are now able to gather a lot of business process's execution data (calledevent log) by managing them with information systems. In addition, they seek to analyze theirprocess by leveraging event log to improve their processes (e.g. reducing waiting time and costs).However, there are many cases where process management systems record work suspending, butdo not record specific reason for that suspending. Hence, we propose the extracting representativeworkers' activity patterns method for understanding the activity of workers during work suspend-ing. The method takes into account the order of work with Damerau ? Levenshtein distance.Analysts are able to understand the cause of the work stoppage by comparing the extracted pat-terns with the process flow. We could find the workers' activity patterns which is along work flowof the process in the empirical evaluation.

  • 許 慏碩, 村上 英治, 高橋 大志
    原稿種別: 研究会資料
    2021 年 2021 巻 BI-018 号 p. 14-
    発行日: 2021/03/05
    公開日: 2022/01/17
    研究報告書・技術報告書 フリー

    Gas and electric power are one of the essential social infrastructure for social and economicactivities. The purpose of this research is to construct a gas' consumption model suitable for thedistribution and sales market of LP gas in Japan by utilizing temperature. As a method of predictingdemand, we would like to first create a model of power consumption and then apply it to LP gas. Analysisof customers' gas consumption can infer from what standpoint they were consuming public goods.Through this, we would like to offer users' reasonable price and usage to expand their choices andcontribute to the efficient use of social resources. Furthermore, analysis of LP gas usage through deeplearning is possible.

  • 工藤 大輝, 後藤 裕介
    原稿種別: 研究会資料
    2021 年 2021 巻 BI-018 号 p. 15-
    発行日: 2021/03/05
    公開日: 2022/01/17
    研究報告書・技術報告書 フリー

    In general, baking companies develop a large number of new products, but only some of them becomeestablished as standard products, and the production of new products may result in a large amount of wasteloss due to unpredictable sales. Therefore, it is necessary to predict the future order volume based on theshort-term order data of new products. In this paper, we propose a method for forecasting and evaluationusing a combination of multiple explanatory variables and models, using shipment data from bakeries fora short period of time, such as one month after the start of sales. In the proposed method, the most accurateresults were obtained by constructing a model with variables that took into account the characteristics ofrandom forests and bread making, and explanatory variables that were modified for the low number ofstores.

  • 堀井 博夫, 小畑 崇弘, 瀬之口 潤輔, 倉橋 節也
    原稿種別: 研究会資料
    2021 年 2021 巻 BI-018 号 p. 16-
    発行日: 2021/03/05
    公開日: 2022/01/17
    研究報告書・技術報告書 フリー

    Since the Japanese electricity retail market's liberalization started in 2016, the numbers ofpower generation companies and retailers joining the market are increasing. Power generation companiesand retailers must pay imbalance charges, which depend on the amount of error between planned supplyand demand volume and real supply and demand volume. Imbalance unit prices are calculated from spotprices and day-ahead market prices in Japan and volatile because imbalance unit prices are influenced bymany parameters like the imbalance between supply and demand, climate changes, etc. Forecastingimbalance unit prices are essential for risk management. Analyzing and forecasting imbalance unit pricesusing generalized additive models (GAM) and future work to improve the price forecasting estimation areshown in this study.

feedback
Top