日時: 9月14日（木）1限目（9:20-10:50）

abstract: In many commercial enterprises, improving organizational performance is an important issue. Therefore, research results on organizational performance have been accumulated in business administration and organizational behavior theory; specifically, various organizational performance factors have been clarified. On the other hand, these research results need to be more practical to be used in actual company organizations. In this presentation, we introduce two studies as specific examples of such issues: Study 1 deals with the issue of not being able to continuously grasp Work Engagement (WE), which is the level of employees' enthusiasm for their work, and Study 2 deals with the issue that we cannot discover human who can activate their teams.
(Study 1) The SARS-CoV-2 or Covid-19 pandemic has transformed our lives. In order to aid in the prevention and spread of infection, a remote work style has rapidly proliferated. As this remote work style has proliferated, new work-related problems have come to light such as long working hours, mental health problems, and decreased communications. One issue that team members in the company face is that they cannot fully grasp the WE level (WEL) of subordinates such as in terms of absorption, dedication, and vigor due to limited in-person communications in a remote work setting. On the other hand, as a substitute for in-person communications, online communications via text-based chat tools such as Slack and Microsoft Teams have become popular. We propose a new approach that estimates the WEL of users on text-based chat tools by embeedding them in a feature space using graph neural networks (GNNs). Specifically, in this embedding process, we do not use the content of text communications but utilize only communications network architectural information, which expresses who is talking to whom and how often. In this study, we conduct two studies using Slack data to evaluate the proposal. The results of the first study reveal that the conversation content does not influence the WEL, but the properties where they are in the communications network do have an influence. Specifically, there is a typical tendency that the similarity of graph architectural feature vectors becomes low as the difference in WE becomes large than in the case of the similarity of the linguistic feature vectors. According to this result, in the second study, we develop a machine learning model that estimates the WEL using only the architectural features of the communications network in which a node represents a human and an edge represents a communications log, i.e., if person A talks to person B, the edge between node A and node B is stretched. The model estimates the WEL using true and predicted values at a correlation coefficient of 0.60. Since this model in the proposed approach uses only the communications network and does not use linguistic information, it is valuable in actual business situations.
(Study 2) Many managers and human resource departments transfer a person in an attempt to increase the performance of an organization or team. When we define performance as team efficiency, the performance is influenced by the density of team communications. However, whether or not the candidate transferee will actually increase the density of team communications is an unknown. In this paper, we propose a new approach that estimates whether or not a person who joins a team will improve the density of team communications based on an instant messaging system (IMS). In the proposed approach, we embed people in feature space using graph neural networks. In this embedding process, we do not use the content of text communications but utilize only communication graph architectural information that expresses who is talking to whom and how often. Additionally, the proposed approach does not require a questionnaire to indicate the density of team communication as in some previous studies. In the proposed approach, we develop a machine learning model classifying whether or not a transferee will improve the density of team communications. The model classifies transferees at an accuracy of 0.57 and precision of 0.58. Since this model does not use text contents from the IMS, it is valuable in actual business situations.

発表題目: Instant Messaging Systemを用いた組織パフォーマンス向上への実践的アプローチ
発表概要: 多くの営利企業において，組織のパフォーマンスを向上させる方法は重要な論点である．そのため，経営学や組織行動論の分野では組織パフォーマンスに関する研究成果が蓄積され，組織パフォーマンスの様々な因子が明らかになっている．一方で，これらの研究成果を実際に会社組織で利用するには，十分に実践的ではないという課題が生じる．本発表では，そのような課題の具体的なものとして，従業員の仕事への熱量であるWork Engagement（以下，WEと呼ぶ）を継続的に把握できない，という課題を扱った研究（研究1）と，チームを活性化させる人材を発見出来ない，という課題を扱った研究（研究2）を紹介する．
（研究1）リモートワークの拡大によって対面コミュニケーションが減少し，部下や同僚のWEを十分に把握できなくなったという問題が発生している．一方で，対面コミュニケーションの減少と同時に，SlackやMicrosoft Teams等のIMSを用いたコミュニケーションが活発になっている．本研究では，IMSを用いてWEレベル（以下，WEL）を推定するアプローチを提案する．提案するアプローチでは，グラフニューラルネットワーク（以下，GNN）を用いて特徴空間に人を埋め込むが，その過程においてIMSでの発言内容を用いずに，誰と誰がどの程度の頻度で話しているかという情報のみを利用するため，企業における秘匿情報の観点から実際のビジネスシーンで活用しやすい．本アプローチによって構築された機械学習モデルは，WELの真の値と予測値との相関係数0.60でWELを推定することが確認できた．
（研究2）企業において，チームのパフォーマンスを高めるための一つの手段として人事異動がある．パフォーマンスをチームの効率性と定義した場合，パフォーマンスはチーム内のコミュニケーションが活発になるほど高くなることが知られている．一方で，誰がどのチームのコミュニケーションを活性化させるかを事前には分からない．そこで本研究では，IMSのデータを用いて，チームに新しく参加する人がコミュニケーションを活性化させるかを算定するアプローチを提案する．本アプローチにおいて，コミュニケーションを活性化させるかを判定する機械学習モデルを正解率0.57，適合率0.58で構築した．さらに，本アプローチでは，テキストコミュニケーションの内容を利用しないため，プライバシーや秘匿情報の観点から実際のビジネスシーンで活用しやすいという特性がある．