Supporting Time-Constrained Student Sports Journalists: Smartwatch Flagging and Match Visualization for Better Interview Questions

Supporting Time-Constrained Student Sports Journalists: Smartwatch Flagging and Match Visualization for Better Interview Questions Ai Hagihara, Satoshi Nakamura Meiji University

Contribution 1. Understanding difficulties of student sports journalists 2. Match review method using subjective flags − Helps journalists reflect on the match and create better 3. Smartwatch-based flagging system − Allows journalists to flag while taking photos 1

Background Survey Good-Q Proposed Method Flagging Question Experiment Experiment Recent Progress “How was your condition today?” “What was the reason for today’s victory?” Vague questions are difficult for athletes to answer − Athletes are happy when asked questions that show careful observation of their play. − Ideally, interview questions should dig deeper into the match. 2

Background Survey Good-Q Proposed Method Flagging Question Experiment Experiment Recent Progress Why it is difficult to formulate specific questions Multitasking • • • • • Taking notes Planning the article storyline Preparing questions Taking photos Posting on social media Time constraints • Very little time before the interview starts 3

Background Survey Good-Q Proposed Method Flagging Question Experiment Experiment Future Work Why it is difficult to formulate in-depth match questions Multitasking • taking notes • planning the article storyline Because of multitasking and time constraints, student sports journalists have little capacity • preparing questions think of ideal questions. • taking to photos • posting on social media Time constraints • very little time before the interview starts 4

Background Survey Good-Q Proposed Method Question Experiment Flagging Experiment Future Work Interview Question Survey Distribution of Question Evaluations Player High Score Player Low Score Journalist High Score 48 9 Journalist Low Score 28 24 Both High: Focused on tactics or specific plays Referred to past matches or comparisons Both Low: Vague, abstract, or incomplete questions 5

Background Survey Good-Q Proposed Method Flagging Question Experiment Experiment Recent Progress Our Goal To support student journalists in creating tactical and play-specific questions that are highly rated by both athletes and journalists We propose a system that helps them understand the match effectively 6

Background Survey Good-Q Proposed Method Flagging Question Experiment Experiment Recent Progress Fencing 7

Background Survey Good-Q Prototype System Proposed Method Flagging Question Experiment Experiment Recent Progress Left Player's Position Right Player's Position Visualize player’s positions on the piste 8

Background Survey Good-Q Proposed Method Flagging Question Experiment Experiment Recent Progress Prototype System Flagged scene 9

Background Survey Good-Q Proposed Method Flagging Question Experiment Experiment Recent Progress We investigated two points: Whether reviewing the match using videos and play data helps journalists come up with more specific questions. Whether journalists can actually add flags while performing their usual reporting tasks. 10

Background Survey Good-Q Proposed Method Question Experiment Flagging Experiment Recent Progress We investigated two points: Whether reviewing the match using videos and play data helps journalists come up with more specific questions. Whether journalists can actually add flags while performing their usual reporting tasks. 11

Background Survey Good-Q Proposed Method Question Experiment Flagging Experiment Recent Progress Creating Questions User Study Overview Purpose Participants To examine how flag-based review affects the content of interview questions Two student journalists Setting Watch a match video and create interview questions Condition Notes / Video / Video& play data(with flag) Time Limit Two minutes 12

Background Survey Good-Q Proposed Method Question Experiment Flagging Experiment Recent Progress Analysis Method ◆ A total of 98 questions were analyzed ◆ Each question was coded for five elements: − Position − Time − Scoring Method − Score − Win/Loss 13

Background Survey Good-Q Proposed Method Question Experiment Flagging Experiment Recent Progress Analysis Method ◆ A total of 98 questions were analyzed ◆ Each question was coded for five elements: − Position − Time − Scoring Method − Score − Win/Loss 14

Background Proposed Method Survey Good-Q Question Experiment Flagging Experiment Recent Progress Score Win/Loss Results – Elements Included in Questions Method Position Time Method Notes Only 0.04 0.02 0.37 0.27 0.27 Video Only 0.25 0.07 0.24 0.35 0.35 Video&Graph 0.34 0.04 0.38 0.23 0.17 Key Findings: − Proposed method produced more questions referring to player position − Fewer questions focused on win/loss outcomes 15

Background Proposed Method Survey Good-Q Question Experiment Flagging Experiment Recent Progress Results – Comparison of questions referring to “Method” Method Position Time Method Score Win/Loss Notes Only 0.04 0.02 0.37 0.27 0.27 Video Only 0.25 0.07 0.24 0.35 0.35 Video&Graph 0.34 0.04 0.38 0.23 0.17 Notes Only: Questions mainly confirmed what happened. Video & Graph: Questions explored the player’s “It looked like you were hit by a counter in the last point. thoughts. Was that right?” The proposed method may help create deeper questions. 16

Background Proposed Method Survey Good-Q Question Experiment Recent Progress Flagging Experiment Results – Comparison of questions referring to “Method” Method Position Time Method Score Win/Loss Notes Only 0.04 0.02 0.37 0.27 0.27 Video Only 0.25 0.07 0.24 0.35 0.35 Video&Graph 0.34 0.04 0.38 0.23 0.17 Notes Only: Questions mainly confirmed what happened. Video & Graph: Questions explored the player’s thoughts. The proposed method may help create deeper “In the last two points, you moved forward and aimed at questions. opponent’s feet. What were you focusing on?” 17

Background Proposed Method Survey Good-Q Question Experiment Recent Progress Flagging Experiment Results – Comparison of questions referring to “Method” Method Position Time Method Score Win/Loss Notes Only 0.04 0.02 0.37 0.27 0.27 Video Only 0.25 0.07 0.24 0.35 0.35 Video&Graph 0.34 0.04 0.38 0.23 0.17 Notes Only: Questions mainly confirmed what happened. Video & Graph: Questions explored the player’s thoughts. The proposed method may help create deeper questions. 18

Background Survey Good-Q Proposed Method Flagging Question Experiment Experiment Recent Progress We investigated two points: Whether reviewing the match using videos and play data helps journalists come up with more in-depth questions. Whether journalists can actually add flags while performing their usual reporting tasks. 19

Background Survey Good-Q Proposed Method Flagging Question Experiment Experiment Recent Progress Smartwatch-based Flagging System ◆ Single tap operation → Record timestamp as a flag ◆ Haptic + sound feedback 20

Background Proposed Method Survey Good-Q Flagging Question Experiment Experiment Recent Progress Field Study Overview Purpose Target Sport Participants Test feasibility of smartwatch flagging in real match enviroment Fencing 2 student journalists (3 years of fencing coverage experience) 21

Background Survey Good-Q Proposed Method Flagging Question Experiment Experiment Recent Progress Results of Field Study − Journalists can add flags while doing other tasks such as taking photo. − Our proposed method was feasible in real match environment. 22

Background Proposed Method Survey Good-Q Flagging Question Experiment Experiment Recent Progress Relationship Between Score Difference and Number of Flags Number of Flags 10 8 6 4 2 0 0 2 4 6 8 10 12 14 16 18 20 22 24 Point Difference (Points) A strong negative correlation was found between score difference and number of flags (r = −0.88) 23

Background Survey Good-Q Proposed Method Flagging Question Experiment Experiment Recent Progress Limitation Small number of participants − The number of participants was small − Including beginner journalists Analysis time − Tracking the players took a long time − Real-time analysis was not possible yet − Hard to use in actual interviews 24

Background Survey Good-Q Proposed Method Flagging Question Experiment Experiment Recent Progress Recent Progress (Additional result not written in the paper) Smartphone Application − Real-time analysis achieved by detection only (no tracking) − Conducted an actual interviews field test 25

Background Survey Good-Q Proposed Method Flagging Question Experiment Experiment Recent Progress Journalists’ Comment “By using the system, I was able to ask questions that focused more on the match than usual, and that felt really good. Also, during regular interviews, player A usually look serious, but when I said, “You didn’t move forward and backward much in that match, did you?” player A’s expression suddenly brightened and he said, “Yes, exactly!” He talked much more than usual, and I was really glad that I could ask that kind of question it made me happy, too”. 26

Background Survey Good-Q Proposed Method Flagging Question Experiment Experiment Recent Progress Journalists’ Comment Focus on the match − More focused on the match than usual − To flag important moments, they needed to watch the match carefully. It changed not only how to make questions but also how to watch the match. 27

Background Survey Good-Q Proposed Method Flagging Question Experiment Experiment Recent Progress Athletes’ Comment “When he asked about the scene where I scored consecutive points at the end, I was especially glad because I also felt that I was able to bring the flow of the match to my side at that moment”. “She mentioned my playing tendencies that I hadn’t noticed myself, and that gave me new insights into my own performance. I thought that was a really good question”. 28

Background Survey Good-Q Proposed Method Flagging Question Experiment Experiment Recent Progress Effect of Proposed Method Watching the Match Making Questions My Reviewing matches System Writing the Article 29