Knowledge-Assisted Ranking: A Visual Analytic Application for Sports Event Data

Abstract

Organizing sport video data for performance analysis can be challenging, especially when this involves multiple attributes, and the criteria for sorting frequently changes depending on the user's task. In this work, we propose a visual analytic system to convert a user's knowledge on rankings to support such a process. The system enables users to specify a sort requirement in a flexible manner without depending on specific knowledge about individual sort keys. We use regression techniques to train different analytical models for different types of sorting requirements. We use visualization to facilitate the discovery of knowledge at different stages of the visual analytic process. This includes visualizing the parameters of the ranking model, visualizing the results of a sort query for interactive exploration, and the playback of sorted video clips. We demonstrate the system with a case study in rugby to find key instances for analyzing team and player performance.

State of the Art of Sports Data Visualization

“…As a result, with few exceptions (e.g., [PBV16, TSLR07, PYHZ14]), most of the research papers we surveyed do not contribute a new generic technique but instead a tailored adaptation of an existing one. The most frequent visualization technique extensions we found were of node‐link graphs (e.g., [PVF13b]), heatmaps (e.g., [PSBS12, Gol12, Bes16, AAB*16, MB13]), small multiples and glyphs (e.g., [HZ16, LCP*12, CPG*16, CF13, PYHZ14], treemaps (e.g., [JB97, CS06, Tur94], arc diagrams (e.g., [OJK13]), parallel coordinates (e.g., [OJK13, CPG*16], and line charts (e.g., [Woo15, PBV16, Won16]). For the same reasons, sports data visualization is also propitious to the design of new interaction techniques .…”

“…This is the case with The New York Times' Peyton Manning's touchdown passes record [AQ14]; Paths to the Top of the Home Run Charts [Tim07]; comparison of baseball players who reached 3000 hits in their career [CQW11] or of NBA players in terms of number of points scored during playoffs [Pea17]; and with the comparison of NBA players performances in terms of points, rebounds, and assists per game over the season [Gol17a]. One notable example of this technique is Five Thirty Eight's sumo project [Con16], showing the history of sumo from 1761 to 2017, using overplotting of line graphs with color coding and annotations. The density of information in those visualizations creates a baseline that usually gathers the majority of players from which a select few top performers visually stand out.…”

“…With Buckets [Bes16] and BKViz [LTB16], it is possible to explore the points a basketball team scored against each other team in the league, among many other facets of the data. Chung et al's knowledge‐based ranking system features multiple coordinated views to explore the multiple facets of rugby games [CPG*16]. These approaches result in powerful tools in which many dimensions can be explored simultaneously—as opposed to more static infographics—but at the cost of being more complex to use.…”

“…A History of Sumo[Con16] is a stylized interactive visualization that presents the history of sumo using storytelling and exploration capabilities. Image courtesy of the author, used with permission.Showing relative differences rather than absolute values.In sports such as skiing, running, and racing, time is what matters most.…”

KAVAGait: Knowledge-Assisted Visual Analytics for Clinical Gait Analysis

“…1) The tree structure of the EKS while selecting a category ('Ankle' in this case) for adjustment 2) based on the 'Hatching Range Slider' (HRS) representing the limits or norm-ranges of the spatio-temporal parameters from the patients included in the category. 16 STPs in the 'Parameters in Category' column. Additionally, the system calculates how newly loaded patients match to the stored 'Categories' in the EKS.…”