TweetCric: A Twitter-based Accountability Mechanism for Cricket
This page contains supplementary material for the paper:
Arjumand Younus, M. Atif Qureshi, Michael O’Mahony, Derek Greene, Naif Aljohan "TweetCric: A Visualization Approach for Searching and Exploring Twitter-based Public Sentiment within Cricket" , 2017 (Under review).
Summary
Cricket is a popular international sport that has a massive fan following. Recently, cricket fans have begun to utilise Twitter to share their opinions during live cricket matches. This paper demonstrates a Web service called TweetCric to uncover insights from Twitter with the aim of facilitating sports analysts and journalists. The proposed system arranges crowdsourced Twitter data within an exploratory search interface through a comprehensive event plot and enables users to drill down into more detailed information about each player in the game through summary visualisations (i.e., interactive time series and polarity plots). Furthermore, the system also incorporates domain-specific approaches to sentiment analysis within a retrieval system which helps in conveying significant information about cricketers’ performances. The current implementation of TweetCric shows the strengths and weakness of players during a certain game while highlighting overall team performance.
TweetCric Flow (Case-Study)
The entry point to our system comprises a dashboard whereby summary information for an entire game of interest is displayed and the user has the option to navigate significant entities at various points of the game (see Figure 1). The text (in red) shows various sentiment modalities supported by TweetCric.
Figure 2 shows an example illustration for the 19th March 2016 ``India vs. Pakistan" game; here the top graph is indicative of significant events within the match as measured by tweet volume while the bottom graph is indicative of sentiment scores for our team of interest at various points in the game. The user can tap into the exploratory and interactive features of TweetCric by hovering over various peaks of a game, and each hover reveals a word cloud that indicates significant terms for that particular point. Moreover, clicking on each peak reveals the players for that particular point in decreasing order of their tweet volume, and this gives an overall picture of players' roles at various points of the match; users can further click on players' names for each point thereby exploring a player's strengths and weaknesses in significant detail.
As a case-study, we illustrate the example of Pakistan's bowler Muhammad Sami, who was highly criticised by cricket fans at the beginning of the match when Pakistan was batting. The pitch was supportive of spin bowling whereas Muhammad Sami is a fast bowler. The situation however reversed when Pakistan's fielding started and Muhammad Sami was able to show a good performance during his initial bowling spell. Figure 2 shows the word cloud for Muhammad Sami's bowling spell with words such as ``wicket", bowled", ``beauty", ``hattrick" highlighting strengths of his bowling spell; ``SureshRaina" and ``ShikharDhawan" were the two Indian batsmen bowled by Muhammad Sami. Figures 3 and 4 show the sentiment-based Likert scales for these two time intervals, which illustrate the changes in Twitter's opinion regarding Muhammad Sami as the match progressed. A sports analyst may find these findings interesting in terms of making future predictions about a cricketer's performance thereby aiding future decision-making in the game of cricket.
Sentiment-based Likert scale for bowler
"Muhammad Sami" during Pakistan's batting
Sentiment-based Likert scale for bowler
"Muhammad Sami" during Pakistan's fielding