Paper #4: Summarizing Sporting Events Using Twitter

Summary

    The Almaden IBM research group is researching into ways of using algorithms to track tweets during sports games so the collected tweets can be used to quickly create a news document faster than a reporter can by themselves.  They look at Twitter statistics and determine which tweets to mine by using Keywords and activity spikes.
 

        They use activity spikes to find appropriate tweets based on the logic that when something important happens people feel that they should comment on it. Afterwards they use Heuristics to filter out alot of the tweets, including tweets in other languages, tweets with URLS, and tweets that are replies to other tweets. The researchers used several heuristics and eventually created an algorithm which created game summaries based off tweets. Afterwards , They used People to read and evaluate said reports

Related Work Not Referenced in Paper

• Using  Lexical  Chains  for Text  Summarization 
• AUTOMATED  TEXT  SUMMARIZATION AND  THE SUMMARIST  SYSTEM 
• Generic Text Summarization Using Relevance Measure and 
  Latent Semantic Analysis
• Seeing the Whole in Parts: Text Summarization for 
  Web Browsing on Handheld Devices
• T h e   T I P S T E R   S U M M A C   T e x t   S u m m a r i z a t i o n   E v a l u a t i o n  
• What is Twitter, a Social Network or a News Media?
• Social networks that matter: Twitter under 
  the microscope
• Twitter Power: Tweets as Electronic Word of Mouth
• Twitter and status updating
• Using Twitter to recommend real-time topical news

I looked up both Text generation and Twitter related articles on the internet. Twitter seems to be a hot topic, there was no shortage of reports of Twitter and how it can be used as a news source.    

Evaluation

    In the report they evaluate both the important moment detection and the summarized reports. They evaluated the moment detection subjectively and quantitatively by comparing the moments the twitter algorithm outputted with news articles from various news sources such as ABC.com, ESPN.com, etc. They found the certain events were less likely to be tweeted ( such as yellow cards ) meaning games that had alot of these events, we're summarized less accurately.
   They also evaluated the summarizes both objectively and subjectively by using software and humans to evaluate the summaries. The summaries did a good job recapping important events but didn't quite pack in small details like a human made news article does.
 
Discussion

I think this will be a cool technology which could really make news reporting for sports very fast and up to date. I think the evaluation is appropriate because news is something that is enjoyed by humans, and therefore needs to be evaluated objectively.