Coffee Twalk Three: Measurement and Analytics for Social Media
RateMyNewBot
1.
JOUR779V Computational Journalism | Ramesh Balasekaran | Assignment #3 1
News Bot Design & Development
(https://twitter.com/Rate_Tweets_Bot)
The AutoReply Bot(@Rate_Tweets_Bot) created for this assignment responds with three tweets
whenever a user mentions the bot in their tweet. The bot looks for a specific format in the tweet
to perform four different function.
Search a phrase/word from the public stream and summarize and reply back top three popular
tweets
Input : ‘@Rate_Tweets_Bot #Brussels Attack#’
Output: Top 3 Tweets containing the Phrase ‘Brussels Attack’ and its rating and exposure
statistics.
Method: The bot collects tweets with given phrase from all the accounts till a set limit is
reached. The tweets are then rated based on tweet’s potential exposure or audience and also the
rate of interaction with the tweet and tweet owner’s account. It also takes in account tweet
owner’s verification status. After the tweets are rated, top three tweets with their statistics are
replied back to the user.
Search a phrase/word from the list of user timelines and summarize and reply back top three
popular tweet
Input : ‘@Rate_Tweets_Bot @Cnn #Brussels Attack#’
Output: Top 3 Tweets containing the Phrase ‘Brussels Attack’ and its rating and exposure
statistics.
Method: The bot collects tweets with given phrase list of accounts the bot follows and also the
account the user have mentioned in their tweet till a set limit is reached. The tweets are then
rated based on tweet’s potential exposure or audience and also the rate of interaction with the
tweet and the tweet owner’s account. It also takes in account tweet owner’s verification status.
After the tweets are rated, top three tweets with their statistics are replied back to the user.
Find top 3 tweets among all the news accounts
Input : ‘@Rate_Tweets_Bot send me trending news tweets’
Output: Top 3 Tweets among the set of news tweets account and its rating and exposure
statistics
Method: The bot collects recent 50 tweets from all the news accounts that the bot follows. The
tweets are then rated based on tweet’s potential exposure or audience and also the rate of
interaction with the tweet and the tweet’s owner account. After the tweets are rated, top three
tweets with their statistics are replied back to the user.
2.
JOUR779V Computational Journalism | Ramesh Balasekaran | Assignment #3 2
Find top 3 tweets among all the news accounts and also include tweet account that interests
the user
Input : ‘@Rate_Tweets_Bot @cnn @fox send me trending news tweets’
Output: Top 3 Tweets among the set of news tweets account and its rating and exposure
statistics.
Method: The bot collects recent 50 tweets from all the news accounts that the bot follows and in
addition to the predefined list it also collects tweets from accounts that the user has mentioned in
their list. The tweets are then rated based on tweet’s potential exposure or audience and also the
rate of interaction with the tweet and the tweet’s owner account. After the tweets are rated, top
three tweets with their statistics are replied back to the user.
Purpose of the Bot
Twitter provides features for viewing hashtags that are trending, however, the features only
provides options to view the stream of tweets that contains the hashtags. When the stream of
tweets is viewed it does not provide a clear picture of the trending story or displays the tweets
from a credible account at the top. In order to overcome this problem the bot is created which
will find a credible tweet and present to the user to understand trending news. This bot looks for
the best tweet that summarizes the trending story by calculating the validity of the source, the
exposure and interaction score of each tweet and sends the top three tweets to the user so that
users can understand the trending story better by directly looking at the top three tweet’s content.
Input
Output
Algorithmic Transparency
The Bot omits retweets while collecting the tweets. After collecting the tweets, the algorithm
looks for each tweet’s number of retweets , number of times the tweet has been marked
3.
JOUR779V Computational Journalism | Ramesh Balasekaran | Assignment #3 3
favourite, tweet owner’s number of followers at the time of tweet and the total number of
owner’s tweet that have been marked favourite. While calculating the rating, more priority is
given if the tweet owner is verified account and has many followers compared to other accounts.
While rating the tweets from a set of news channel, accounts that has more followers are given
higher rating because all news accounts are verified. The algorithm also takes in account number
of retweets for each tweet and number of times the tweet has been marked favourite to rate a
tweet.
Verified accounts are given higher ratings due to the availability of very few verified accounts
and also since the verified accounts are obliged to tweet meaningfully. The accounts with more
followers also are required to compose meaningful tweets to reach their wide audience. While
rating tweets from news accounts, the interaction of users with each tweet either by retweets or
favouring the tweet indicates that the tweet is meaningful and provides full information hence
tweets with higher interaction are rated higher.
Limitations and Alternatives
The bot has a limitation in terms of calculating a tweet’s interaction score while collecting tweets
from public stream since during collection the interaction is very less. The alternative solution to
overcome the problem would be to collect tweets from the public stream with a delay in time to
calculate a tweet’s interaction score better. The challenge i faced while creating the bot is the
importance that should be given to the tweet’s attribute and tweet owner’s attribute, and i felt
option to modify the importance based on user’s need would attract more trust among users.