Sentiment Analysis using Twitter API

  • Published on

  • View

  • Download

Embed Size (px)


Sentiment Analysis on Android Lollipop Using Twitter APIPedram BashiriRohit RuparelJoseph CaprinoIntroductionHuge growth in the use of microblogging platforms such as TwitterOrganizations mining Twitter for information regarding what people think about their products and servicesTwitratr (, tweetfeel (, and Social Mention ( advertise sentiment analysisOverviewSentiment analysis on Googles Android LollipopWe are using big data techniques to better do this analysisUsing Twitter API to download tweetsDownloaded unique tweets from hashtags #Android5 #AndroidLollipop as textilesSystem Architecture

Achievements30,000 unique tweets downloaded via Twitter4j library interfacing with Twitter APICode checks for duplicate tweets and ensures we have unique tweetsPerformed a word count analysis utilizing MapReduceAchievementsStored the results of the word counter utilizing MapReduce into HBaseUtilized HBase Java APIGathered the word count analysis data and visualized itAchievementsProcessed a number of tweets and determined their sentimentIgnored bad tweetsUtilized Stanford Core NLP for sentiment analysis on remaining tweetsSaved results of scores and their frequencyVisualized the sentiment analysis resultsIssuesSentiment Analysis:Ads without URLs capturedScores differ for almost identical tweetsFuture WorkDistributed sentiment analysisMore advanced visualizationsCreating our own dictionary to do sentiment analysis