Characterizing Dutch Job Advertisements in Twitte

Group 06 (Aaqib Saeed, Muhammad Arif Wicaksana, Alexandru Serban)

In this project, we try to characterize Dutch job advertisement tweets in period 2014-2015. Questions we try to answer:

1. What kind of jobs are posted on Twitter?
2. How is the job trendings over the period?
3. Which area over the most jobs on Twitter?

Our screencast:

##Files:

• geo-tagged.pig is for geo tagged tweets generated by twitter-geotagged.jar.
• nongeo-tagged.pig will sample the tweets generated from map reduce job by twitter-nongeotagged.jar.
• word-freq.pig will generate word counts/frequencies and the output will be used by R script.
• The folder bigdata-0.2 contains MapReduce code.
• folder parsed_job_tweets contains job tweets parsed from the datasets.

##Usage: ###MapReduce hadoop jar twitter.jar nl.utwente.bigdata.TwitterR <INPUT DIRECTORY> <OUTPUT DIRECTORY> ###Pig Latin

pig –x mapreduce wordfrequency.pig
pig –x mapreduce general.pig

###R

dataset <- read.delim("[FILE PATH]", header=FALSE, quote="", stringsAsFactors=FALSE)
dataset$V1 = tolower(dataset$V1)

wordcloud(words = dataset$V1, freq = dataset$V2, colors=brewer.pal(9, "Dark2"))