Chat analysis (google hangout and whatsapp)

little experiment I did last year, using R NLP librarires and hangout and whatsapp data.

Getting/Cleaning hangout data

Hangout data can be downloaded using google takeaway. The data that I used and tested is from 2014-2015 timeframe. JSON file provided by google is very huge and requires knowledge of data format to extract the conversation so I used https://bitbucket.org/dotcs/hangouts-log-reader/ tool to extract the convertion

after processing the output of 'hangout_reader.py' I had the file in this format:

2014-02-18 16:44:22: <1author-1> message
2014-02-18 16:44:33: <2author-2> message

Getting/Cleaning whatsapp data

Getting whatsapp data was straight forward. Just emailed the conversation to my inbox. only problem was that the format slightyly changed in between 2014 and 2015. so I had to use two different parsing code.

2014 format

09:53, 21 Jan - 1author-1: message
09:57, 24 Jan - 2author-2: message

2015 format

20/05/2015, 1:18 PM - 1author-1: message
20/05/2015, 1:18 PM - 2author-2: message

Final data format

Final data was exported to individual CSV files (cleanChat.c) with following columns

• year
• month
• day
• hour
• minute
• author
• message
• source

Data input

datain.R : to input data

WordCloud

wordcloud.R : Generates a basic wordcloud libraries

• slam
• tm
• wordcloud

GGplot2

ggplot2.R : Generates a basic graph library

• ggplot2

Other details

R version : 3.2.5

further work

will add more experiments soon.