Booksoup allows you to analyse and traverse your downloaded facebook data, including features such as sentiment analysis and message frequency analysis over time.
Booksoup allows you to analyse and traverse your downloaded facebook data, including features such as sentiment analysis and message frequency analysis over time.
Booksoup requires BeautifulSoup4 and TextBlob, and requires matplotlib to run the demo graphs.
Initialise a new instance of the
BookSoupclass, passing in the top-level path of your facebook data folder as an argument.
from booksoup import BookSoupme = BookSoup("facebook-data")
Get a conversation by name
convo = me.load_conversation("Jane Doe")
Print participants of the conversation
print(convo.participants)
Print messages in the conversation
for message in convo.messages: print(message.date, message.timestamp, message.name, message.content)
It's possible to see how often messages are sent in a specific conversation at each hour of the day using
interaction_freq. This returns a dict with each key being an hour in the day, and the corresponding value being the number of messages sent at that time over the history of the conversation. ```python me = BookSoup("facebook-data") convo = me.load_conversation("John Smith")
times = convo.interaction_freq() ```
Using the
demo_interaction_frequency.pycode, this can be visualised:
It's also possible to view how many times a specific person within a conversation sent messages from the beginning to the last point of the conversation using
interaction_timeline(name). The following example shows how often I sent messages within a group conversation.
me = BookSoup("facebook-data") convo = me.load_conversation("Lewis, Andrew, Michelle and 4 others")times = convo.interaction_timeline(me.name)
Using the
demo_interaction_timeline.pycode, I can visualise in one graph how often everyone in the conversation spoke by building a separate timeline for each person.
Another example below with one friend over a longer timeline:
Booksoup can also perform sentiment analysis. Average sentiment for a user in a specific conversation can be calculated using
Conversation.avg_sentiment(name), or a timeline of average sentiment can also be built using
Conversation.sentiment_timeline.
convo = me.load_conversation("David Grocer")Print the average sentiment of David Grocer in the conversation
print(convo.avg_sentiment("David Grocer"))
Print the timeline dictionary of my average sentiment in the conversation
print(convo.sentiment_timeline(me.name))
A conversation can either be loaded using either the title of the conversation (as in all the previous examples) or the numerical ID of the conversation (the filename of the conversation's html file).
convo = me.load_conversation(40)
In all of the timeline examples, the interval can be specified as either
monthor
day, with the default being
month. To switch to daily intervals for timeline operations, set the
intervalargument, e.g
convo = me.load_conversation("David Grocer", interval="day")
Booksoup can extract and categorise event information. This includes title, description, location, timestamp and a 2-element array containing the latitude and longitude of the event if available.
me = BookSoup("facebook-data")events = me.load_all_events()
Events are organised into attending, maybe, declined and no_reply:
for event in events.attending: print(event.title, event.description, event.location, event.timestamp, event.latlon)