Any one worked on Sentiment Analysis
What could be resources of data for working on sentiments analysis if the targeted region doesn’t use the mainstream social media like Facebook or Twitter?
They might use whatsapp or text to communicate, but is there a system to access these (which i think is not)
By design, native WhatsApp is encrypted end-to-end making separate analysis of those conversations impossible. To use WhatsApp or text messaging, you will need to feed the communication channels through another system, which could then expose the conversations via API. For WhatsApp, the WhatsApp for Business service allows this, but is currently in only limited release (or so I’ve been told…) There are third-party companies that can assist with getting this set up (Twilio, for example, can assist with WhatsApp, text, and other channels.)
For the record, I am the IT architect for a project in my company looking at these kids of services… not specifically for sentiment analysis (yet), but rather customer helpdesk interactions that allow us appropriate levels of visibility into the discussion regardless of the customer’s preferred channel. But I definitely see sentiment analysis as a roadmap for my company’s initiative within the next year or two.
Sentiment Analysis is number one tool to understand the issues for say Elections, Brand Engagement, Competitor Analysis etc…Unfortunately the data is closed source for many channels.
My long term goal is to work with political establishments to analyse the issues.
Yep… in the age of increasing data privacy demands, it will become more and more difficult to access such data sources. And those demands don’t have to come from within the US. The EU’s GDPR protects all EU citizens, regardless of where they reside in the world or where the company that collects/stores their data is based. With penalties of up to 4% of global gross annual revenue, it’s too serious to ignore.
OSINT is rather hard to do for the average person on the street. Usually one has to do a lot of screen scraping and api crawling bots to build out a data set.
Luckily, Python (Juypter + NTLM + Scrapy) shines in this area.
These few examples should get one started and while they do show how to use things like imdb and twitter’s APIs think of them as the building blocks that one can utilize to build more targeted applications.