Many libraries exist that allow you to extract noun phrases from text. It’s almost a given in english language that nouns and noun phrases are the best candidates for features. If the purpose of the extraction is to mine features, then it is advisable to check for the existence of adjective words (JJ) that may be actual sentiment words – representing opinion about the noun phrase.
Consider the following example: Say we wish to extract noun phrases from the sentence “extremely good customer service” using the TextBlob library in Python. The POS pattern for the sentence will be “[(u'extremely', u'RB'), (u'good', u'JJ'), (u'customer', u'NN'), (u'service', u'NN')]”. Prima facie, you’d think that the noun phrase and the feature here is “customer service”, while the opinion about this feature is “good”? Well that’ll be somewhat wrong according to the implementation of TextBlob (and my current understanding of feature extraction and opinion mining). My idea of a more accurate extraction will be to filter out noun phrases that have adjectives which are actually sentiment words. That way, your feature will be “customer service” while the opinion will be “extremely good” or just “good”. Of course this is open to interpretation.
In a different world, Outlook would’ve been my favourite mail provider. My love for them died a horrible death when they had identity crises, confusing us with Hotmail, Live and now Outlook.com. In that period switching to Gmail was the most reasonable thing to do, and I haven’t regretted it since then.
Although Outlook.com has arguably the best intuitive and well-designed minimalist interface, I prefer Gmail due to their effective spam filtering algorithms and the tabbed inbox, which categorise your mail into sections. Let’s not talk about the targeted ads in your inbox; they’re now deprecated, I think. The only thing Gmail doesn’t do well is to help you clear up your mailbox.
Outlook.com has a Sweep button that allows you move and delete your emails in batches based on the email sender’s address. You can even schedule a clean up routine where Outlook.com intelligently deletes or moves your email based on some specific rules. I use this tool all the time to maintain an almost-zero inbox. It saves me the time I’d have spent searching for emails I want to delete (LinkedIn updates and the likes).
While the mail team at Yahoo are dicking around moving pixels instead of fixing login problems and improving their spam filters, I hope that the Gmail team improves their already-perfect product by copying the Sweep feature from Outlook.com.
Google can do anything, except creating human beings. I hope they find reason to shamelessly copy Outlook.com, just this once.
As I’ve now decided to pursue a career in academic research, I’ve decided to conver this blog to serve as a monograph for my academic ramblings and other notes about technology. Yes, I have shown no love to this blog, I won’t give excuses. That said, I’ll still be posting about Microsoft’s tools and technologies, Andriod development, Python language an related framework. Hopefully, someday I might even write about client-side technologies too. For now you can keep up with me as I blog on Tumblr.