NLP & Social Media

What’s the Role of NLP in Social Media?

According to John Rehling, ‘NLP can analyse language patterns to understand text. One of the most compelling ways NLP offers valuable intelligence is by tracking sentiment — the tone, or opinion, of a written message (tweet, Facebook update, etc.) — and tag that text as positive, negative or neutral.
Much can be gleaned from sentiment analysis. Companies can target unhappy customers or, more importantly, find their competitors’ unhappy customers, and generate leads. We can call these discoveries “actionable insights” — findings that can be directly implemented into PR, marketing, advertising and sales efforts.
Considering the prolific spread of thoughts & opinions via Social Media platforms, the opportunity to analyse and glean meaning of this ever-increasing amount of data through NLP offers insights into people’s behaviour, attitudes and decision-making’.

Social media holds unique challenges for NLP

However, Social Media holds unique challenges for the application of NLP both through the use of symbols such as emoticons to convey sentiment and the usage of casual speech as opposed to the more rigorous texts of official documents and mainstream media. Ambiguity is a particular challenge in relation to Twitter and other Microblogs as their restrictions on post length prohibit textual context, implying an assumed knowledge that, though not available explicitly, can add tone to an otherwise non-descriptive post. Furthermore, due to the dynamic nature of Social Media, sentiments can fluctuate very quickly, and analysis can become outdated before long. This means that the chosen NLP schema must not only be able to accurately identify emoticons, abbreviations, double negations etc but also be able to identify newly added abbreviations, intended misspellings and much more.
Another challenge relates to identifying what impact a sentiment might have. This includes processes to define authorship and linked networks as well as quantifiable indicators such as retweets, shares, likes etc.

Benefits of analysing social media data

The benefits of using NLP in analysing social media data include opinion mining for companies & organisations (perception of products, competitors, added services) and sentiment analysis regarding historical and current events. Social Media due to its very nature (dynamic, user generated, networked etc.), when analysed through NLP, can quickly identify perceived problems with a product / service, define trends and help develop & improve companies’ strategies.

Approaches to NLP on social media

Using NLP on Social Media data commonly involves a series of steps as detailed in part 1 (setting up of polarity lexica/gazetteers, establishing polarity scores, rules to deal with emoticons, etc.). It also includes mechanisms for language detection, topic clustering algorithms (to identify and cluster relevant posts) and prioritising target identification.
Further reading on NLP and Social Media:
Diana Maynard, Kalina Bontcheva, Dominic Rout: Challenges in developing opinion mining tools for social media
How Natural Language Processing helps uncover Social Media Sentiment