Can you automatically detect affect in linguistic media such as speech and text? In December, I mentioned a Scottish company that claims to identify mood in voice and use the knowledge to power a safer driving system. Spotting affect in text as an indication of a product/company/brand/policy/person’s reputation, say, is possibly harder. You can’t just go for keywords such as “bad” or “brilliant” in case they’re embedded in a negative expression, as the New Scientist kindly explains in a news item on UK company Corpora Software.

“Corpora has come up with a program called Sentiment, which uses algorithms to tease out grammatical components, such as nouns, verbs and adjectives, and identify the subjects and objects of verbs. It can even analyze pronouns like “it”, “he” and “her” to work out what words or concepts they are referring to.

Having an understanding of grammatical structure makes it possible to filter out words that are not relevant to the sentiment of the article, Jacobi says. So instead of assuming certain words, such as “unpredictable” or “rubbish”, are positive or negative it allows the structural context to disambiguate them.”

When the web first came along, everyone used purely formal indicators such as page hit numbers as a sign of reputation. Now we have entered the actual content, and use keywords in search engines that can dredge up an apparently relevant advertisement and place it on your page. Unsurprisingly, if you Google on “stupid” “hateful” and “rubbish”, you get no ads. Since humans can perform metalinguistic operations on a “bad” word to produce an ironical effect, we are now obliged to start parsing texts to see whether the grammar of words can automatically hint at attitude. Presumably, a first step towards some sort of semantic processing of linguistic viciousness. Another ten years ( as they proverbially say in the NLP industry) and we could be reading about a Piranha Brothers engine. Remember Monty Python’s Doug Piranha:

He used… sarcasm. He knew all the tricks, dramatic irony, metaphor, bathos, puns, parody, litotes and… satire. He was vicious.

Might have been a description of computer HAL in the 2001 movie?