30 April 2010
Expert System, leading provider of semantic software that searches, discovers, classifies and interprets unstructured text information, today announces a report showing how the words and concepts used by the three leading United Kingdom general election candidates – Brown, Cameron and Clegg, can be analyzed through semantic technology as well as from a quantitative perspective.
The company processed the transcripts of the 2010 UK general election debates using COGITO, the semantic software which understands the meanings of words and is able to assess the correlation between different unstructured texts. The analysis was focused on what was actually said by the three politicians, starting from a simple question:
“From a semantic standpoint, did Brown, Cameron and Clegg say basically the same things?”
After analyzing the results obtained at the end of the automatic analysis, Expert System has developed a report focusing on the elements which had appeared to be significant; lexicon and the grammatical structures, lexical frequency distribution, the amount of words, phrases and prepositions the three candidates have used, the rhythmic structures, the most frequently used adverb, adjectives etc.
The report describes the findings from the Cogito analysis, and some interesting considerations (i.e:
– Cameron tended to use phrases which were a bit more articulated than the other two candidates: perhaps he has a greater aptitude for public speaking?;
– Clegg pronounced some very lengthy phrases: perhaps he is a more emotional speaker?
– And Brown, tended to use less, and a lower variety of, adverbs: perhaps he is a more controlled speaker?)
“People are often heard to say that political election leaders of the three main UK parties all say the same. Perhaps many even use this as their excuse not to vote”, said Bill Porter, Expert System Business Development Director UK. “So we at Expert System decided to put this assertion to the test with our Cogito semantic technology. We analyzed the transcripts of the recent televised election leaders debates show whether, from a linguistic point of view, there are differences or similarities between what the three candidates said. We limited ourselves to “chopping up” what was said and evaluating the most abstract essence of the concepts discussed, the use of the lexicon and the grammatical structures. Our conclusion was that the correlation between the three was not very high, although it is interesting to see how this varies between issues. There are differences, so placing a vote does make a difference!”