Innovative tools and programs are now available to online businesses to gauge the mindset of their customers. These tools pinpoint their perceptions and thus help astute business heads respond with relevant public relations and Internet marketing strategies.
The Financial Times has just recently introduced an experimental program Newssift, which tracks sentiments about certain business topics currently in the news, along with a specialized search engine.
This lets users organize their queries on basis of topic, place, person, organization and theme. For example, employing Newssift, a search for Wal-Mart suggests that sentiment about the firm has recently turned positive. However, when the search is refined with another suggested term ‘Labor Force & Unions’, the ratio gets unfavorable.
The simplest algorithms function by scanning relevant keywords to categorize a specific statement as negative or positive, on basis of a simple binary analysis. Simpler incarnations of such sentiment analysis are also becoming available for casual Web visitors in the form of more ‘lightweight’ tools such as Twendz, Twitrratr and Tweetfeel. These let users gauge the mood of Twitter users about a range of topics.
However this approach does not succeed in capturing the subtleties, which bring complex human language to life: sarcasm, slang, irony and many other idiomatic expressions. The more advanced algorithms by Jodange, Newssift and Scout Labs use refined analytics to achieve an accuracy of bout 70 to 80 percent. Importantly, users can reclassify inaccurate results on their own for the system to rectify its mistakes.
Sentiment analysis algorithms will gradually get more sophisticated to produce more accurate results. The trend may ultimately lead to thoroughly sophisticated filtering mechanisms, which could well become an integral part of everyday Web usage, an aspect that needs to be kept in mind by businesses in Winnipeg and Portland, while devising their Internet marketing strategies.
