Wright, A. (2009). “Mining the Web for Feelings, Not Facts.” The New York Times.  Retrieved on 30 June, 2010.

http://www.nytimes.com/2009/08/24/technology/internet/24emotion.html

Online presence is a valuable commodity in today’s digital market.  As companies seek to track exactly how their brand is discussed via the web and where these discussions appear, it becomes apparent that even a team of employees devoted to such research cannot tackle the shear size of the medium.  Thus, algorithms are being employed by marketing research firms as well as companies themselves to handle not only the amount of information present on the Internet, but also in what context it amasses.

These algorithmic tools are applied all over the web, but are concentrated on social networking sites like Facebook and Twitter, as well as sites that allow large amounts of user-generated content. Theoretically, in this way a computer can track not only when a company is mentioned but also in what connotative context it appears. Differing from previous brand tracking, these new programs seek to determine subjective opinion as well as objective knowledge.

By programming computers to scan the Internet for words that hold certain connotative meanings, marketers and brands can preemptively address user satisfaction issues as well as ensure that their brand awareness strategies are reaching the desired audiences.  While sophisticated, these programs are not as able to pick up on subtle linguistic language such as sarcasm, slang, and idiomatic expressions, which inhibit their current use online.

Although still in their infancy, such algorithms now have up to 80 percent accuracy in defining audience sentiment on the web.  The attractiveness of such a tool to marketers and PR firms ensures that more research will be done to improve both accuracy and wide applicability to all types of brands.  These programs are further equipped with user generated feedback options which allow the program’s accuracy to be improved with increased use.  Thus, like Facebook and Twitter, these programs become highly adaptable and able to recognize and modify their behavior based on user generated content.

The goal of these programs is to allow companies to foresee potential threats to brand reputation and user satisfaction before they become too big to address.  Furthermore, adapting objective versus subjective defining programs for broader use could ensure greater transparency in search engines, movie and food reviews, and e-commerce.

USER GROUPS: Obviously the Internet is only widely used by certain socio-economic and cultural groups, so such programs will only be able to attract these certain types of users.  Also, angry or enthusiastic users are more likely to talk about a brand online, which also inhibits marketers’ understanding of how their brand is accepted across large user groups, especially the ones who may have greater brand loyalty but may not necessarily be online.  The system must take this into account so the company does not mistakenly believe that all product users are adopting the same attitudes, unwittingly changing the product or addressing issues that may upset a broader consumer base not present online.

CONCEPTUAL DESIGN: These programs seek to not only aid marketers, Public Relations firms, and companies themselves, and also address the concerns of the consumers of such brands.  By attending to potential problems in user opinion the brand can preemptively act in accordance with user wishes. At the same time, these strategies can also be applied to manipulate public opinion or distract the public from certain emotionally charged issues, protecting the brand from backlash from its consumer base.  New specialists must be created who understand the system and its functions as well as its role in predicting market response.

INTERACTION DESIGN: Although the idea of the widespread use of “sentiment trackers” on the web does have its benefits to both marketers and consumers, it has yet to be proven statistically accurate at a level that would make it usable.  70-80% accuracy is not enough to be sure that the program is reading the public correctly.

Also, the article does not address exactly how user input is used to improve the system.  More study must be conducted into how to overcome the programs’ shortfalls in order to market the system to potential clients.  Marketers or companies who use the application must be confident in the abilities in the product as well as be well versed in how it works in order to describe its workings to the user.  If user generated content is to be used, public error potential and education must also be taken into account.

INTERFACE DESIGN: The definitions of “objective” “subjective” language can vary from culture to culture and social group to social group, making the roots of failure for the system a concern.  It is also imperative to know exactly whom the product seeks to interpret and who is running the system in order to determine possible areas for miscommunication or product failure. Designers will have to create a product that is translatable to a wide range of users, from corporations looking to guard against brand tarnishing to individuals wanting an objective opinion on the best pizzeria in their area.