Author: Alexandra Cernian – Lecturer, Faculty of Automatic Control and Computers
Companies are turning to sentiment analysis to learn what customers think and to boost customer-centric strategies for sales and marketing campaigns. Predicting consumer behavior can be a difficult task, especially with significant amounts of data coming from so many and varied channels of communication, one of the most representative being social media. Therefore, marketing departments rely on using predictive analytics and devising marketing campaigns based on various criteria, in order to improve brand awareness and loyalty and to approach consumer interactions at a higher and more personal level.
Why is it important for the Internet to know if we are happy or sad?
Research conducted by IBM shows that 2.5 quintillions (2.5 x 1018) bytes of data are produces each day, that is the equivalent of 530,000,000 songs or 5 million laptops. 90% of the datat in the worlds were genertated in the last two years, and the rythms of growth is very alert. The sources of data are very varied and heterogeneous, social media being one of the most significant data generator these days. In this context, the Big Data phenomenon becomes more and more significant on the Internet and leads to the need to create innovatove analysis methods to process these data.
It is important for those who use the Internet for marketing purposes to know whether we are happy or dissatisfied about what they have to offer. Depending on the feedback, they immediately know how are positioned on the market and what adjustments need to be made in order to appeal to consumers. Romania is in the lead in terms of Internet usage. Results of a study conducted by the National Authority for Management and Regulation in Communications (ANCOM) shows that 90% of users in Romania use the Internet to search for information, read news and keep up to date with events, while 77% they use it to access social networks. It is therefore a natural step fot hose who „sell” anything to be up to date with their online users..
Sentiment analysis brings new insights
- Proactive attitude in cutomer service. Recent studies conducted in the US showed that 81% of Internet users have researched over a product online at least once. The analysis of opinions expressed online by buyers gives a 360-degrees image of the market and gives companies real-time feedback about their products and services. Thus, customer service consultants can have a client-centered and customized approach.
- Improved product and marketing strategy. Sentiment analysis tools can be successfully used by companies to retrieve product feedback, find out what is being said about their brand and the competition on the market, in order to use this information in developing marketing campaigns that address the needs and perceptions of their target audience.
- Increase brand reputation. Sentiment analysis can help companies monitor consumers’ opinion about the brand, to enhance their reputation and increase their visibility. Thus, they can anticipate changes in perception among social media users and prepare their business strategies one step ahead of time. Moreover, they can easily identify the influencers among users in order to reach them and increase their level of awareness on the market.
- Competitive advantage. Sentiment analysis tools provides an overview of the entire market, from which you can quickly learn the status of competitors and identify new opportunities.
What are the current challenges in sentiment analysis?
Understanding human emotions is a complex process. Every language comes with specific aspects of grammar, syntax, rules etc. In addition, the analysis of feelings is set back by taking into account the context, the nuances of words, irony, grammatical errors and so on. Many variables are involved in the equation and it is difficult to integrate all of them into a universal pattern. Detecting the meaning of a word based on the context requires advanced semantic technologies for interpreting texts. A word can have a positive connotation in context, however negative or ironic in another.
People have different ways of expressing their opinions. Sometimes small differences between the two texts make a big difference, and a sentiment analysis tool must take into account these situations. Moreover, people can express contradictory ideas in one post. The majprity of online reviews contain both positive and negative comments, which becomes more difficult to manage if both are found in the same sentence. Currently, sentiment analysis systems cover the classification of texts as expressing a positive, negative or neutral opinion, while the spectrum of human emotions is much more extensive and nuanced.
Conclusion
Sentiment analysis tools can certainly bring added value in marketing analysis and can help companies to better meet the expectations of the market and develop customer-centric strategies. The vision that I promote in this field through the research projects that we coordinate assumes that current limitations can be overcome and we can get a more refined classification of social media opinions by identifying human emotions and framing users in various personality and behavioral profiles. This information can be used by marketing specialists to target their campaigns and work on a more customer-centric approach.
