The Intelligence of Social Media (Part 2)

In the first part of this blog, I mentioned that sentiment analysis measures the polarity of opinion—positive, negative, or neutral—regarding a subject, a product, a service, etc.

Two main approaches can be used to perform sentiment analysis or text mining: a knowledge-based approach, which uses linguistic models to classify sentiments; and a learning-based approach, which uses machine learning techniques to classify text. The concept of sentiment analysis opens a great number of possibilities and opportunities for introducing BI strategies to analyze the enormous amount of data flowing through the Web.

In fact, some software solutions have been designed to address this type of analysis. These tools are called “social Web analytics.” According to the definition provided by The Social Web Analytics eBook (2008), by Philip Sheldrake, social Web analytics are “the application of search, indexing, semantic analysis and business intelligence technologies to the task of identifying, tracking, listening to and participating in the distributed conversations about a particular brand, product or issue, with emphasis on quantifying the trend in each conversation's sentiment and influence.”

Many organizations are aware of the importance of measuring this information and analyzing it. Currently, sentiment analysis has a strong potential to be used jointly with BI applications making it possible to apply traditional BI techniques to visualize what a sentiment-based tool has discovered on the Web. Some vendors are already offering analytics services (radian6, Sysomos, BuzzLogic, and Attentio) to measure and analyze social media content.

Now, there is also an existing trend regarding traditional BI providers to address these tasks:

•    Teragram (a division of SAS Institute) unveiled its Sentiment Analysis Manager (SAM), which is a social media analysis tool that applies BI techniques to sentiment analysis. This tool can analyze content from social media sources like Amazon or Twitter. It also analyzes ratings from Web surveys in order to evaluate products and services.

•    BusinessObjects Text Analysis from SAP is able to examine text in 30 different languages to allow users to gain an understanding of product opinions from Web customers around the world. It can extract and analyze text automatically from several sources based on preconfigured data items (people, places, dates, etc.). It can highlight sentence relevance and prepare executive reports based on the content found.

•    SPSS Text Analysis for Surveys 3.0 from IBM can help improve text analysis. It enables companies to extract and analyze survey responses.

The effort focused on sentiment analysis is customer-oriented, which in turn has more impact on customer relationship management systems (CRMs). This enables more operation related users to be more involved directly in BI related tasks. As a result, the use of BI is widespread among a larger number of users in each organization. Social media is a catalyst that gives information empowerment to users. Organizations are not only increasing the amount of information generated, but they are also increasing the number of knowledge-based workers whose capabilities include gathering and analyzing this information. In brief, employees are also part of the social media networks.

Due to the nature of social media sources and its behavior, it is difficult to calculate social media return on investment (ROI). Social media are means by which individuals can interact; they are also segmented channels—therefore, there is no direct ROI nor is there a general consensus on how to obtain it accurately. Some companies think social media analysis adds value to the company. Other companies assess overall media ROI efficiency and try to evaluate how social media affect overall productivity.

Sentiment analysis and text mining may still be under development, but there is no doubt that traditional BI vendors are addressing sentiment analysis  to target other possible markets in the BI field. Companies are realizing that understanding a customer’s position as expressed or represented in social media can help gain knowledge of people’s opinions regarding their products and services.

Other aspects still need to be determined regarding the evolution of text mining and sentiment analysis, but today it is clear that social media represent tools that can be used to understand market trends. Traditional BI vendors are making it possible for companies to address these types of tasks directly with their own set of tools, or making the necessary arrangements to do it.

While BI is being “socialized”, social media will be analyzed more from a BI perspective. In the near future, every commercial initiative will probably have very robust and mature social media analysis. It will be important to follow the trends created by Social Media Analytics in order to determine how sentiment analysis technologies will evolve enough to encourage more companies to embrace social media analysis initiatives.
comments powered by Disqus