Find Research
Our World View
Industry Perspectives
Research Program
Parallax View Magazine
> stay tuned
View our collections of research around key subject areas:
>
>
>
>
ERP
>
>
>
>
>
>
>
SRM
>
>
WMS
 
 
Article
Social Data Mining: Beyond Sentiment

Quite a few companies are focused on mining public social media data for consumer sentiment indicators. However, there are many other insights and information of value that can be gleaned from Tweets, blogs, and Facebook pages. We are just beginning to understand how various companies can exploit these data--whether for risk management, financial investments, or other uses.


Full Article Below -
Untitled Document

The volume of social media data has exploded in the last few years. The number of active Facebook user accounts surpassed 1 billion last year. People are sending over 500 million tweets per day. There are at least 250 million blogs on the web and countless discussion forums. Furthermore, the trend line for all of these continues upward as far as anyone cares to predict.

Mining for Sentiment

A significant portion of this social media data is free for the taking by anyone who wants to mine it. And lots of companies are mining it. Sentiment analysis systems attempt to use social data to analyze how much consumers like or dislike a particular product (or political candidate, or stock, or other entity). More sophisticated sentiment analysis tries to discover exactly which aspects of a product or service consumers do or don’t like, by how much, why, and the importance of each aspect.

These analyses treat the world of social networks and social media as a sort of giant focus group. This can be very valuable to companies in a number of ways such as continually improving your products, fixing bad service, or quickly identifying influential negative comments in order to take steps to mitigate the problem(s) directly with those who are complaining before it goes viral. Political candidates may also continually refine their message based in part on such sentiment analysis. Some companies, such as 3 Tier Logic, incorporate social media sentiment analysis into web/social engagement and promotional campaign management. Others, such as First Insight, create their own social media sentiment data by engaging consumers in various online games to determine the most attractive product attributes and price elasticity before a product launches.

Beyond Sentiment

However, there are many other practical uses for social media data besides assessing sentiment about products.

Crime-fighting—Police are beginning to use social data to detect and counter criminal activities. This has been largely a manual effort in the past, but is starting to be automated, such as X1 Social Discovery which enables law enforcement to search across multiple social media streams as well as crawl, capture and instantly search content from websites, webmail, and YouTube. In addition to helping identify and solve crimes, social data can sometimes be used as evidence in prosecution, such as when a criminal brags about their crime online or (astoundingly) posts a video of the crime online (yes, some have actually done that).

Financial Trading—Traders and investors have long combed social media as part of their research. Now tools are emerging such as from startup Alphamatician which uses social data not just to gauge changes in market sentiment, but also to alert about events that can impact the fundamentals of a business.

Supply Chain Risk Management—A potential use of social data is helping supply chain risk managers become aware of and monitor supply chain risks, such as natural disasters, explosions or fires in a manufacturing plant, labor strikes, transportation lane disruptions, and political unrest. People often tweet about these events well before the official media starts reporting. Social media may provide details not available elsewhere, though of course the reliability and accuracy of social data has to be verified.

Epidemiology—Academics and government agencies are exploring how to use social data to help identify, track, and predict disease outbreaks.

National Security—The US government is mining social data to try and identify threats to national security, prompting debate over the extent this invades the privacy of ordinary citizens. Project Quantum Leap1 conducted an experiment combining social data with other transactional data to detect and counter the financing activities of terrorists, insurgents, international organized crime, human traffickers, and weapons proliferators. More controversially, the NSA uses social media, geo-location information, and other public and private sources combined with analysis of phone and email records to identify threats to national security.

Demand Sensing—Companies like ToolsGroup are adding social media data and machine learning to their other data and tools to improve demand sensing.

Emergency/Disaster Response—Response coordinators can become aware of and assess damage, gaining a more complete picture sooner, using social data, combined with their other monitoring tools such as remote sensing and video. Oak Ridge National Laboratory is researching ways to mine social media data for facts relevant to disasters, quantify the uncertainty in that data, and integrate that information with remote sensing and GIS data for rapid response.

What’s Next for Social Data Mining?

We expect new types of social data to emerge. Twitter, Instagram, and Tumblr won’t be the last new social media phenomena. An exciting trend (I’ll leave it to others to debate whether or not to call it social data) is crowd-sourcing of valuable information, such as safety information in LiveSafe, which includes a mobile app that lets people report all manner of incidents such as theft, auto crashes, fires, riots, sexual assaults, medical emergencies, and other incidents that the public and authorities should be aware of. The same app will show users what is happening and help them avoid trouble spots.

Social data can be an important element of creating real-time situational awareness,2 which is a key component of many of the above use cases, as well as others. TransVoyant’s Continuous Decision Intelligence ingests and evaluates social media feeds along with other real-time data feeds such as sensor, weather, traffic, video, preference and asset location and uses rules and decision algorithms to detect specific events and provide this type of real-time awareness and intelligence for use in many of the applications listed above and more (security, crime fighting, logistics, supply chain risk, etc.).

We are in the early stages of discovering the uses for this data. As the volume and variety of social data, continues to increase, businesses, governments, and individuals will uncover more and more use cases for extracting value from that data. We’ve only just begun!

___________________________________________

1 See Mining social media: The new way of life -- Return to article text above
2 Recently we wrote about this in Continuous Real-Time Situational Awareness for Time-Critical Operational Decision Making.


To view other articles from this issue of the brief, click here.




MarketViz powered.