5 ways to analyze text
Finding the connections in your unstructured data
Kathy Lange, Senior Director in SAS’ Business Analytics Practice, says at least 75 percent of an organization’s data is unstructured – including images, content of Web documents, standard documents, audio, video, e-mails and call center or claims notes – and the volume is growing rapidly. “With text analytics, we are able to extract meaning out of large quantities of various types of textual information,” says Lange. “We combine it with structured data within an automated and repeatable process to enhance business insight.”
With more and more people talking about unstructured data, you may have started to encounter terms such as natural language processing, content categorization and semantic analysis. But what business value could you possibly get from such academic-sounding technologies? You'd be surprised. Leading organizations including The Associated Press, Forbes.com, HP, NY Times Company, washingtonpost.com, Wolters Kluwer, the World Bank and Yahoo! use SAS to analyze unstructured data and understand customers, market opportunities, internal operations and supply chains.
Use the following examples as inspiration for ways your organization can use text analytics:
1) Pair online borrowers with investors: Unlike traditional loan applications, Agata's forms for lenders and borrowers include many open-ended questions. SAS Text Miner made it possible for the online social lending and credit scoring marketplace to analyze unstructured data and interpret the traditional score using information from community participants and their experience profile taken from the Web.
2) Automate online recruitment: ChinaHR.com, the first company to offer online recruiting services in China, uses text analytics to extract data from résumés regardless of what format they are in, and identifies key information requested by employers. “SAS helps us stay a leading force in online recruitment in China and across Asia,’’ says VP Hai Yang.
3) Reduce workplace injuries: AFA Insurance is analyzing free-form, handwritten text on insurance claims forms to find connections among thousands of documents and terms and present results in meaningful population clusters. Text analytics has decreased administration costs, reduced workplace injuries and improved the quality of analyses.
4) Anticipate and address public complaints: The Hong Kong Government's Efficiency Unit analyzes citizen complaints in multiple languages to better understand underlying issues and quickly respond even as situations evolve. “The solution consolidates all information and uncovers hidden relationships through statistical modeling analyses,” says the Efficiency Unit’s Assistant Director, W. F. Yuk. “It helps us understand hidden social issues so that government departments can discover them before they become serious, and thus seize the opportunities for service improvement.”
5) Quickly classify and tag content: Major US media companies automate the process of grouping and tagging content for hundreds of items per day at TV and newspaper websites across the country. One leading daily paper reports that all content is automatically tagged, articles are showing up higher in search engine displays, and e-mail news alerts have a 50 percent click-through rate.