KM past and present:
The focus is on integration, scope and analytics
The software analyzes social data and tries to determine who is actually on a buying journey and then decides whether the consumer is expressing awareness, purchase intent or commenting on a previous purchase. Survey brand health trackers may only measure that factor once or twice a year, whereas SDL’s platform carries out its analyses in near real time. An in-house data science team has a range of skill sets from structured analytics, regression and advanced expertise in natural language that supports the process.
Technological advances have made an enormous difference in the analytics that can be carried out. “Now that we have more computational power and bigger data sets, deep learning can be used to further enhance traditional analyses,” says Greg Council, VP of marketing and product development for Parascript, a company that provides image analysis and data extraction. “The concepts were there in the past but it was not practical before.” The best tools are the ones that combine multiple techniques, such as Bayesian statistics and neural networks, according to Council.
The more advanced the analytic techniques, the less likely it is that the customers will understand the inner workings. If concern arises on that front, Council suggests that the customer consider it from the viewpoint of any technology that is used without being understood. “You don’t need to know how an elevator works to go from one floor to another, but you want to know that it does what it’s supposed to before you get on.” So for black box applications, a good approach is to test it on known information and check its performance before deploying it, which in fact is typical for machine learning applications.
Big search capability for SMBs
Many enterprise knowledge management solutions are now within reach of small- to medium-sized businesses (SMBs). That accessibility is partly thanks to the cloud, which has eliminated the need for capital investment in large, expensive solutions by allowing a pay-as-you-go approach. But others are targeting the market with affordably priced products that may be either on premise or cloud-based. For example, SearchBlox is a search product priced at $5,000 per server that will handle 5 million documents and unlimited users within an enterprise.
Out of the box, SearchBlox is much more powerful than the basic search products were just five years ago, even though it is competitively priced. It identifies all metadata, whether e-mail (date, subject, sender), HTML (tags), PDF (document properties) or any of the other 30 formats it can process. That capability allows users to set up searches on any of those parameters to narrow their search.
SearchBlox also offers something that none of the entry level products did five years ago—text analytics. “We have been increasingly finding that after indexing their text, our customers want the ability to analyze it,” says Timo Selvaraj, co-founder and VP of product management at SearchBlox. With the amount of data pouring in, it is easy for users to figure out what information is in what location, especially if they have multiple data sources to manage. SearchBlox provides visualizations and insights in a variety of forms, such as word clouds, tree maps and pie charts.
“Search as we know it is transforming itself,” says Selvaraj. “Customers want to extend search so that alerts can be set up, thresholds identified for negative sentiment and real-time visualization provided that helps make sense of the information.” The boundaries between search and analytics products are blurring, and integration of functions is occurring within affordable levels for SMBs.