We know the drill. When we have a general question, we turn to Google with an imprecisely worded query. More likely...
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
than not, we locate useful links that lead to the answer. The content is out there. Google quickly sifts the results to discover the needle in the haystack.
Yet, when it comes to work-a-day situations, we have very different experiences. Whether deployed in the cloud or hosted on premises, the needed content is maintained within our firm's existing servers and systems. We rely on various applications, databases and content repositories, each with their own methods for accessing and querying information. It turns out that knowing what and how to ask the questions is as important as finding the answers.
Missing is our ability to "Google the results" in an intuitive, easy-to-use fashion. The underlying enterprise content management analytics that pinpoints what we want to know is usually absent from our application environment.
Scoping CMS analytics
Content analytics encompasses an ever-expanding range of business intelligence and analytics tools, technologies and practices.
At its core, enterprise content management analytics adds structure to unstructured data to compute the connections among disparate sources and make sense out of data lakes and information streams. Content management system (CMS) analytics blends tried-and-true techniques for organizing and finding things with new approaches just coming to market.
Once deployed within an organization, content analytics promises to raise the bar on operating efficiencies and produce innovative solutions that encompass machine learning and artificial intelligence. But building these next-generation applications requires engineering discipline.
Text analytics and semantics
CMS analytics has its roots in text analytics. Think about the words in a document created using a favorite editing tool, such as Microsoft Word or Google Docs. The document contains text strings -- alphanumeric characters stored in a file. Transparent to end users, but accessible to developers, it also contains tags for formatting that determine character attributes (bold, underlined), paragraphs (breaks, indentation), and page layouts (margins, pagination). Algorithms use these tags to transform the text strings into readable experiences, rendered on screen displays and printers.
And tagging isn't limited to formatting. Through the virtuous cycle of metadata enrichment, it's also possible to tag text strings for meaning -- embedding terms and concepts directly into a file. Thus, individual entries within a business directory might include tags for "last name," "first name," "company" and "email address." End users automatically add these tags when they edit predefined templates. Algorithms process the tagged content, respond to queries and render results on screen displays and printers.
Some tag sets embed semantic knowledge about the interrelationships among different items. For example, Schema.org defines the web-wide tag sets for a range of commonly recognized "things"; it is designed to enhance the experience of search engines and provides the foundations for the Google Knowledge graph. As another example, the Financial Industry Business Ontology encodes the meanings of financial terms, concepts and related items of business knowledge. Algorithms can then produce smart financial services applications, such as automatically determining the credit-worthiness of loan applicants.
Moreover, text analytics doesn't always rely on explicit tagging. Natural language processing (NLP) recognizes patterns in text and assembles enough clues to make inferences about the relationships between words and phrases and thus determine meaning. NLP relies on a range of linguistic techniques such as text-sentiment analysis, entity extraction and linking, latent semantic analysis and Bayesian text classification, each relevant for certain kinds of document collections and business situations. Not surprisingly, it's a foundational technology powering many cognitive computing solutions.
Deducing digital patterns
But text is only one data type. There are other techniques for deducing digital patterns from various kinds of content -- including photographs, images, full-motion videos, audio tracks and 3D objects -- and then extracting the underlying meaning of things.
Take image analysis, for example. Raw images are collections of bits, organized as pixels and commonly stored in standardized formatted files, such as JPEG. Image analysis tools can process a photo collection to recognize faces; highlight emotions; identify colors; and classify the principle objects images contain, such as dogs, wild animals and fluffy clouds. When the files contain the geocodes, content analytics can determine locations.
The end result can be a next-generation digital asset management platform that automatically organizes a collection of branded assets, with men, women and children wearing different styles of clothing, walking in fields and cities, together with backgrounds from different places around the world. A marketing department can then enhance productivity by making it easy to find just the right photos for personalized marketing campaigns.
Getting started with enterprise content management analytics
Notice a familiar refrain -- deploying pattern recognition technologies with a purpose. As far as business users are concerned, smart applications deliver the business value. CMS analytics is the means to an end and depends on insights about how digital experiences enhance productivity, reduce costs and mitigate risks.
How can an organization get started with enterprise content management analytics? Consider the following:
- Know how to ask the right questions. Be task-specific and situation-aware. The more focused, the better. Rely on subject matter experts to model the problem domain. CMS analytics should encompass a well-defined model.
- Optimize for transparency. Make sure end users don't need to do anything outside their familiar, everyday tasks. Simplify the business environment wherever possible. The algorithms behind the pattern recognition should just work intuitively and be appropriate for the situation at hand.
Remember, enterprise content management analytics platforms aren't for the faint of heart. When done the right way, they can be configured to produce powerful and profitable results.
Predictive analytics help companies look ahead
Digital experience rules over web design
WCM, AI make for better digital experiences