When you hear the words metadata, your eyes might be tempted to glaze over. Don’t let them.

While not as sexy as predictive analytics, enterprise search and other advanced data mining techniques, metadata management plays a key role in the world of big data. Need proof? Microsoft today said it plans to file a formal complaint with the European Commission against Google because the search giant won’t share – you guessed it – metadata.

Specifically, Microsoft contends that Google will not make metadata connected to YouTube video’s available to its Bing search engine. That means Bing users may not see relevant YouTube videos show up in search results. Not only does this negatively affect consumers, Microsoft contends, but it also hampers competing search engines from attracting advertisers.

Microsoft also says that its inability to access YouTube metadata means its own YouTube app for Windows-based mobile phones “is basically just a browser displaying YouTube’s mobile Web site, without the rich functionality offered on competing phones,” according to a blog post by Microsoft’s top lawyer, Brad Smith.

Metadata, at the heart of Microsoft’s complaint against Google, is essentially data about data, as Wikibon defines it. It includes the date and time a piece of data was created, the author, keywords related to its contents, and so on. When commercial search engines mine the web looking for content, it is metadata they are often looking at.

Metadata, and managing it diligently, is equally important inside the enterprise as it is on the web for a number of reasons. First, enterprise search tools – tools that help business users find relevant documents, emails and other content within the workplace – use, among other things, metadata to determine and display search results. Like commercial search engines, enterprise search tools mine metadata to find content relevant to keyword searches. But they organize and display results in a much different way than their web-based cousins, and metadata again plays a central role.

Search engines like Google determine search results partly through metadata that describes how popular a piece of content is. The more clicks a YouTube video gets, the higher it will end up in Google’s search results (though not perhaps in Bing and other competing search engine results thanks to Google’s metadata hands-off policy.) The number of clicks is recorded as metadata.

In the enterprise, however, the relevance of a piece of content is not determined by its popularity, but other factors including the author and who else has viewed the content. A piece of content is likely more relevant to someone in the accounting department, for example, if it was created by his or her boss and viewed by colleagues that have similar job functions – all important metadata. Another piece of content that is very popular in, say, the sales and marketing departments is likely less relevant to the accountant, even though it has garnered more downloads or views throughout the company.

Second, data analytics engines use metadata to help business users gain insights from their data. Data that reveals, for example, how many customers viewed a marketing campaign online during a certain time period, are forms of metadata.

Some metadata is created automatically and some manually. It is the manual metadata creation that can often be tricky. For example, when a business user creates a new document, depending on the application or system involved, he or she may be responsible for tagging the content with metadata, often including descriptive terms. So organizations must decide on uniform metadata definitions and models, and train workers responsible for inputting metadata to do so in a standard way. If not, enterprise search results and data analysis will be skewed. Inaccurate and incomplete results and analysis can also result from missing metadata, as Microsoft’s pending complaint against Google illustrates.

[Cross-posted at Wikibon Blog]