Multidimensional tagging, a key component in social sharing sites, can potentially help enterprises manage large stores of information. In this article, I'll examine the ways that multidimensional tagging will be implemented using Open Source tools.
As storage costs have continued to decrease, organizations create and retain more information. It is easy - and useful - to keep information online, just in case it's needed. The challenge has become managing information not simply storing it. Structured systems, such as relational database management systems (RDBMS), have well-developed tools for indexing, locating, and retrieving information. Unfortunately not all information fits well into structured systems. Graphics, presentations, spreadsheets, and word processor documents have all proliferated, no longer bound by storage costs. This is especially true of desktop systems where 200GB hard drives are common.
Finding exactly the information needed at the right moment has become difficult. A typical desktop drive has tens of thousands of files. A file server or NAS device could easily have hundreds of thousands or even millions of files. Even if system files and applications are discounted, that's still a lot of files to wade through.
File Systems and Search Engines Provide Some Relief
File systems impose a certain degree of order on unstructured information. By allowing information to be placed in hierarchical directories, file systems group files that share some relationship into meaningful categories. The downside to categorization using a file system is complexity for the user. The user is forced to descend through more and more layers of directories to find what he wants. Just remembering where the correct directory is becomes a chore. Information organization via the file system becomes less useful when information is spread across an enterprise. Incompatible systems, individual ways of building directory structures, and shear scale make this a difficult way to manage information assets across an organization, even a small one.
Search engines provide some respite from our information organizational woes. By indexing keywords and storing references to their source in a database, it's possible to look quickly through all the available files. The vendors of searching engine technology, drawing from vast experience in indexing hundreds of millions of Web sites, provide tools that let users find information spread across a huge number of desktop disks and enterprise storage systems.
Tagging Provides Necessary Clues
The trouble with the typical search engine is that the user must first know what he's looking for. He has to have some idea what keywords are indexed for a particular piece of information. If the indexed words don't match the words that a user thinks apply to the information, then the search engine won't find what the user is looking for. A marketing brochure, for example, may not have any words in it that say "marketing brochure" but that's how the user thinks of it. The way the information is categorized in the users head doesn't always match the strict keyword index of the engine.
This has already become something of a problem for Web sites that allow users to share large amounts of information. Cues are needed to help visitors who come to a site find what interests them without knowing the exact nature of the content. In typical Internet fashion an organic solution has arisen called multidimensional tagging, labeling, social tagging, folksonomy, or just simply tagging. It's multidimensional because a single piece of information can have many different tags, reflecting the different dimensions that users apply to it. Tagging lets users assign a set of categories to a piece of information when they create it. The tag system, also known as the tag cloud, grows as people use the information and see relationships in it that the original author might have missed. Users categorize information according to how they view the information, which makes it useful for groups of people who don't always think alike, such as engineers and marketing people.
Tagging is a key feature in social sharing sites such as Yahoo's del.icio.us and Flickr as well as You Tube, and Userscripts.org. Whether it's sharing interesting Web pages, photos, video, or Greasemonkey scripts, all of these sites rely on user categorization. Without tagging, no one would find anything of interest and the site would fail. Unlike simple storage sites (such as Yahoo Photos and Yahoo Briefcase), they require a way of presenting information to users that lets them find it quickly. Users can find information even when they're not really sure what they want. Social sites don't abandon search engines. Instead they integrate searching with tagging to provide a breadth of information retrieval options. Most let users search the tag cloud as well as scanned keywords, providing a rich search environment.
Tagging: New to the Enterprise
Tagging technology for the enterprise environment is new and not widely deployed in products. That's unfortunate. Not only is it extremely useful for finding information, it's also a natural way to do it. This is especially true for people used to social sharing sites. The tagging methodology facilitates the efficient sharing of information across many users and a large file space. It's exactly what enterprises need to make best use of the great stores of unstructured information on their corporate networks.
There are three ways that tagging is being implemented in corporate environments: integrated into applications; as a part of a standalone information management system; and, eventually, as a file system feature. The first kind of implementation is readily available. Image management systems, even ones directed at the desktop environment like Google's Picasa, include tagging as a core feature. The next version of the Thunderbird e-mail client (version 2.0) is expected to include e-mail tagging, augmenting its current search capabilities. Of course, once users get used to tagging for managing certain types of information, they will wonder why they can't use it for all the information that they need to access. They'll expect tag clouds that span all kinds of information in the enterprise.
Tagging is also being implemented in targeted information management tools. Tools for searching large stores of information in a corporate network are still at an early stage but tagging should be expected to make an appearance in information management and search engine tools in the near future. Consider this, Yahoo uses tagging in Flickr and del.icio.us as well as the upcoming My Web 2.0. Is it a stretch to expect it to implement tagging in the corporate search arena? The same is true for Google, which uses tagging on its eBlogger site, GMail Web-based mail service, and Picasa image management tool.
Finally, tagging can be expected to become a feature of the file system and operating system. Some aspects of tagging already exist in operating systems. The ability to attach keywords to files in Microsoft Windows is an example of file system-level tagging. These keywords are currently read by Microsoft's desktop search engine, creating a crude multidimensional tagging feature. Of course, entering and displaying tags is clunky and tags can't be displayed in and of themselves, rendering it more of a hack than a real feature. It does, however, point the way to future features of the operating system. Fully integrated into an operating system as normal metadata and using standard visual cues such those used on social sites, tagging will become a typical part of most corporate environments.
Tools Exist, File System Hooks Don't
The tools for corporate tagging capabilities already exist in the Open Source community. Most of it is encapsulated in the tools used by social bookmarking sites, which are often based on the LAMP stack. They're typically written in common scripting languages, such as Perl or Python, or Java. One such Open Source tool is unalog. Ostensibly a social bookmarking system, it's written in Python and the source is readily available on SourceForge. While the core tools exist, the hooks into the file system are still mostly missing.
A somewhat different but innovative approach is evident with Flickrfs or the Flickr File System. Based on FUSE, it creates a virtual file system with tagging for the Flickr digital photo management service. A fusion of file system and service, Flickrfs lets Linux users access the Flickr service as if it were any other mounted Linux file system. Photos can be accessed through the same tags available on Flickr using standard Linux commands such as cp. Flickrfs represents another way that tagging may come to information management - as a specific application or service but integrated into the normal file system.
Conclusion
Multidimensional tagging provides an opportunity to let users manage information more in line with their natural way of thinking. By sharing tags across the enterprise, users will spend less time looking for information and more time making use of it. Unlike other collaborative systems, users do all the work without legions of editors making decisions that users find mystifying. The social sites on the Internet have shown this to be a viable information management model. It's a matter of how and when, not if, these features become available to the corporate enterprise.
References
Subscribe to Open Source Magazine Email News Alerts
Subscribe via RSS
