Data of the People, By the People, For the People

So maybe I am a day late and a dollar short, but I recently started an account with the online bookmarking service, Delicious. A latecomer to Web 2.0, I started the account because of the difficulty I experienced managing the many bookmarks in my browser. You can see my Delicious account here. If you look on the right-hand side of the screen, you'll see the number of tags that I have entered to date. "What are tags?” I'm glad you asked.

In Web 2.0 and other computer applications, tags consist of keywords assigned to a data file that describe the corresponding file content. Tags are a form of metadata that is useful when searching across a database. These data artifacts are used in a variety of applications; for instance, I am creating this blog entry in MS Word, and have assigned metadata to this file (which I viewed by clicking “File”, then “Properties”). It looks like this:

In this graphic, you can see some of the metadata is assigned to this file, which may become useful if I need to find this piece at a latter date. In this instance, the keywords identify the content. While I don’t often use this function because my database is small, it provides utility to a company involved in identifying and managing content across one or more organizations. The result is that specific tags can be searched within a database to facilitate retrieval of similar content. Doing so requires that the tags be organized in a way that allows them to be easily found.

Organizing systems of information can be approached in various ways. One source I read cited the Dewey Decimal System as a familiar system for establishing a hierarchy for the information contained within your local library. In the biological sciences, taxonomy of the known species of flora and fauna organizes our knowledge of the plant and animal kingdoms. In the same way, tags placed as metadata on database files can be used to classify and organize the files in a database. By the way, clicking on the tree in the link below gives rise to specific instances of biological classification – go to this link and click on the frog, for instance . . . the Tree of Life is an interesting Data Visualization.

Web 2.0 applications similarly use databases to organize tags, but their use figures more prominently in the functionality of the site. Hierarchy is established by folksonomy, where increased the frequency of tag use by the viewer community elevates the tag in site’s tag hierarchy. This can make for some other interesting data visualizations – Delicious has established a tag folksonomy for my personal account, which uses text color and relative size in a Tag Cloud to denote my most frequent tags. It looks like this:

Delicious’ site-wide tag cloud, as of the wee hours of October 21, 2009, looked like this:

Looking at this tag cloud, it’s easy to see that “design”, “blog”, “video”, and “software” are the most frequent terms used in tags by the Delicious community. This hierarchy is established by data the user’s entered, and implies that that these are the topics that Delicious users are seeking on the Internet. It’s a pretty solid statement about what is important to people. It is data of the people, by the people, for the people.

These hierarchies are not without imperfections. Anything derived from the actions of people is subject to human error – such as when I was looking for information on wrench design and entered the term twiting force instead of twisting force – see my personal tag cloud above.  As such, errors can be incorporated in the database, although this one should get buried at the bottom of the tag cloud pile, unless every other web-surfing mechanic is as bad a typist as I. This illustrates another Web 2.0 principle - the applications get better with increased use. As other tags are elevated in the hierarchy, the impact of my error will become essentially nil.

Other problems occur in folksonomies.  One of them has to do with the idiosyncrasies of the language used for tagging.  In English, we have synonyms. When I searched on lemon, two date-ordered bookmarks that came up had to do with automotive defects and lemonade, two results that are clearly unrelated. 

Searching on bush included items about the former U.S. presidents and a development project with bush in the title, but nothing about plants. See below:

So people aren’t perfect and neither is our language. Other aspects that make folksonomies less than perfect are discussed in this interesting paper by Adam Mathes. But the imperfections do not negate the power of folksonomies – knowing what your audience is thinking can be an important and powerful tool.

In business, reviewing a folksonomy against a firm’s own marketing goals could increase its success. For example, a software giant ready to introduce its Next Big Program could review a folksonomy on its site and find the most common searches are crash, corrupt and security patch. These descriptors could be indicative of the desired qualities for new products as well as an existing quality and public relations problem. A chain of auto repair stores that pushes brake and oil change services can review a car repair site and find “check engine light” and “state inspection” are the most popular tags assigned by site users. How might that inform their marketing plans going forward? A political organization could use folksonomies to gauge the interests of the public. Opening a tag cloud is a whole lot easier than accessing and reviewing website logs on user habits, and is more meaningful because the data and language are input by the collective usership.

So Delicious got me thinking about folksonomies and audience analysis, which resulted in the piece you are reading.  Good technical communication is all about accurate audience analysis. Folksonomies offer a feedback mechanism like no other - Data of the People, By the People, For the People. And imperfect or not, I cannot think of a more direct method for gauging audience interests, a key to creating A Better Message.