Best way for users to explore non-visual content

Question

I'm working on a project that offers a library of content to choose from. The challenge is that the content is purely textual. My users will roughly know what they want, but they'll also want to browse.

Let's say, for example, that my site was focused on pre-made letters to grandma. As a user, I'm too lazy to write my own letter, so I want to browse a library of content and choose one that is pretty close to what I want. From there I can tweak it.

A traditional approach would be for me (the programmer) to brainstorm what categories most letters to grandma will fit into. But that's tough - content could expand by tone (e.g. "Lovey-Dovey" vs. "Formal"), purpose ("Update on my family" vs. "How are you"), etc. As the programmer, I'm not sure I could come up with a category tree on day one that wouldn't have to be retrofit constantly.

An alternate idea would be some kind of tag cloud. This way as new content is added, it sort of categorizes itself. My team will be in charge of generating content for the most part, so we can sheppard content into existing tags when needed. If we do let users create their own content, we can require approval of new tag creation.

I'm leaning towards a tag cloud idea, or perhaps a multi-mode tag cloud like http://cssglobe.com/tag-clouds-styling-and-adding-sort-options/. I'm open to different/better ideas on how users can browse this non-visual content. Obviously a gallery of thumbnails won't work, so I'm open to new ideas.

Answer 1

I do like your tag cloud option and seems like a good interactive way to get information across. However the challenge with auto tagging is that an letter can getting tagged incorrectly (assuming the example of letters to grandma).

E.g. The letter might read like this :

Dear Grandma

Hope you are doing good. I am doing well though I am not happy as my pet hamster died.

Now the tone of this letter is going to be sad but the tagging could assume it was happy due the word "happy" in the letter. Unless you can find a way to automatically analyze sentiment (which is not an easy task) your tag cloud could give a lot of false results. I would recommend that you do some brainstorming with your team and ask them to come up with a list of categories under which the letter would fall (e.g. Happy,sad, reflective,funny,cute) and present it in a simple list format like this :

You can also encourage user participation by asking people to tag the letter with a limited set of choices like how yelp does it

This tagging can also ensure that a letter tagged incorrectly will be moved to the appropriate section.