jump to navigation

Tagging vs. Searching May 23, 2006

Posted by Steve in : tagging, search / 1 comment so far

Two recent articles have helped me reframe my thinking a little regarding the relationship between tagging & search.
Quoting Tony Karrer quoting me:

I personally have found that because I’ve switched to Yahoo MyWeb that has full-text search across my bookmarked pages, I’ve come to use tags mostly to represent two things:

  • Actions - I tag items with “blogthis” if I plan to come back an write it up in a blog.
  • Sharing - I tag items that I plan to share with a specific tag so that others in my group can find it.

So for me, it’s not quite the folksonomy effect that most people talk about, but based on these articles, I’m starting to think that’s what other people are finding as well.

And from Bill Ives’ Where Tagging Works and Where Tagging Doesn’t Work:

If I want to search on a key word, I will still go to Google as the most efficient way. If I have the time to go exploring through multiple links and see the interrelations between key words, I might go to del.icio.us. However, if I want to set up a way to store and share links on a particular topic, I will use del.icio.us which I have done already in co-authoring an article.

A lot of people (myself included) have tended to more or less equate tagging with “del.icio.us”. Delicious doesn’t search well, and google doesn’t tag well. That creates some kind of artificial distinction between “tagging” and “searching”. As a result, people get distracted trying to demonstrate the (re-)findability value of social software (perhaps because search is so generally useful that impact on search has become the easiest measurement of value?)

But when you use a service like Yahoo MyWeb the two are no longer so separate. That scenario better demonstrates the (IMHO) best uses of bookmarking/tagging:

Search benefits from these but (in my opinion) that really isn’t the intended purpose of tagging. And search isn’t the only point of integration that will benefit from the addition of social features…

Tags:

New Google products and their impact on the enterprise May 12, 2006

Posted by Steve in : enterprise, search, bookmarking, google / add a comment

It’s probably worth taking a quick break from enterprise del.icio.us and mentioning some recent “products” that are very relevant to enterprise bookmarking and information sharing:

Google Search History

Since early 2005, Google has (optionally) been automatically saving all of your searches, if you’re logged in to a google account (gmail, etc.) Automatic saving provides desktop-search-like “Web History” indexing to the internet product… it also provides them with all sorts of data for later use in personalization features. And there’s one other side effect: it gives you an easy way to gather and annotate results from your past searches. Items can be starred, after which they will appear in a Bookmarks section, and they can be tagged with keywords. I have read that search history can also recommend similar pages, but this isn’t showing in my history. Overall, the implementation of this feature is actually pretty slick, but not well promoted so I doubt many people are using it… yet.

Google Notebook

This feature is taking direct aim at del.icio.us and friends. After notebook is enabled, all Google.com search results pages have a “Note this” link that allows you to bookmark them with a single click. These links are saved in Notebooks, and notebooks can be exchanged with other people, but I don’t believe they can be tagged per se. Still, this is an important addition because Google has recognized that people are basically creating collections when they bookmark, and they’ve provided tools to do so without the complication of tagging. Even more important is the browser integration that Notebook offers through firefox and IE plugins. It complements Search History by providing a way to gather and annotate results directly from the search UI and result pages, saving you the step of going into the History. I don’t know whether the stars from search history will eventually be consolidated with notebook’s notes.

Google Co-Op

This one is sort of the wildcard of the bunch. Co-Op consists of two parts: the ability to create and/or subscribe to OneBox templates, and the ability to “label” websites with tags defined in those templates. This is a very ambitious project which will result in thousands of new OneBox templates being written to easily expose information in lots of sources; the main winner will be content producing sites who can expose their interface right at the top of Google’s search results!

This post by Philipp Lenssen does a good job clarifying what’s going on, although in my opinion it is slightly inaccurate in that he doesn’t make the distinction between what happens out of the box vs. only after subscribing, and he also overstates how much search federation and clustering is really happening.  Also, labeling is clearly not an end-user play when you have to upload XML files to tag something, so for now it exists only as a way to refine a OneBox template.  [* see below for my response…]

There’s already a lot of progress towards this in the enterprise product; in fact it seems that Co-Op may have been based on that work, with the subscription feature having been added to prevent spam and OneBox overload. My guess is that there will have to be at least one overhaul of the subscription feature before CoOp reaches popular use - it’s already being overrun with junk. This isn’t a concern in an enterprise product, where all users can be automatically subscribed to appropriate, sanctioned OneBoxes.

The takeaway for enterprise use:

Just as Co-Op emerged from the enterprise product, there’s no doubt that the other two features will be integrated into their enterprise and desktop search products, and probably sooner rather than later. Having added all three, their enterprise or desktop search product would make great inroads into the personal benefits provided by an “enterprise del.icio.us”, with the added advantage of that functionality being directly available in search – no need to switch to a new application.

Anyone building an enterprise bookmarking application is on a direct collision course with Google… (and not just bookmarking)

* Updated - here is my response to Philip’s Co-Op article, extracted from his comment section:

Hi Philip, I think you’ve done a great job of explaining the complex Co-op product, but there are some distinctions I feel could be made clearer:

First, Co-Op is basically two components –
1) the ability to build a OneBox element, and
2) the ability to label URLs in support of clusters
But when it comes down to it, these are really the same thing… (more in a bit)

Second, Co-Op is visible in two ways –
1) What you will see out of the box, and
2) What you get after subscribing
It’s important to make this distinction because frankly, it will be a very tiny % of Google users who will ever be subscribing to CoOp templates, given its current form.

The ability to make a new OneBox is really powerful and will be very helpful to sites with lots of content – you can basically expose your entire site through an element at the top of Google! But there are two huge limitations that will keep this from popular use: #1 – it’s not really search federation, since as you’ve demonstrated in your example you basically have to send over your entire entity list in the template XML file. This means you have to do scheduled updates, etc. #2 – No one benefits without subscribing, and I suspect only very techy people will subscribe to things for now. There are some things they can do to make this easier for end users and I’m sure it will improve over time, but right now it’s targeted squarely at high-volume sites.

Clustering and labeling is just junk right now. It’s basically a specialized type of OneBox which allows you to define a topic & vocabulary, and then support it with XML labeling files. This isn’t “tagging” in the popular sense, as again is not targeted at end users. Rather it is done this way so that the OneBox “moderator” (whoever created the cluster) can distribute his work among many publishers. It also faces the #2 problem above – no one will benefit from clusters until they subscribe. (At least in this case they can automatically enable certain clusters for all users after they’ve been reviewed for quality)

So I guess I would have chosen different words than your bullet list, explaining how this will help publishers build OneBoxes & specialized clusters, rather than mentioning the masses, who won’t yet benefit much. (I do think the feature will improve over time of course!)

related: google search history

related: google notebook

related: google coop

Tags:

What’s the personal benefit of social bookmarking?

Posted by Steve in : enterprise, search, bookmarking / 5 comments
Page [1, 2, 3, 4, 5] of this series:

One of the common claims I hear about the value of bookmarking sites like del.icio.us is that “bookmarking benefits the user first.” It’s often said in support of the Del.icio.us Lesson (even if they don’t use that particular term) and the corollary is that after people find enough value in personal bookmarking, they will go on to share information, get recommendations, etc. The individual benefit, in this case, is “re-findability.” Since I’d like to make sure our product reaches widespread enterprise adoption, I’m trying to make sure we have a good personal use case before tackling the social part. And I’m curious: does social bookmarking really offer the benefit of re-findability?

The reason I ask this is because I never use tag search. It’s partially because tag search often doesn’t combine with regular search. Del.icio.us doesn’t index the contents of pages, just the title, tags and notes you’ve saved. This means you have to remember exactly how you tagged something; that’s not always an easy task (and it’s probably worse for a beginning tagger, who likely tags with a few very general tags.)

Even assuming a service that stores and indexes full page content, such as furl or Yahoo MyWeb, I still have a more reliable helper when finding an article I know I’ve seen before: I turn to desktop search. It almost always returns the right thing much faster than I could have found it on del.icio.us, and searches through more content repositories. Better yet, it turns up things that I hadn’t thought to bookmark yet… This is incredible return for zero investment.

If tag search doesn’t work well for me yet, perhaps the list of saved items improves my ability to re-find articles? I’ll admit that I originally joined the service in search of a way to easily synchronize bookmarks between my work and home machine - clearly a personal benefit. However, when I look at my own habits since adopting del.icio.us, I find that I don’t ever use my saved links as bookmarks. For instance, there’s a webpage with a great CSS guide that I revisit a lot. Yet I don’t get there by going through my del.icio.us “css” tag. Why not?

First, getting to my saved links and clicking through takes longer than I’m willing to wait for a very common operation. This could probably be alleviated with a browser extension, although the best Firefox plugins I’ve found don’t solve this problem well yet. The other problem is that I’ll often save new CSS links, pushing my favorite link further down the page, until eventually it’s on page two - at which point it might as well not exist any more. ;) This too could possibly be prevented through some sort of new “sticky” feature in an extension. But really, in the context of a potential enterprise application, it’s a bit optimistic to imagine many people using Firefox, much less extensions.

So if social bookmarking a) isn’t better than desktop search for (re-)finding articles, and b) doesn’t offer the best way to visit my favorite links, then why am I such an active bookmarker? Because I’m often marking up collections of items, usually without highlighting any one article over the others. If you look at the use cases for furl, for instance, you’ll see people saying basically the same thing. Sometimes I’m making the list for myself, more often I’m doing it with the intention of sharing. Sometimes the list is even being automatically created for me, which I’ll talk more about in a later post.

As I see it, this is (part of) a need that everyone has, which isn’t already best served by browser bookmarking or desktop search, and that’s what makes it work learning a new behavior and/or new application. So I don’t see our application as being for social bookmarking as much as it is for social list-making. To me, such an app isn’t really tasked with re-finding, except in the general sense of being able to survey all the items I’ve collected in a particular list. Maybe this is a bit of a very fine distinction… but in my next few posts I’ll explain why I think it’s an important one.

First, I’m looking for feedback. How do you use social bookmarking sites? Have I overlooked a way in which bookmarking benefits you personally?

Then, in my next post: does this really require tagging?

Tags: