More sophisticated would be a self-learning system that suggests tags based on current tagging patterns. Thus, if a user tags a document "foo," the application might return a message saying, "Eighty percent of users who tagged a document 'foo' also tagged it 'barr.' Would you like to tag the document 'barr'?" No such capability exists today.

Bringing it home

Usability also extends to the retrieval of bookmarked results. Optimally, those results should be integrated into the enterprise search engine and displayed alongside regular search results. While every product but Cogenz's Enterprise Edition returns bookmarked results within search results, only Connectbeam does so on leading enterprise search platforms Fast and Google. IBM and BEA use their own search engines, each of which has smaller market share than Fast or Google.

As for ranking bookmarked results, Connections, Application and Enterprise Edition all rank documents based on popularity: The more bookmarks referencing the page, the higher its ranking. But not all authors are equal, and treating results from experts and novices alike is deceiving. The same goes for the bookmarkers themselves: Experts should have more weight.

To address these issues, BEA has developed an ActivityRank for each item and user in the system. Administrators can adjust the relative influence of each rank factor, such as author's and submitter's ranks and the rank and number of people who've edited, bookmarked, tagged and viewed the document. Similar elements can be applied to the ActivityRank of individuals as well.

BEA's solution can also expire links from old bookmarks, forcing inactive, top-ranked pages from dominating search results.

Finally, all the products expose the tag cloud, revealing tags others have used. All but Cogenz's Enterprise Edition use bold and larger font sizes to indicate more popular tags. BEA officials say they are modifying the tag cloud to depict the relationship between content and user, a capability that's expected to be available in the first half of 2008.

Bookmarking price

It can be difficult to get a firm price on enterprise bookmarking systems, either because of additional software requirements or market pressures. To help get a better handle on the deal, eWEEK asked vendors to price out a two-site installation for 5,000 users.

IBM Lotus Connections costs $110 per user for the first year with a yearly maintenance fee thereafter. So expect to spend at least $550,000 in the first year for our 5,000-user scenario, making Connections the most expensive of the products. Then again, Connections includes more functionality than the other players. And, if you average out that one-time charge across the five social software tools in Connections, the price compares favorably with competitors'.

AquaLogic Pathways costs $40,000 per CPU, or $64 per user and $80,000 for our 5,000-user scenario. Once again, that's not the full story, as additional BEA software is required. At the very least, AquaLogic Interaction will be needed for security and searches.

Not surprisingly, the two startups do not require the purchase of sister software packages. Connectbeam charges a yearly fee with tiered pricing based on number of users, starting at $20 per user or $100,000 per year for 5,000 users. Cogenz charges $2 per user per month, or $120,000 in the 5,000-user scenario.

This article was originally published on eWeek.com.

Check out eWEEK.com's Messaging & Collaboration Center for more on IM and other collaboration technologies.