Monthly Archive for April, 2008

Can Zemanta help you write better blog posts?

Published
by
Tom Raftery
on 13/4/2008
in semantic web
. 9 Comments

Zemanta is a really cool Firefox plugin which scans the content of your blog post as you are writing it and suggests related content!

I first heard about Zemanta when I met Jure ÄŒuhalev, Zemanta’s Community Manager, at the BlogTalk 2008 conference here in Cork earlier this year. When Jure explained it to me I was intrigued and interested to try it out.

Subsequently I met AleÅ¡ Å petič, Zemanta’s Managing Director, at the Plugg conference. Having heard a lot about the plugin, I downloaded it to try it out and I have to say I am impressed, especially since the recent release of version 0.2.1.

The changelog for this release is:

- introduced WordPress 2.5 and new Wordpress.com support
- introduced FireFox 3 Beta 5 support
- increased the number of suggestions to non-wikipedia sources
- tripled our index of related articles
- we also started adding our users to the articles index

In the screenshot below you can see that for the post about Microsoft’s Live Earth Zemanta suggested images including Microsoft logos and Virtual Earth screenshots. It found articles and blog posts about the new release of Live Earth and it suggested related Tags and Links (along the bottom). Zemanta also pays close attention to copyright, making sure that suggested content is licensed as Creative Commons or approved by stock providers, so you won’t get into trouble by using Zemanta’s service

Zemanta Firefox plugin

The fact that Zemanta is a Firefox plugin, as opposed to a WordPress plugin is quite clever as it means that it only needs to be installed once and it works across the multiple blog sites I write on. It works on WordPress, WordPress.com, Typepad and Blogger.

I know Zemanta is using some Semantic web technologies so I asked how the plugin works and received the following reply:

Following concepts come handy when trying to understand the engine: disambiguation, entity extraction, hierarchical classification, information retrieval, machine learning, cross-domain background knowledge.

That didn’t help me much but maybe it will be helpful for some!

The Zemanta plugin still needs a bit of work. When writing this post, for example, I had to save the post as a draft and open it again before Zemanta started suggesting content. However, it is very useful when you are writing a blog post to be able to see other sources of related material so I predict a bright future for Zemanta.

Heading to EnergyCamp and Interop

Published
by
Tom Raftery
on 4/4/2008
in Conference and Green
. 7 Comments

EnergyCamp and Interop and on in Las Vegas at the end of the month.

Via James Governor, David Berlind very generously extended an invitation to me to come along to these conferences.

EnergyCamp should be very interesting as it will be the first unconference around Green IT issues as far as I am aware.

Interop also has a Green IT track which has a session on Green data centres. Given the work we have done on this in CIX, I’m really looking forward to this.

PERM_FAILURE: SMTP Error (state 14): 550 Email blocked by ORDB

Published
by
Tom Raftery
on 4/4/2008
in Technology
. 3 Comments

My incoming email was unusually quiet yesterday morning. Great, I thought, I’ll get some work done!

However, later in the morning I realised that anyone tryint to send me email was receiving the following error:

PERM_FAILURE: SMTP Error (state 14): 550 Email blocked by ORDB - to unblock see http://www.example.com/

A quick chat with a few people, including Ross of Rozmic (providers of EmailCLoud - my upstream anti-spam solution) and some Googling told me that my mailserver was mis-configured.

ORDB were one of a number of free blacklists of email spammers. Mailservers could query incoming email against their blacklist and accept or reject email based on the response. However ORDB shut down in 2006.

My mailserver was setup in 2007 but still had a configuration whereby it checked all incoming mail against ORDB (or tried).

Recently, it seems, the blacklist servers were reconfigured to list every ip address as spammers. This was probably to get mailserver admins to once and for-all remove references to ORDB from their config!

When I realised this was where the problem was I went to my mailserver configuration file (Exim.conf), found the following lines:

# deny using ordb
  deny message = Email blocked by ORDB - to unblock see http://www.example.com/
       # only for domains that do want to be tested against RBLs
       domains = +use_rbl_domains
       dnslists = relays.ordb.org

and deleted them.

Sure enough the mail started flowing once more. If you tried to email me in the last couple of days and couldn’t get through, apologies - please do try again.

Enterprise wikis reviewed

Published
by
Tom Raftery
on 1/4/2008
in review and wiki
. 28 Comments

Long story short, PBWiki is yer only man.

Why?
I was chair of the it@cork Conference organising committee last year. The committee is made up of volunteers who are all busy with their day-to-day jobs so getting times to meet which suit everyone is always challenging.

To help with the organisation I rolled out a private wiki where we posted meeting minutes, kept track of action items, posted to-do lists and updates on (potential) speaker status’. The wiki was a hosted PBWiki and it was a great success.

It was then decided to rollout a wiki to facilitate the organisation and running of all it@cork committees and subcommittees. The wiki software needed to be able to:

  • handle multiple wikis (one per committee/subcommittee)
  • handle user and group permissions
  • give stats around wiki usage and
  • be cheap or free (it@cork is a not-for-profit)!

I searched around and discovered that Atlassian’s Confluence enterprise wiki product has a free community license specifically for not-for-profits. PBWiki, and SocialText didn’t have any mention of a free Enterprise class offering on their sites so I didn’t really pursue them.

Boy was that a mistake! The setup of the Confluence wiki was far from straightforward. It took two of us the best part of a day to simply install it. Remember that as I was doing this for it@cork, this was not billable time. I was installing it on my own server and because Confluence requires TomCat as its webserver it had to run on a separate port to Apache. This meant several people couldn’t view it in their organisations.

Worse though was that once it was successfully installed, it was a disaster. Uptake and use of the wiki was minimal because the UI was appalling.
A couple of quick examples:
- a simple task which requires many steps, adding (or removing) users to/from groups goes like this:

  • Go to Administration
  • Scroll to bottom of page and click on Manage Users (why not dynamically have the most used menus at the top or failing that use general stats to move most used features to the top, or have two (or more) columns of options so no scrolling is required)
  • Click Show All Users (why? why aren’t they all listed by default?)
  • Click on the user
  • Click on Edit Groups
  • Select Groups and click Join (or Leave) - there is no feedback to tell an admin that this action has been saved

Editing groups doesn’t appear to be possible at all i.e if I want to remove 7 people from a group containing 25 people, I have to go into each user profile (using all the steps above) and remove them individually. It should be possible to do it in one simple step from within a Group Admin page.

Confluence backend

Other problems with Confluence included the fact that there are almost no stats around use of the wiki available and there is a bug in the users and groups which meant that the permissions applied to the groups were not percolating down to the members of those groups. A pretty serious bug I’m sure you will agree.

I Twittered recently that I was looking to get off Confluence and onto another Enterprise wiki platform and within minutes Ross Mayfield, Chairman of SocialText had contacted me (despite being on vacation!) offering me a SocialText wiki for it@cork. Yes please I said (poor guy, I nearly bit his hand off!).

The backend of SocialText is far cleaner, simpler and more intuitive than Confluence.
SocialText wiki backend

However, for completeness sake I also contacted PBWiki and when they heard this was for it@cork, they immediately offered business edition wikis at no cost. This was spectacular news as it@cork were already familiar and happy with PBWiki.

The PBWiki backend is the cleanest and simplest of the three (it is also the only one which fits vertically on a 1024×768 screen).
PBWiki backend

Not only that but when you compare the edit screens of both SocialText and PBWiki, PBWiki definitely comes out ahead in usability.

So we have a winner - PBWiki.

What surprised me in this is that when I Twittered looking for an alternative to Confluence, not one person suggested PBWiki. I have no idea why this is. PBWiki has both Enterprise class functionality and a UI which is completely non-intimidatory. I predict no issues with uptake by users this time round.

By the way, I did also try out DekiWiki and Twiki but I ruled them out quite early on.

UPDATE: Zoli Erdos posted a response to this post where he highlighted a couple of points I didn’t make clearly enough here:
1. I compared a non-hosted version of Confluence to hosted apps like PBWiki and SocialText. However, hosted Confluence is NOT available under the community license. As this was for a not-for-profit a community license was all we could go for. PBWiki and SocialText had no such restrictions.
2. I tried out DekiWiki but the lack of an ability to create Groups ruled it out quickly.

Tom Raftery’s Social Media is Digg proof thanks to caching by WP Super Cache!