Tuesday, December 18, 2007

Video on Office Live Spaces

Scobelizer has an interesting video on Office Live Spaces. Nice video! I was wondering how this relates to Groove and Sharepoint? Is this just for "smaller groups"? Does anyone know?
Also refer to
TechCrunch's review.

Blogging transparency

Hope you all had wonderful holidays! Back to blogging!

Some time ago I posted on Jeremiah Owyang's blogging approach. I wrote a.o.:

What I was thinking though was: isn't it too bad that these draft posts aren't visible in some way from the start? I'm not saying every idea should be public (total transparency), even if that is possible. But in companies it can be very useful to know that somebody's thinking about something and working on an idea. The 150 draft posts show that Jeremiah has loads of ideas that are not really ready to publish, but eventually they will or several draft posts will be merged into one. I think lots of people can relate to this situation. And a blog is a wonderful place to work on, store and publish them. So, wouldn't it be nice to have a way to publish a draft post (just the title for instance) to give your colleagues (or the world) an idea of what you're thinking about?

Well at least Jeremiah liked the idea and sent all his draft posts to me. This wasn't quite what I meant. What I meant was to be able to publish draft versions of your posts on your blog. These are clearly marked as 'draft', but all that want to read it may and can also leave comments etc. In this way a post could be finished sooner and be more complete. Looking at Jeremiah's list of topics, I'd say the key words aren't enough to elaborate on.

To give you an idea I'll post some of my draft posts here. And I'll be doing that more next year.

1. Managing structured and unstructured information holistically

I'm looking for a toolset or architecture that allows companies to manage the complete product lifecycle, from idea to product to service. Tools to manage unstructured information are abundant as are tools to manage structured product information. But tools that can do both or support switching between both aren't.

2. Managing context of information

I'm looking for ways to easily and concisely manage the context of information objects in companies. To give you an example. For the attached document to an email, the email is very important. The email gives context to the document. It tells the receiver, e.g., what to do with it. Are their tools that allow me to store contextual information and/or add this information?

Happy Holidays!

I'm off on vacation. Going skiing! There's lots of snow in Switserland and the sun is shining. What more could you wish for?! Well maybe one of these "perfect snowball makers"!

I wish you all Happy Holidays and hope to see you in January 2008! I'll be celebrating my first blogging anniversary then!

Implications of the Latest PLM acquisitions

Interesting article about the latest PLM (Product Lifecycle Management) acquisitions and their implications. Although this article is focused on the automotive market, it has broader implications to other market segments. A couple of interesting remarks from the article:
You could argue that now everybody has a few less choices; the market has consolidated to a few major players. (...) PLM technology is becoming less of a factor. PLM depth of coverage, how rich the solution is, is becoming more of a factor.
And I think this remark about Dassault is most important. Being able to support multisite development, engineering and manufacturing is key. Plus being able to collaborate with non-engineers on one platform.
Squire says MatrixOne lets Dassault grant access to detailed product definition to virtually anybody in the extended enterprise, and to standardize business processes that use that information across disparate groups, including non-engineers and other people not necessarily using computer-aided design (CAD), digital mockup, or digital manufacturing tools.

The New Semantic Web wave (5)

I'll continue and round up my posts on "The New Semantic Web wave". In this post I'll comment on Alex Iskold's posts about the Semantic Web on Read/WriteWeb. The first post is titled "Semantic Web: Difficulties with the Classic Approach" and the second "Top-Down: A New Approach to the Semantic Web".

First of all: Alex, thanks for the insightful posts and the historical overview of the development of the semantic web.

Secondly, to my readers, please read Alex's full posts. Summarizing them for you would withhold you of a good overview. They're long posts, but well worth the read.

Now to my comments.

I really liked the pragmatic approach ("simple semantics") to the long quest for the Semantic Web (refer to figure). I agree with your "new approach" too. Along with the examples you give, I find that Twine and Powerset are applying this new approach, right? They're doing "simple semantics", not waiting for full language understanding by computers (in logic). Just do what you can, like identify names, places, structured queries, etc. For the coming years this "new approach" will keep us busy and will improve our web experience. I do wonder if this really is "new" though. It definitely is a "new wave" of semantic web.

My most fundamental remark has to do with what you say here:

The truth is that the understanding of natural language by computers is a really hard problem. We have the language ingrained in our genes. We learn language as we grow up. We learn things iteratively. We have the chance to clarify things when we do not understand them. None of this is easily replicated with computers.

I completely agree with these sentences. However, I missed what this means for your approach. Could you post on these more philosophical problems of the semantic web. For instance, will computers that are based on logic ("strict rules of logic"), truly be able to understand human language (and speech)? I would say: No, not as long as computers are based on logic. (Refer to Katherine Hayles, How we became posthuman and Paul Cilliers, Complexity and postmodernism a.o. for arguments on this "no".) Human language is partly logical, but also has non-logical parts to it. So when we talk about the semantic web, we're not talking about semantics in a linguistic sense, but in a more limited, derived and technical meaning. And indeed, if this is our definition of semantics, than computers can 'understand' the meaning or derived meaning of the text.

To my readers, by the way, a more recent post on Read/WriteWeb gives an overview of the Semantic Application to watch.

First experiences with Powerset

Some time ago I was invited to use Powerset, a new natural language search tool! I'll share my experiences with you.

The user-interface is wonderful! It’s clear, not clogged, lets you type in comments without moving away from the page, etc. Here are two screenshots to give you an idea.

When you start using Powerset you get to watch a couple of short video’s on how Powerset works, how you can use it, what the limitations of this version are, etc. Insightful, good expectation management. (By the way, why not let all be able to see those video's, while they're waiting for invites?)

To give you a taste of how Powerset’s natural language search works they’ve created a set of queries to illustrate the power of their semantic index.

The reference applications are limited to ‘speech’ (quotes), ‘business’ and ‘arts’.

There's an application called Powermouse, which gives us a glimpse into Powerset’s natural language index and lets us see how it extracts structured facts from open text. The index now only contains how things are semantically connected in wikipedia. Powermouse lets you explored these connections.
You see three boxes: in slots 1 and 3 you can fill in a person's name, places or things, the second slot is filled with a connection word. The relationship is searched for in Wikipedia.

In the 'Sports' application you can type in questions like:
"Who did the Chicago Bulls defeat?"

When you type in "defeated" it also finds results with "eliminated".

Search results are presented in a table. On the one hand you get the search results of Powerset, on the other hand, results from "old" search. The user may vote on who's best or say it's "undecided".

During search I was able to leave my remarks on search results and user interaction in an intuitive way. Entering ideas, comments, questions and remarks did not imply that I had to move away from the page.

By the way: querying the discussion forum containing ideas, questions, comments and remarks provided very concise results.

Well, it was nice to play with Powerset. As I said before, I wasn't too hopeful about this new natural language search. I would like to be. And I do hope I will be in the future. Basically, the results I saw relate to the kind of results we had years ago when "question answering" was hot. Looking at the results I see this is still a very difficult task. And these results were limited to structured queries and a limited data collection...

Nonetheless, I'm happy that companies like Powerset are trying to tackle this difficult task!
And, they are doing this in a new way. Using the social component of search. This could be the key to real useable natural language search in the future. By having the user vote and correct in a simple way, Powerset may be able to create a true new search engine. I truly hope so.

And, finally, a little joke. In the Business section I typed in "Who founded Powerset?". Result "no right answer"...

Web 2.0 Framework

Also ran into this older post (why didn't I see this before?!) on Ross Dawson's blog: the Web 2.0 framework. It gives a framework for Web 2.0, definitions of Web 2.0 and a Web 2.0 application landscape. Thanks for sharing!

Supporting structured and unstructured information processes

Wow, this is a really interesting post by Ross Dawson. I was trying to summarize it for you, but it's better you read it all!

It's related to the discussion about the 'enterprise software' being sexy or not. And about how ERP (Easily Repeatable Process) and BRP (Barely Repeatable Process) relate or don't relate. Some time ago I posted on this topic too, relating to the IT Flower by Innovation Creators here and here.

Relating to the IT Flower and my posts, I do find that the ERP (in Rinde's defintion) and the BRP worlds are distinguished too much in Dawson's and Rinde's post. I think one of the problems knowledge workers have, working in 'traditional' companies, is the daily switching between ERP and BRP or between structured and unstructured information processes. There is little or no support for this switching. The ERP and the BRP world in daily work is not distinct, but mixed.

What we do see though is the integration of e.g. Sharepoint with these tools for structured information processes. I don't think Oracle, SAP, etc will truly understand this subtle switching between 'structured' and 'unstructured'. This will come from others, such as Microsoft(?).

What do you think?

Friday, December 14, 2007

Waiting for...

Every knowledge worker sends out lots of emails asking people/colleagues to do something for you. After you hit 'Send' you usually forget about the request. The 'Getting things done' method says you should keep a list of those outstanding tasks. I try to, but it's hard to be consistent... There is an Outlook plugin to make this easier.

However, just recently, a colleague of mine (thanks Roel!), pointed me to an even easier (and free) solution to this problem. Just follow the guideline and you're all set!

Thursday, December 13, 2007

Automatically Classifying Unstructured Information

CMSWire posted that IBM updated it's Classification Module. It "automates the categorization of large volumes of enterprise information" and integrates with the FileNet P8 CMS platform. Well it doesn't automate it upfront: it gives suggestions and learns from corrections. You can find more information on this module here.
This is interesting stuff. Neatly organizing our structured information is taking up most of our time in organizations. But what do we do with all our corporate unstructured information (- assuming we manage this information centrally). This Classification Module can help us sort it out and check our taxonomy and classification.
One great use of such a tool is when a company worked on a certain topic, left the topic for some years and wants to come back to it. Where do you start reading and sorting through the information? Having this Module suggest a classification of that information could then be a big help.

I was wondering: Is this Module based on IBM Information Miner technology (in the past it was called 'IBM Intelligent Miner for Text')? And do they also use this to analyze their Jam sessions?

The Semantic Web Wave (4)

Scobelizer shared this first-time demo of Twine. While waiting for an invite I'll chew on this!

UPDATE Dec. 18, 2007: Just watched the video. I'm really looking forward to getting my invite. This technology would be very interesting to use in companies. I think the limited scope would make Twine even more useful. Is Radar Networks thinking about offering an enterprise-version of Twine?

Thursday, December 6, 2007

Beyond the Ordinary with Christmas...

Nice video on the O'Reilly Radar promoting the 'do it yourself' mentality and, in Océ terms, 'beyond the ordinary' with a Christmas flavor. Enjoy.

Saturday, December 1, 2007

Blogging explained

It's already all over the Internet, but in case you missed it: a new Common Craft video explaining blogging.