Vision of the semantic web

Michael Marshall in evancarmichael.com explores some fascinating examples of implications of the increasing use of the semantic web. He explores the issues surrounding fraud, SOE and spamming, associated with tagging. Expect this to become a real issue soon as where new developments go, not far behind is someone trying to make a quick buck.

I was also very impressed with his view on tagging and truth/fact and its relative perception. The semantic web will rely on a sub-level of information providing an additional layer of information (through tags) in tandem with with web page information. In effect this information, or tags, tell the computer what the page 'means'. Michael Marshall makes the point that who is going to inspect the tags to make sure they are correct. This is very relevant to the measurement of online media.

While we should never expect perfect information; the hope is the semantic web will permit more accurate, intelligent searching and data interpretation. In the world of media measurement could this lead to computers undertaking more of the qualitative tasks we currently reserve for humans? This helps illustrate some of the challenges for auto-evaluation, and I hope undermines those who say they currently have an automated process for undertaking qualitative measurement.

Getting a pipe on social media measurement

Since discovering Yahoo Pipes a couple of months ago it has really been off my to-do-list. Whether I am progressing is debatable but there is a feeling that the bits are starting to form a shape, if not necessarily fitting together easily.

Following the news from the head of the stream I have been playing with Pipes to create an effective monitoring capacity, with some success. On the face of it it seems simple enough but the truth seems a little less clear. There are a selection of examples other have put together and as the general instructions are not so good it seems best to reverse engineer one of these.

In a future post I am planning to concentrate on one of the Pipe apps, Open Calais semantic tagging tool. Its a brilliant concept and with time I am determined to make use of it.

The other element I want to use is a Google Docs spreadsheet, which I understand can receive RSS feeds. Going right to the end of the 'food-chain' I also need some sort of admin. This might need to be custom derived using something like PHP / MySQL or Drupal(all about which I know very little).

The one thing I am determined to do is blog about the experience as these new tools and are all free and very current I hope it will useful information.

Doing the Zurich Test

Is the semantic web now becoming a reality. Sir Tim Berners-Lee has said all the tools are in place, it just needs people to collate the data and make the applications. As mentioned yesterday we have been trialling one of the recent tools developed by Reauters called Calais.

The techno language bit is a bit baffling to a non-techie like me but I have understood enough to run a few queries on what I call challenging texts. Would it be able to distiguish between company names, places and people? The image above is an extract from the output and I must admit I am pretty impressed. Although it failed a few times yesterday, on this extract from the FT it was able to distingussh between Zurich the financial services company and the Swiss town - pretty good!