Pixelated Semantics

A schizotypical inventory

BOB the non-Bomb meets Newsbot

The "bomb-scare" that forced an airliner to return to Sydney last night may have resulted from a worst-case interpretation of a note scrawled on a sick-bag. The "note" consisted solely of the letters "B-O-B" and was interpreted by the pilot as "bomb on board". However, according to the ABC, the acronym is more commonly used to describe "best on board" - that "that someone sitting in a particular seat... looks good".

Running "B-O-B" through MSN's beta news search verbatim produced little by way of result, except for the ABC source, and a completely unrelated item on the British Labour Party that happened to feature acronyms that were hyphenated. Syntactically related, not semantically. The release of yet another non-semantic search engine tends to confirm that the major "engines" have become more concerned to control information than to freely disseminate it. MS explains:

"Who decided to feature this story and photo? The computer did it! [...] Newsbot uses computer algorithms to determine which stories and photos to feature on the website. Newsbot uses MSN Search Technology computer algorithms that take account of many factors, such as how many sources are covering the same story, when the story was published, who published it and how many people have looked at a particular story."

In other words, MS decides the relevance by way of the criteria it applies to the Bot's weighting of results. "The computer did it!" is pure escapology - it takes a human after all to instruct (program) the computer to make those decisions. By making publisher a criteria it becomes very simple to exclude sources, and elevate others, using corporate values rather than end-user. Wrapping the service in pseudo-geek PR does not disguise the implicit ability to bias results. Of course, there is a presumptive caveat: "as we all know, computers aren't smart enough to really understand all our human idiosyncrasies -- so please don't take offense..." - or in other words, perhaps this is for the mainstream who prefer not to think too much about the information they are presented with.

Later Musings: several hours after the first post, there are still only a handfull of references for "B-O-B" though this was the biggest story in Australia today. Presumably the Bot does not crawl and trawl with great frequency. It occurs to me that MS has missed again the direction of today's infotech developments: the future of the net "browser" as its currently being shaped is not entirely WWW, its a search-based interface which offers the flexibility to read RSS, HTML, RDF metadata and more, to launch applications, and also be capable of useful things like relational client-side topic maps of the information gathered during online sessions. And there definitely are semantic search engines available that would appear to be much smarter in their linguistic department. Information maps and models stored client-side can be used to leverage searching. But Newsbot does not even match the functionality of the annoying search built into the current IE. There is no reason why search criteria such as source and relevance cannot be configured and sorted by the user - client. Simply providing a very basic form-based search in a web page is nowhere near good enough and does not innovate - in fact it draws back. Newsbot is adequate - for 1995, barely.

Item posted by AutoEditor at 10:28 am ::