IReader Uses Semantic Analysis to Summarize Linked Pages
Marshall Kirkpatrick
10 comments »
IReader 2.0 is a semantic analysis and link preview tool that launched today. Syntactica, the company behind iReader, hopes the browser plug-in will bring its technology back to life after a previous incarnation as a search engine was allegedly shut down by Google. If you’ve seen CoolIris, Browster or Snap then you’re familiar with the idea of a browser plug-in that provides a preview of the page behind a link. What makes the iReader unique is that the software analyses the text behind the link and gives readers a handful of bullet points intended to summarize the most important parts of that text.
Compatible with IE or Firefox on Windows or Mac, the iReader installs very easily. It can be turned off at any time with a right click or control click. The program’s performance varies at launch; some summaries are good and some aren’t. The two second delay between hovering over a link and seeing the small pop-up window is probably unavoidable but is sometimes longer than I care to wait. Of course seeing a pop-up as soon as any link is passed over is annoying as well, so perhaps what’s at issue is that this whole class of tools is unappealing to me.
A quick test of the semantic analysis provides some interesting examples. The iReader does a great job of summarizing OpenID.net - perhaps that’s its true calling, who else do you know that can summarize OpenID clearly? Hovering over the company’s own link results in text with bad enough grammar that it’s distracting. A Google News link for “Brittany Spears Checks Into Rehab” gets no summary at all - perhaps that headline speaks for itself.
I’ll give the iReader an extended trial and perhaps it can surprise me. I can’t help but wonder what else its semantic analysis could better put to use for. (See the previous post on Adaptive Blue for an example of semantic analysis put arguably to better use.) In most cases, the title, context and URL that appears for a link when I hover over it provides sufficient preview for me.
It will be interesting to see what kinds of business models beyond obtrusive advertising emerge for services like this. Analytics could offer interesting opportunities, but pop-up previews could lead to an unhelpful reputation for intrusion that would leave users hesitant to willingly expose their traffic and attention data. Give me a readable summary of the text of all the links I’ve hovered and clicked in a day and I might be interested.
See also Read/Write Web, where Richard MacManus gives iReader a more positive review.
I’m delighted to see technology like this become practical. Once the semantic Web is everywhere, we have the Web 3.0.
http://en.wikipedia.org/wiki/Web_3
Unless Larry Page was serious about the claim that AI is “not as far off as people think” in which case the singularity will cause the Web 3.0 to be irrelevant.
http://news.com.com/2100-11395_3-6160372.html
I found it totally annoying and less than helpful on most links I tried and it really needs to discriminate links, it will try to pop a summary on every link even nav, header and other page elements. The mouse-over is a problem as usual, would be better if it just added a symbol one could click when wanting a summary. You can turn it off but at this point I found it easier to just uninstall, didn’t think it was ready for prime time.
This inspired me to write a blog post:
“link previews (like snap) are completely backwards”
http://bla.st/site/blog/65/
To summarise: “This functionality should be built into the browser, not individual web pages.”
The best similar tool that I have come accross is Clearforest Gnosis that has a rich scheme to classify the type of text, and loads its classification in the sidebar.
I have tried the iReader software and it’s VERY annoying. Mike, please hold a poll about what your users think about page previews - do they consider them useful, or annoying? I would be surprised if more than 5% of the users like them.
Software Man — yeah, iReader can be annoying if you don’t adjust it to your needs. You can toggle it on and off by just right-clicking. And you can set sensitivity by going into tools/iReader or Options/iReader, entering how many milliseconds to wait before showing the preview window. So, if you really want to, you can set it to a second, or maybe several seconds, so you get fewer accidental popups of the window. I do some work for this company, and one of the developers told me they’ll soon have a control panel with slider controls instead of numeric input.
Am I the only one getting tired of naming convention rip-offs? Every new web site either starts with i, my, or ends in r.
iReader, iRooster, i-…
mySpace, my-xxx
flickR,mappR,zoomR
If everyone is so creative with these new sites, why can’t they think up an original name?
I’d like to see a really good semantic analysis tool that you could pump a blog post through and it would automatically come up with tags / keywords.
There probably is one, I’m just ignorant about it.
Thanks for the feed back, we need it to find the best way to make this
technology useful and not annoying. We are using your feedback to
this end.
Thanks,
Henry