Phil Bradley's weblog

Twurdy for Readable Results

Twurdy Search - Search for Readable Results. Interestingly this is a 'Tw' search engine that for once has nothing to do with Twitter. It's a play on the words 'too wordy'.

"Everyone has different reading abilities. Some people searching the web are university professors and others are 5 year old children. Twurdy has been created to provide people with access to search results that suit their own readability level.

Twurdy uses text analysis software to "read" each page before it is displayed in the results. Then Twurdy gives each page a readability level. Twurdy then shows the readability level of the page along with a color coded system to help users determine how easy the page will to understand.

Twurdy's goal is to provide web searchers with information that is most appropriate for them. This will mean that 10 year olds doing school assignments don't have to click through difficult material to find something they can use. It will also mean that phd students do not have to click through websites designed for kids in order to find what they are looking for."

OK, that's their blurb. To be fair, it's pretty accurate as well - a search leads to a summary of results, with a background colour of light orange through to very light orange - the darker the shade, the more difficult the page is to read, or rather - the more the page will appeal to an older or more experienced reading group. I found it very difficult to match the particular shade to the reading key, because the differences are very subtle. I'd have preferred more distinct colours, or to have a page of results ranked from easy to read to harder to read. I tended to use the 'Readability level' instead, which I found much more useful. It also doesn't help that the reading key scrolls off the screen as well, which I think is a basic design error.

To give you an indication of 'readability levels' the BBC Homepage has a readability level of 539.  The News Front Page however has a level of 633, and the CBeebies stories and rhymes is 512. Slightly odd, and while I couldn't see any information on exactly how they arrive at these figures I did find this in the FAQ - "The structure and nature of some pages will mean that Twurdy's readability level calculation is less useful than in many other pages."

Nice idea, and useful for teachers, school librarians and parents perhaps, but I'd be happier once it's got some more fine tuning under its belt.

Related articles by Zemanta

• Twurdy.com - Having Access To Readable Search Results (killerstartups.com)

Duck Duck Go adds new feature

The strangely named search engine Duck Duck Go has added a 'quick peek' feature. It addresses the problem of having to hunt through a page to find the reference to the thing you're searching for. To access, just hover over the magnifying glasses next to results. In
real time, they:

1) Go out and grab the current version of the page.

2) Turn it into paragraphs of text.

3) Look for the most relevant paragraphs given the search terms.

4) Highlight those paragraphs in the preview window.

It's an early release and currently works best for informational queries and on informational sites, e.g. Wikipedia, review sites, news sites, blogs, etc.

Works well - see the screenshot for my hubble telescope search:

Timeline of Search Engine History

Timeline of Search Engine History. Nice little visual history of the development of internet search engines, which hits the major highlights. May be useful if you're running an introduction to the internet.

Zakta - Home

Zakta - Home. Fairly standard search engine with a personalisation approach. Register, contact friends, search, vote for results, write your own guides, share them with friends. Looks fine, but the guide that I looked at hadn't been updated in 5 months (and it was about Twitter) and the whole site seems to have a slightly musty aroma about it.

The problem with any of these resources of course is that you have to get a fair number of people using them and the majority just can't be dragged away from Big G to make enough of a difference.

Search engine interfaces then and now.

Popular Search Engines in the 90s: Then and Now. Fun listing of a lot of search engines - what did they look like then, what do they look like now. Short answer if you don't have time to view the resource - then they were all different, now they all look the same, pretty much.

Navisso Search

Navisso Search. This is an interesting search engine that I discovered thanks to AltSearchEngines. Their main value (according to them) is the thing that I like the least, ironically enough. They say: "Navisso is an independent web search engine that (unlike many other popular search engines) does not rely on external search feeds. Our focus is to build a reliable, content-driven searchable index while reducing spam and other undesirable pages within our results... We are interested in building a true search engine by providing accurate, real, and relevant search results to our visitors. That said, we will never push any results to the top, unlike many other search engines."

This is supposed to ensure that 'official sites' don't get top billing, because they're not always the best. This is true, and I've certainly seen times when a fan site is better than the official one, particularly with films or books. However, one of the problems facing is the searcher is 'do they want the best resource or the most authoritative?' The truth is usually somewhere between the two, and I think this is where Navisso falls down. My martin luther king search didn't return any of the key sites that I expect to find, though the racist site was actuallly on the first page - clearly a case when an unofficial site, put together for devious means gets a better ranking than the good stuff.

So what DO I like about it? The functionality is good. Their 'OR' option is slightly strange, since it's {cat dog} but you can do a wildcard search based on * (librar* for library, libraries etc) and there's a nice unknown character ? and they also provide options for nested logic.

Still in its early stages yet, so the database isn't that large, but it's an engine I may well revisit at some point in the future to see how it's developing.

XAPHOO Search - Home

XAPHOO Search. is a fairly basic search engine, but it works well enough. It provides access to the web, images, video, news, sport, audio, wikis, PDFs and something called 'sweepstakes'. There's a basic advanced search function (limit to specific sites for example) and an option to limit a search to a specific country. The results are good enough, with a thumbnail option, an email link and a 'quick look' to load the site into the search engine window.

It does what it should do, but in all honesty I can't find anything particularly exciting about it.

SimilarSites.com

SimilarSites.com - Easily Find Similar WebSites. This one works in the same way as all of the other search engines in this category. You type in a site that you like and it'll find well - similar sites to that one. (However did you guess!) Works well, nice and clear, easy to see the results, which are presented with a thumbnail to the left and brief summary to the right. You can vote sites up or down.

It's nice enough, but not good enough to knock my favourite, Web Insuggest off its perch. Of course, if you don't like either of those you can try Similicio.us or Who is like it?

Hunch - a decision search engine that works

I really wasn't sure what to expect when I looked at Hunch as the home screen asked me where I lived. It then asked me a bunch of other questions about what I liked, was interested in, age, sex and so on. I dutifully answered the questions (didn't take long, no more than a couple of minutes) and then Hunch told me that it knew enough about me to start being helpful.

You can look at one of the preset questions, or search for other questions to help you out in decisions you have to make. Let's take one (almost) at random, and I'll walk you through it.

'Where should I live in London?' By a pub/nightlife? North or South? Tube zone? Happening or Quiet? Tube line? Quick answers to these then gave me Wimbledon, Battersea or Dulwich, with a 'wild card' of Richmond. Interestingly I've lived fairly close to ALL of those locations, so you can't get much more accurate than that.

I answered questions on British comedies (again, spot on) and SF films (pretty much a failure), and then explored the service, and said that I wanted to make a decision on the search engine I wanted to use. Wild card answer came up with Exalead which was pretty good. Also tried 'which URL shortening service should I use' and it got that right as well.

If you have a decision to make it's worth trying this out - and it's quite fun as well.

Search engine comparisons

There are a few resources that have popped up in the last few days that allow you to compare searches from the major engines in a blind taste testing exercise. The first one that I came across was Blind Search. Run a search and get a three column set of results - simply choose the one set of results that you're happiest with. You'll then see the engines popped up afterwards; Bing, Google or Yahoo. You have to keep a note of the engine that you like the best, since a tally option that the site originally had has now been removed.

The second engine, which just displays two screens from Bing and Google is called Bingandgoogle which doesn't score much for originality, but is very accurate. Bingle does exactly the same job and you cannot really tell the two apart; the only difference being that you can switch results entirely to one search engine or the other.

Of course, if you don't like any of these you can always try Tripleme which does a search across Yahoo, Google and MSN Live (which I'm guessing is pulling results from Bing).

GrabAll allows you to compare two pane results for Google, Yahoo, Ask, Bing, DMOZ, AltaVista, Gigablast and Looksmart.

Scour lets you check out results from Google, Yahoo and Bing with a nice sort option.

Searchboth will compare results from Google, Yahoo, Ask, Dogpile, MetaCrawler, AltaVista, Looksmart and Websearch again in a two pane window approach.

Soovle takes a different approach by providing on page access to Google, Wikipedia, Amazon, Answers, Yahoo, Bing and YouTube, Type in your keyword and Soovle comes up with alternative terms - click on any of them and your search will be run on the appropriate engine.