Phil Bradley's weblog

Harvester 42: integrating meta search engines

Harvester 42: integrating meta search engines. Straightforward multi search engine, covering about 50 different resources divided into 16 different categories. Results are inframe, so you just scroll down the page to see the next frame.

YOHOGO! - Multi Search Engine

YOHOGO! - Multi Search Engine. Well, it's a multi search engine, and you know what these are like, right? This one has tabs for the web, jobs, news, images, videos, music, torrents, social, reference and er.. adult videos. It's got a reasonable number of options available - 15 different engines in the web category, and the same number in images. Interesting it's providing access to Exalead images, but not web results.

My main gripe is that they pop results into their own frame, and overlay it with adverts. As a result I get to see only the top three results from Google for example, and have to start scrolling to see the rest. This is not good, and for that reason 'I'm out' as they say on Dragon's Den.

Struly multi search engine

Had a chance to play with Struly (as in 'yours truly) today. This is what they say about themselves:

"STRULY (rhymes ‘yours truly’) is a search engine. A multi-search engine to be precise.

But why would you want to use a new search engine? Well, Struly isn’t anything new. And most of all Struly isn’t an attempt to wean you away from your favorite search engine.

For the same effort/time it takes you to search Google or Yahoo, Struly throws up results from both these as well as other search engines…and in the process shows you some exciting developments in the search industry."

It worked well; I was impressed with the speed. It's basically a tabbed engine, so if you search the web you get Google results displayed for you, with tabs to move to Yahoo, Bing, Carrot2, Wikipedia, YouTube and Flickr. There are also search options for Images, Videos, News, and Torrents. If you're looking for a 'quick flip' set of results, this isn't at all bad!

More on KartOO

You may recall that I blogged about KartOO the other day and ended by wondering how results were chosen, and in what order. Angelica Marchand has kindly filled me in on the details:

Let me explain briefly how the online reputation and K.Visu work (Our search engine follows this principles too).

E-Reputation
Using text mining technology, the applications analyses web content (blogs, forums, websites, etc)  and defines whether they have a positive, negative or neutral connotation. The analysis is based on a dictionary that has predefined words but also a learning system with Bayesian filtering to continously enrich and update the vocabulary.Our customers can define the sources of information to consider (Bing, Yahoo, etc) and if needed, the keywords to look for, in addition or in substitution of the custom dictionary.

K.Visu
The KartOO technology allows to represent a list of objects (Web pages, documents, database info...) in the form of a visual map, each connected by links (common words, database criteria, or a customer's dictionary). The result is a 2D map on which the objects that have the most common points are brought closer. Information is thus organized in two dimensions contrary to the classical lists, which impose a linear course. The more documents within a topic, the greater the surface; topics with close content will be place next to each other.

For example in your map, LILAC has more key words in common with Library than with Services. The size of the surface indicates the amount of information available, thus Professional and Managers have the largest size.

Another function is targeting keywords, some customers want to find certain terms, so we include them in the main dictionary and when a search is conducted this terms will be highlighted if they come up in the results.

Thanks Angelica - that was most helpful.

Kartoo has a makeover

The visual multi-search engine Kartoo has had something of a makeover recently. It used to employ a genie type figure, with results appearing on the screen as larger or smaller webpages, with links between them to indicate complimentary information. That's all gone, and we're presented with a new, slicker search results screen, which although it works on the same principle is much easier to see and use. Screenshot below:

 

There are various options you can use to change the appearance of the screen; news, sponsors and so on. A mouseover reveals a little more information about the particular resource you're looking at.

Kartoo has also added a new resource, based as best I can understand it around e-reputation though how this works the demo doesn't explain. However, my lack of knowledge apart, the resource works quite nicely. The image below gives you a hint of it:

 

This is the result of a search that I did which started with librarians. I then clicked on a link in the image element of information, and then conference. Seemingly good results, but alas, still lacking in any more information as to why those results were chosen, where they came from and so on. This is clearly still a work in progress, and if anyone from Kartoo wants to pop in a comment to explain in more detail, that would be grand.

Panabee

Panabee. I'm liking this. It's a classic concept with a slightly different twist. Run your search and Panabee displays a split screen of results from 2 different engines. (During the search stage it also provides suggested search terms.) However, you can then choose to display results from various subject areas such as images, shopping, reference, real-time, gossip and so on. In each category you can then choose from various different engines - in the video section for example is the chance to search from 14 of them.

The nice thing is that you can have one pane displaying results from your favourite image search engine, while the other displays web results for example. It's a little like creating your own merged or blended set of  results. Nice, simple fast and effective.

Yebol for knowledge based smart search

Just had a chance to take a look at Yebol.com and I really like it. 5 search options; Smart, Web, Images, Video and News. The results page is quite busy, with related topics down the left side created on the fly (and they work), categories, expanded searches, Twitter postings and adverts on the right. In the centre we have links to top sites, adverts, news and search results. So lots of stuff, and you need to spend a few moments looking through the page to see exactly what you want to do next. No advanced search option, no help screens and no RSS option, which is a shame.

Their mission "is to build human-like world's knowledge base and provide knowledge based search (semantics) and services. Yebol allows for a multi-dimensional search result instead of the normal one-dimensional search seen by most web search engines today.  This provides a more accurate summary of top sites and categories; a wider array of related search terms; a longer and richer expansion for query results; and a deeper base of links and keywords in search result pages.

Worth trying out!

Startpage Metasearch is the new Ixquick

Ixquick, one of the old timer multi/meta search engines has undergone a makeover, and renamed itself to Startpage Metasearch. It looks just like Ixquick used to look however, so no difference there. It's a good search engine, and one of its main claims to fame is that it doesn't record IP addresses.

It's currently pulling results from 12 other engines, including All the Web, Ask, Bing, Cuil, EntireWeb, Exalead, Gigablast, Lycos, Open Directory, Qkport (no idea on that one, will investigate!), Wikipedia and Yahoo. No Google though you'll notice.

It also has a 'Phone' option, which is actually a people search for addresses and numbers. There's a video option, powered by Blinkx. Amusingly, while Bing is carrying on something chronic about their mouseover function, Startpage just shows the videos in the thumbnail, though admittedly and to be fair, you just get a few seconds of clips.

There's also a picture search, which is probably the poorest of the options, with a limited amount of background data.

Advanced search is catered for, as there's an advanced and an expert search page with even more options, but nothing amazingly exciting that you won't already be aware of.

Oddly the Ixquick site is still up and running and isn't redirecting, which seems a little odd. If you've not tried the engine, it's worth giving it a go.

Fefoo multi search engine

Fefoo is a neat little multi search engine that provides access to over 250 different search engines. Straightforward interface as you can see below:

In most cases I found the choice of engines reasonable, although there were a few surprises - no Exalead in the websearch selection for example. When you run a search Fefoo loads the appropriate search engine into the main element of the screen, leaving room at the top for little ribbon search element:

Nice and easy to use, with a few command line switches for specific engines or types of search.

PolyMeta - Intelligent MetaSearch and Clustering Engine

PolyMeta is a search engine that someone mentioned in a recent blog comment, and I went along for a look. It searches in the areas of Web, News, Images, Twitter, Video and Blogs. It pulls content from Ask, Bing, Cuil, Google and Yahoo. This could be expanded out to include another 3 engines through their 'sources' option. Choices differed according to search option chosen, so News included content from Topix for example. It also provides a total number of results from each engine together with a click through to go directly to the search engine site.

Content is gathered from the first page of SERPS (which given that Bing has a larger default for the number of results on the home page would seem to give it a slight bias), and then arranged in some way that made no sense. I did a search for a hashtag for a specific conference, and the top result had been found by Cuil, second result had been found by Ask, and the third result by Ask, Google and Yahoo. Unique results are highlighted, which was a nice feature.

Information was displayed in a triptych with clusters/sources on the left, results mid screen and images on the right. There's also a 'did you mean' option which it's culled from Google.

The web search was ok, but having recently looked at Scoopler, it was a little lacking in currency, and didn't hold up any better when I actually checked out their Twitter tab. News was ok, but couldn't hold a candle to the more specialised Silobreaker. Image search pulled content from 10 sources, but was quite lacking in detail - just the image and source, with no further details. The video option was also ok, but again not as good as my preferred option of Trooker. The blog option only pulled content from Ask and Google, which was disappointing.

In summary I think that Polymeta almost makes it, but misses out all around - it's just not quite good enough at any single thing to pull it out of the masses, and doesn't have any kind of unique feature to overcome its other limitations.