Phil Bradley's weblog

tuSavvy - social search engine

This is a rather nice new engine - tuSavvy which I've been playing around for a while and it's quite impressive. It works along the 'what other people have found and bookmarked' line instead of searching the web; they describe it as "We are aggregating some quality URLs and public bookmarks someone already takes the time. With those, we generate search results from tag index and community factors, not link structure" (As you can tell, it's not an English language based group that's created it - it apparently originates with Korean entrepreneurs.)

As a result of pulling data from different sources it's coming up with some really unusual, but accurate results. As usual I did a search on my name (what, me ego search? Surely not!) and early results were from sites such as Pageflakes, Slideshare, Zimbio, and Rollyo. All places that I've put material and identified it, but not stuff that most other search engines have turned up. In addition there's also all the usual web based material that I'd expect to see. A mouseover on any result results in a small popup allowing me to email the result, bookmark it or 'Save to mybox'.

There's a social element that, while not yet fully emergent is hinted at. I can have buddies, create projects and import bookmarks. My feeling is that it's going to be the kind of engine that will work well with a group of people working together on a search project, pulling in content and sharing it in a central resource.

It looks intriguing and is an engine that I'll be coming back to.

Social Mention

Social Mention  is a social media search engine that searches user-generated content such as blogs, comments, bookmarks, events, news, videos, and micro-blogging services.

It allows you to easily track what people are saying about you, your company, a new product, or any topic across the web’s social media landscape in real-time.

Search results are aggregated from numerous popular social media sources, including Google blog search, Twitter, Delicious, FriendFeed, Flickr, Digg, YouTube etc. and remixed as a single stream of information. The data is fresh, which means you can track conversations as they are happening in real-time.

In addition to web-based search results, Social Mention also features email alerts and personalized RSS feeds for automatic and instant updates.

At least - that's what it says on the website. It's a good engine - very quick and easy to use, with clearly presented results. One of the things that I really enjoyed, which is very silly, is that while the results are being prepared the engine provides you with some hysterical messages -

Accelerating textual scanning
Recalibrating data parser
Scaling back-end networks
Engineering open source bandwidth

and so on.... Well, I think they're amusing anyway - I was doing searches just to see the messages!

delizzy - del.icio.us Bookmarks Search Engine

This is a nice idea. Basically delizzy  will grab all your delicious bookmarks, create a small little universe of the content and allow you to search through it. With delicious itself you can only search title, tags and descriptions, not the entire content of the page. Now, as long as you can recall bookmarking the page and something on it, you have a better chance of finding it again.

Of course, if you use FURL you'll know you've been able to do this for years, so quite why it's all of a sudden a big story I'm not entirely sure.

Delver - a new community based search engine

I've played with Delver today, which has a slightly different concept of search. It's based around social search, and it's really quite impressive. The idea is obvious - you'll like what your friends like (which may or may not be true) and so you will get more from searching if you can see what they're talking about - which of course only works if they are talking about stuff.

Before you can run a search Delver wants to know all about you - but it doesn't require password access or anything like that - it simply wants to identify you. So you can give it your name (great as long as you're not John Smith, but even with my name I was taken aback as to how many Phil Bradleys there are!) and other details, like email addresses, or you can go further and provide it with profiles at places like delicious, Facebook, LinkedIn and so on. When it's done that, it creates your community for you, which looks rather like this:
If you're sharp eyed you may see one or two people that you know, or they might even *be* you. Once this community generation has taken place you're ready to search. Stuff that is found will not be what you expect, if you run the same search on a regular basis, because Delver takes more account of your community, and also how you describe yourself. It's easy to change the results that you get by changing your own description - if I call myself an 'Internet Consultant' then I get a bunch of results from people who also call themselves that, while if I call myself a librarian, the results differ again.

In order to get the most out of this search engine it's really going to be necessary to fine tune your profile and community. I added in details on my Flickr account for example, which is going to skew results in a personal, rather than professional capacity. Delver is also going to work or fail individually based on how active your friends are in their own communities. However, it was fairly easy to tweak, but I have to wonder how many people are going to want to do that.

Is Delver going to make a success of this? My guess is not, and it's down to the registering and tweaking. Most people are simply not going to bother. However, the concept is good, and all it needs is for a search engine to take up the concept - a search engine that already logs you in, that has a note of what you search for, which blogs you write or comment in and so on. If Delver doesn't get bought up shortly, or one of the big engines doesn't embrace the concept I'll be quite surprised.

Stumpedia goes where Cha Cha failed

 

Stumpedia, which describes itself as "a social search engine that relies on human  participation to index, organize, and review the world wide web" has now started to provide what it terms 'instant answers'. These are not the same kind of small potted answers that you get with a search engine like Ask however, oh no. These are questions that you can ask that are routed through to a human being. One of many I should point out - I don't think it's one poor soul chained to a keyboard.

However, which person or persons this may be, we have no real way of knowing. You can ask a question and Stumpedia then utilizes a resource called Muchobene to pass the query onto someone who may know the answer. "The Muchobene plug-in allows you to share your knowledge with other users by responding to their questions."

I asked a fairly straight forward question - 'when was the American Civil War' and after a few seconds up popped a chat window and someone (no idea who) gave me the answer 1861 - 1865. Well yes, but not great, though to be fair, after a few seconds they came back with a fuller response. Which was vaguely correct, depending on how pedantic you feel like being. You can see this in the screen shot.

Flickr Photo Download: Stumpedia instant answers via kwout

Well yes. Except no. No citations were offered. I am left with no idea who the person is who answered my question. I have no idea if it was right or wrong. This is just plain insanity, and quite frankly I can't see this working at all. It's really nice that people are prepared to give up some of their time to help others but really - go and add something to Wikipedia if you want to do that. I couldn't use this information with cross checking it, it's really unusable in this format. Cha Cha tried it for a while and they've now dumped that in favour of mobile answers. It didn't work for them and with the best will in the world, I can't see it working for Stumpedia. Of course, the only exception I can quickly think of is where the person at the other end of the chat gives you a URL or other reference that you can check; my second query 'is there a crystal mineral that forms long, flat fragile shapes' was answered rather better, since I did get a URL I could use. However, I still remain to be convinced.

Can Social Bookmarking Improve Web Search?

Can Social Bookmarking Improve Web Search?. This is something that I discuss on most of the courses that I run nowadays - if you want to get a different view of search results, don't rely on traditional algorithm-led search engines.  Instead, start looking at the  results produced by (shock horror!) real live human beings. Now, I don't mean the social  search engines,  as I'm still far from convinced about their  value,  but instead, think about using something like delicious. 

Did you know that you can type in a URL such as http://del.icio.us/tag/librarians and get a listing of pages that have been tagged with that term? If you want more terms, just whack in a + between each word like+this and you'll get some useful stuff. It'll be useful because when people tag pages they want to remember them, so the pages and tags will pretty well relate well to each other.

Anyway, I'm pleased to see that some work has been done in this area by Paul Heymann et al for a conference paper. I've not been able to read the entire thing, but the abstract is very useful:
"Social bookmarking is a recent phenomenon which has the potential to give us a great deal of data about pages on the web. One major question is whether that data can be used to augment systems like web search. To answer this question, over the past year we have gathered what we believe to be the largest dataset from a social bookmarking site yet analyzed by academic researchers. Our dataset represents about forty million bookmarks from the social bookmarking site del.icio.us. We contribute a characterization of posts to del.icio.us: how many bookmarks exist (about 115 million), how fast is it growing, and how active are the URLs being posted about (quite active). We also contribute a characterization of tags used by bookmarkers. We found that certain tags tend to gravitate towards certain domains, and vice versa. We also found that tags occur in over 50 percent of the pages that they annotate, and in only 20 percent of cases do they not occur in the page text, backlink page text, or forward link page text of the pages they annotate. We conclude that social bookmarking can provide search data not currently provided by other sources, though it may currently lack the size and distribution of tags necessary to make a significant impact."

My emphasis there of course, and I'd certainly agree with their findings.

Twing - Search and Discover Communities

Link: Twing "aims to be the most comprehensive and highly targeted forum search engine for real people seeking real discussions and real information in real time. And of course, the information we index - content-rich discussions in numerous forums worldwide - is provided by real people. As a result, what Twing offers that traditional search engines don't is the ability to more completely identify important information in the deep discussions enabled by the many-to-many communication format of forums."

I've played around with it, and it's not at all bad. Lots of results, which can be arranged by date or relevance. I would like to see an indication of which and how many forums Twing searches - that information may be available on the site but I couldn't see it. Nice option to limit results with additional terms, categories, filters (forum, people, company, language, media) and a date range.

Worth taking a look at if you're having difficulty getting the information you need elsewhere.

Topicle Search Community

Topicle is the latest in the line of create your own search engines. However, it's based on the Google version with the slight twist that people can edit and comment on the engine, as well as suggesting new URLs. However, these features alone don't make me want to change from using the straightforward Google version.

Stumpedia - social search engine

Stumpedia. This is another social search engine, where people add their own links. They've only recently launched and have about 1,200 links. There were 2 for 'internet' for example, so there is a long way to go before it becomes useful. I really do think that if you're going to launch something like this you really do need to have seeded the engine with thousands of results to begin with.

inSuggest - Personal recommendations

inSuggest - Personal recommendations. This is very nice. There are two options, both of which work in the same way; web pages and Flickr images. I'm rather more interested in the Web side of things, so that's what I'll look at. You type in a URL of a site that you like to start with (basic URL seems to work best, it gets a bit confused if you go deep into a site). This then sits in a 'stack' at the top of the page.  It then produces a list of other pages that are matched to that one in the form of thumbnails produced by thumbshort.org. Next you skip through a few, and when you find another that takes your fancy you click and drag it onto the stack (sounds like a card game!) and the search is refocused with more offerings.

You can do this a total of 4 times to get some very precise matching. My main complaint is that sometimes you're shown pages that you can't actually get to because they're forbidden, which is irritating.  However, I liked the look and feel of this, and it would be very useful if you're stuck for sites to use. Quick and intuitive - give it a go!