Back in the mists of time, a couple of Stanford PhD students called Larry and Sergey came up with an idea to make searching on the web better. This they undoubtedly did. But skulking in the low-rent dens of Silicon Valley and elsewhere an army of wannabe moguls work tirelessly on usurping the Mountain View giant.
Whether for niche searches or for the whole web search prize, whether by human-powered or visual methods, whether by developing cunning semantic algorithms or by Social means, all these platforms have one thing in common: they all want to take a bite out of Google.com's posterior.
Visual Search Top Dog:
Quintura provides contextual visual search, and has long been touted as a genuine contender for the next-gen search king crown. You enter a search string or hover your mouse over a tag in the visually represented cloud. The cloud drills down and offers you more tags to hover over, and drills down again until you have reached the tag you want.
High-profile converts like ReadWriteWeb.com and the Maxim
stable of websites may just bring Quintura into the mainstream. Others:
Kartoo, a visual meta-search engine, has attracted more than its fair share of attention. Supposed to make search 'fun', or something, by lobbing in a Flash interface for your search map. Takes too long to load, but on the plus side boasts a feature-rich interface and page previews, not to mention a genie riding a surf-board.
'Search The Tail takes Google a step further by allowing users to narrow their search results by using either popular keywords or obscure terms. The keyword list, namely Google Tail, is in order of popularity.'
Whatever this is, it looks like nothing you've seen before. It's like Michelangelo and Steve Jobs got together to do visual search. If there was ever a website that's gonna help you to pick up chicks down at Starbucks just by looking at it, you've found it. So that's good enough for us then.
Apart from being stunning-looking, you can target search to stuff like stocks, news etc. Well worth a look
Exalead is a French SE which provides thumbnail page previews of results, and shows various extra information on the results page including maps, file types, categories, rich media and so on. Apparently you can also use it to cheat at crosswords.
Human Search Top Dog:
Mahalo is touted as 'The First Human-Powered Search Engine', and it's here to help. Set up by Weblogs co-founder Jason Calcanis last year, the Mahalo community filters out garbage links and provides you with quality pages and multimedia for your searches. Popular topics have their own 'Guide Notes' with all the essential information you are likely to need (eg here
). Of all the current human-powered Web search alternatives, Mahalo is probably the most likely to succeed in the niche, given exceptional growth over the past year. Others:
Trexy is a meta search-engine which relies on the 'Search Trails' of users to streamline the search process. If someone has made a particular query before you, and had to wade through several resources to get to the information required, then that's information that is useful to you. This search tool provides you with this information, so you can in theory avoid the mistakes made by those coming before.
To see how the trails work, look at this query for 'how did Columbus discover America'
Trexy generated a lot of hype soon after launch in 2006... and promptly fell off the radar.
Clustered search results. 'Instead of delivering millions of search results in one long list, our search engine groups similar results together into clusters. Clusters help you see your search results by topic so you can zero in on exactly what you’re looking for or discover unexpected relationships between items.'
True Knowledge's 'Answer Engine' aims to streamline the process of getting a specific answer to your question. It does this by 'structuring data in a way that enables computers to work and think like humans do, drawing inferences and conclusions when needed to find the information that's requested.'
This thing is in Private Beta, so we can't tell you how good it is. But you can already use its API
to utilize the platform in your Web services. In fact there are already a small army of specialized API services for stuff like IP Geolocation and 'phone-number-to-location'.
Members essentially vote on documents which relate to a specific search term, so you should get good results. Unfortunately, Stumpedia's Web search has a long way to go -- we tried 'Google history' and 'computer virus', and both returned the dreaded 'No Results Found' message.
But there's better news elsewhere. A recent new addition is the 'Instant Answers' service, which simply connects people to the Stumpedia community, with a view to answering queries. You use a sort-of IM client to connect with people, ask them a question, and wait for a reply. .
Xoost is one of the front-runners in the burgeoning 'social search' scene, which overlaps in a significant way with the 'human search' movement. You enter a search string as per normal, and get your regulation results. But... your experiences in finding relevant information are saved and shared with other users, and so the system learns how to hone in on better results. Go to Xoost to grab an invite to the Beta, if you're interested.
ChaCha invites you to ask any question in plain English from via Cell phone SMS or over the Web. Currently in Beta, ChaCha relies on human 'Guides' to answer questions, and then send the answer back to you via SMS/Web.
Semantic and Natural Language Search Engines
A Note on Semantic Search: Semantic search relies primarily on content being in RDF format, which is a particularly rich descriptive format. One drawback to Semantic search is that results take aaaaggggeeesss to get back to you. These things have to do a lot more drilling to get more accurate information from the deep recesses of the Web. Top Dog:
Microsoft recently opened up its fat wallet to buy Powerset
, a search engine which enables natural-language queries. The move ended lengthy speculation surrounding the company's future, and put to bed persistent rumors that Google was willing to break the bank to acquire the start-up. VentureBeat
explains the attraction:'Powerset, of San Francisco, has developed a technology that attempts to understand the full meanings of phrases you type in while searching, and it returns results based on that understanding. By buying Powerset, Microsoft is hoping to close the perceived quality gap with Google's search engine.'
At the moment, PowerSet only works its natural-language magic on Wikipedia articles. Like this
Hakia is one of the most serious players in the Semantic Search space, and up to Dec '07 Hakia had raised $18m in funding
. Hakia's quest for quality results hinges on three simultaneous criteria: that " It (1) comes from credible sources (verticals) recommended by librarians, (2) is the most recent information available, and (3) is absolutely relevant to the query". Services
include Web search, embeddable search, Web services and an API are available. In public Beta.
Swoogle is a Semantic Search research project run out of the University of Maryland, and has a logo and name which is likely to attract some stern legal correspondence with a Santa Clara, CA, postmark if it ever goes commercial. As well as browser-based Web search, 3rd party Web services can utilize Swoogle's crawl.
Semantic search powered by RDF Gateway. And Web services with sample agents and a Web service interface
to get you started.
'For computers, our site provides a web service with similar capabilities. Intelligent software agents submit precise queries to our service describing the information they need to perform a task. Our service returns to them only the information that exactly matches their request in a form that they can understand.'
Another research project, this time from Southeast University, China. A strikingly pristine and simple interface enables 'object search' or, more interestingly, 'Concept Search.'
An interesting addition to the Semantic stable. At a very early stage of development, so the website itself is not exactly brimming with information. Check out David Peterson's coverage at Sitepoint
SWSE (pronounced 'swizzy') is a 'search engine for the RDF Web on the Web'. This project is based in Galway, Ireland and uses a version of MultiCrawler to crawl the Semantic Web,
A 'Very preliminary' semantic search engine. A research project at the Open University in the U.K.
Next-Gen Meta Search Engines Top Dog:
Goshme is a Meta with a difference -- it will query multiple search engines, but also decide which ones are best for different kinds of queries. And some other top secret stuff. In Private Beta. Others:
A newish meta-search engine set up in 2005 with browser plugin support. In public Beta.
Music Search Engines Top Dog: Everyone's favorite playable music search engine. Just enter a song/artist, and let Seeqpod find you a playable version somewhere across multiple repositories like YouTube etc. Great, because you never have to download, and so are never breaking the law. But you can still be a cheapskate and never buy anything on iTunes. A super-simple Flash interface makes Seeqpod a pleasure to use. Others: This thing makes us feel like Beavis and Butthead when they saw the auto-flushing urinals at the Grand Canyon. Midomi allows you to find music with your voice. Just sing or hum the tune, and Midomi will tell you what it is, and where to get it. The service relies on comparing your tune with previous queries made by community members, so it will require a high level of adoption before it becomes really useful. But when it does, you'll never have that annoying 'What was that song called again?' moment ever again.
You can also perform a conventional text search, but where's the fun in that?
There's a whole bunch of other cool community-oriented stuff on Midomi, which is still in Beta. Check it out. Another music search engine with a vast search database (500m songs according to TechCrunch). Still in Beta, and highly impressive. A much loved and highly popular music search engine. Search and find tracks in mp3 and other formats. But don't download them or anything, you know. You don't want to break the law. Find sounds, pure and simple. If you can describe the sound you're looking for, like a cow mooing or a car backfiring or something like that, come here. A real time-saver for the .001% of people who actually need something like this for building Flash games or something.
Video Top Dog: An excellent, versatile and comprehensive video search engine. Which you can also get delivered to your desktop . Others: Video search and aggregator where you can set up your own feed. Pixsy is an image and video search engine that licenses its technology to external media sites. The company provides white label multimedia search, yet maintains its own search site at Pixsy.com.
Images / Photo Search Engines Top Dog: Now this is cool. Very cool. TinEye allows you to search for images on the Web, and 'Does for images what Google does for text'. You upload an image file or submit a link to an image, and they figure out where on the web that image is. It's not done by tags or keywords or any of that cave-man stuff, the algorithm uses direct image recognition technology. This thing is gonna be big, boys and girls. If it works. More on TinEye here.
There's also a FireFox plugin, and it's supposed to work on cropped images too. Our spies tell us that Digg has been using the technology of late. Others: This thing is amazing. Sketch the outline of an object/image in a sketch box provided, and Retrievr will go get matching images on Flickr. In real time. Our efforts were not altogether successful, as we were misguided enough to try to sketch a banana in the box provided, and Retrievr justifiably thought we were searching for something with phallic connotations. Rather an unpleasant episode, on the whole.
How they do this, we have no idea. But you've really got to check it out. Another image search in Beta, this time slightly more traditional, but possibly more obviously useful one than some of the babies so far covered.
You can search people, objects (very useful), and My Photos. The 'My Photos' bit is more than a bit clever. It has inbuilt facial recognition, so it knows, for instance, every time your wife or husband is in a photo. Allowing you to query your photo database for all your images which shows somebody in particular. An effective, no nonsense image finder.
International Stars Top Dog: Baidu is the biggest search engine in China. Which makes it a behemoth, even if you've never heard of it. There's nothing particularly interesting in terms of its algorithms or methods, but hey, so what. Sue us.
Baidu usually ranks in the Top 10 in Alexa's traffic list, and the eponymous company which owns it is listed generated revenue of close on US$ 230 mil in 2007, and is listed on the Nasdaq. It also runs a Wikipedia clone called Baidu Baike. Others: Yandex is the biggest name in search in Russia, where it enjoys 44% market share. Google (33%) doesn't much like coming second. Anywhere. And so a bitter 'Search Cold War' has erupted in search there.
People and Organizations Search Engines Top Dog: Searches the deep web for people and also gives Lycos people contact details. Initial experiments suggest this thing works pretty well. Hard to get out of the habit of just googling someone though.
Others: 'Spock is a people search engine which collects data from all corners of the web with a heavy focus on areas with dense people information such as wikipedia, photo sites, blogs and social networks.' Boasts an immaculate start page. Yellow Bot seeks to be a 'hyper-local search engine and community discussion resource.' Wink is "where people find people." You can search by location, name, school etc. TechCrunch's wikibase of startups and other technology companies. If you're going to visit some industry bod's office for a cup of coffee and a chat, check out their organization's essential statistics and TechCrunch coverage here first. This being TC and Michael Arrington, the platform is water-tight and brushed with hyper-efficient, professional sheen. Search, grab a widget, or get creative with the App engine. Similar concept to Crunchbase, from the guys at Inquisitr.
Enterprise Search Top Dog 'What if… you could search among your vice-president’s favorite Web sites? Or through internal documents—invisible to Google? Or within the contacts of your sales representative?' Get the picture? In a recent blog entry, these guys reckon they're already up to 'Search 4.0', which makes us feel very old and out of touch. Thanks a lot! Others: What's a Swicki? A search Wiki, of course, where you ask and answer questions on specific themes. IBM is bringing semantic search into the Enterprise and even into personal email, and that is more than a bit significant. We've covered both Enterprise OmniFind and the Email version (for Outlook-- which is free): 'The newest version is OmniFind Enterprise Search Edition 8.5 and ... it supports clients who use Lotus Quickr and Lotus Connections, in addition to WebSphere Portal. With these enhancements and others, OmniFind Search continues to keep alive IBM’s cross company Information on Demand Strategy.' Yes, we know. We're cheating a bit here, using a Google product and all. The Google Search Appliance is a piece of hardware that corporations install on-premise so that employees can search enterprise data the same way they search the web with Google. It "crawls your content and creates a master index of documents that's ready for instant retrieval using Google's search technology, whenever a customer or employee types in a search query."
There is also a 'Search Appliance Mini' aimed at the SME scale.
Blog Search Engines Top Dog Ask.com Blog Search 'combines Ask.com search technology and Bloglines subscription data. They index the most popular blogs that people across the web subscribe to daily to ensure higher-quality results and minimizes blog spam.' Ask's blog tool shows top feeds from the domain, 'Binoculars' previews, and direct subscribe and share (with digg, reddit etc.). There's a lot to commend this to bloggers, and A-Lister Marshall Kirkpatrick professes to use it every day. Others: BlogScope is a blogosphere 'analysis and visualization' platform run out of the University of Toronto. You can also get popularity curve and other widgets for your blog or keywords, and a FireFox plugin. Twingly was founded in 2007 with the aim of becoming the first spam-free blog search engine. This doesn't sound like a big deal, until you consider the applications which Twingly uses its platform for.
Namely, the Twingly Blogstream widget, which shows blog posts linking back to your original article. A widget which shows these backlinks, while filtering out the link-farm, rss-sucking garbage, is a worthy addition to anyone's community-building strategy, and also helps your readers by widening the scope of the conversation. Nice work.
Get the widgets here, or just go to Twingly.com to check out the search engine. Twing is a search engine for finding content on forums. And also does widgets and a FireFox plugin for embedded search. More. Wildly popular blog and MySpace search engine, where you can also build RSS feeds, get free statistics, and keep tabs on blog trends.
Twitter, Mobile, Open and Social Top Dog: Summize is an extremely useful search engine built on the Twitter API. It searches Tweets for keywords, and returns in reverse-chronological order. Simple as that. And yet, not least because of the simple device of including RSS support for queries, it is a stunningly effective research and monitoring tool, which can tell you what the blogosphere really thinks about your product or service. Others: 'NearU Search is a local SMS search company. The company aims to provide fast and easy search for texters. NearU’s content partners give them info on local deals, dining, movies and activities.' BrightKite isn't a search tool at heart -- it's a geolocater, which keeps tabs on where people are geographically by SMS. But once the service is built out (it's still closed Beta, and the whold world is beating on the door to get a peek inside) this will provide exceptionally useful search capabilities and applications. 'Sightix is providing a unique search service specifically to Social Networking Sites (SNS) which filter search results through the user’s social graph. As such, each user’s who search will receive different, individually personalized and relevant results relating to the social entity that they are looking for.' 'Vtap is a mobile search service that’s fine-tuned for video but can be used for audio and text as well.' This thing is so bleeding edge that all you will find at hooja.com is a subscribe field for beta testers.What they are doing behind closed doors is anyone's guess, but TechCrunch reckon its got something to do with ' allow[ing] people to use their mobile phone to capture information on the go and save it in a central online account they can access any time.'
Paypal's Peter Thiel recently pumped $1.5m into this Israeli startup, and this is not a man renowned for throwing his money away. Another Twitter search engine. Sky Grid mostly does news search. 'Founded in 2005, SkyGrid emerges from stealth mode this week with a search tool that sifts through hundreds of web and mainstream media to show you just one thing: whether the balance of the news on a public company is good or bad, and how the “mood” is changing. Naturally, hedge fund managers are eating it up.' (GigaOm)
There does seem to be a lot of cool features in this thing. Mind you, it crashed our browser, which makes discussing them problematic.
Miscellaneous / Useful Search Engines As one might imagine, this one's a specialized search for PDFs. Surf Canyon, in Beta, offers a plugin for FireFox/IE which 'augments other search engines with recommendations, personalized information and digging.' It drills into search results, personalizes in real time, and adds preview images to Craigslist pages. Pretty cool, if your browser isn't fully laden with plugins already Copyscape allows you to search the Web for other places where your content has shown up. So you can sue the b**tards who stole it. Every blogger should have this website bookmarked. Krugle is a code search engine. Check out the result here for 'php shopping cart' and judge for yourself. Enter a URL, find similar sites. Better than Google Similar? Er... debatable. A slightly strange but overall useful recipe search engine which allows you to search recipes by what you have in your refrigerator. Big does one simple thing: provides your Yahoo!-powered results in BIG FONT SIZE. If it also blew up the page you're going to, this thing could be huge with the vision-impaired. So why don't they? (And what an incredible domain name?)
Search Platforms /Tools Top Dog: Whoever said Yahoo is behind the times? Yahoo Searchmonkey is just one of a myriad of brilliant products doodling along under the radar at Sunnyvale, and is the company's new open search development platform. We covered this at CMSWire a little while ago and were justifiably enthusiastic. Others:
Anywhere Anywhere is a development tool to enable search in Web and mobile applications. It is an open source project which 'enables developers to include location search within their code' using a variety of sources (e.g., Yahoo Geocoding, Live Search, MaxMind GeoIP etc.).
This very, very young dohickey (v126.96.36.199!) allows location finding based on IP addresses, and enables 'conversion between different types of location information - latitude/longitude can be changed to zip code, zip code to city/state, etc.'
Bizarre / Strange Search Engines Okay, maybe they're not all that useful. But they are quirky. Top Bizarro: A highly attractive but very strange search engine that sings to you, insults you and lives under a highway flyover in a futuristic cityscape. Others: Come on, you just knew there had to be one of these out there. Yes, that's right. A Star Trek search engine. Which you can add to your FireFox search box.
Bonus search engine stuff: In case there isn't enough reading here for you, we enclose a link to Larry Page and Sergey Brin's original paper which laid out the Google algorithm. Enjoy!