Job search engine Jomea.com on Orbit TV

November 20th, 2009 by Charles S. Knight
Posted in CEO Views, Global, Job Search, Verticals | No Comments »

jomea-logo

Online Real Estate Search – the Elephant in the Room

November 12th, 2009 by Guest Author
Posted in CEO Views, Guest Authors, Real Estate, Verticals | 3 Comments »

roost-logo

Get ready for a shocker – the vast majority of online real estate sites are a waste of time for a real life home buyer. If you were buying a home in your town of choice and you knew there were 10 homes that met your criteria – would you use a site that only showed 6 of them?

Sadly the vast majority of consumers have no idea that when they’re on the major Real Estate search sites this is exactly what’s going on.

I’m not going to name names in this post, but the bottom line is that virtually all the real estate vertical search players as well as the real estate sections of the top horizontal search engines get their home listings via broker/agent syndication feeds.  What this mean is that in order for these search engines to have complete coverage, they have to get every agent or broker to upload every new listing, every day. And as a result, no matter how terrific their User Interface (UI) is or how much other information they throw around the properties on their site – they are missing big swaths of current listings that consumers want and need.

They’ll deny it, they’ll tell you that they have “92%” of the listings – but what this really means is:

They know there are x number of active Multiple Listing Services (MLS) listings in the country;

Their database has y # of total properties;

x/y = 0.92;

That is very different from having 92% of the actual MLS listings that are active in any given market!

What it means is that they have a ton of other stuff in the database … foreclosures … homes that were already sold … strange hybrid listings … and so on.

The truth is that only those websites owned by, or operated on behalf of, an actual Real Estate Brokerage or Agent have access to the full MLS data set.  There is one exception, Realtor.com, which by virtue of its relationship with NAR gets access to the vast majority of MLS listings.

So here’s a tip - there are two real estate data acronyms to look for that will help you as a home buyer determine if you’re likely looking at the complete MLS data set.  One is IDX (Internet Data Exchange) and the other is VOW (Virtual Office Website).  These types of data sets typically (although sadly not always, more like 9 out of 10 times) mean you are viewing almost everything available from the local MLS.  Look for them in the home listings themselves or at the footer of the real estate websites you’re using.

“Wait a second,” you say … “isn’t your company Roost.com a vertical search site subject to the same limitations?”

Well I’m glad you asked! If you poke under the hood, Roost is actually a hosted network of individual Broker and Agent websites with a smart search engine on top.  That allows us to do a lot of cool things, but most importantly for this discussion it means that when you’re searching on Roost, you’re actually looking at a Broker or Agent’s website which has access to the good stuff.

Go ahead … give us a try … click here.

Alex Chang
CEO of Roost
Roost post

What is Social Search? By Zakta CEO Sundar Kadayam

October 28th, 2009 by Guest Author
Posted in CEO Views, Guest Authors, Social | No Comments »

The Official Zakta Blog
By Sundar Kadayam
Founder & CEO, Zakta
The Personal and Social Web Search Engine

It is hot!  So hot that Google legitimized it with their recent update.  Buzz is building on social search like never before, as this handy trend graph from BlogPulse indicates:

screenshot-trendgraph-socialsearch

But what is social search?

According to different industry voices, social search …

“… involves combining social graph information with pure algorithmic search results.

“… combines traditional algorithm-driven technology with online community filtering.

… helps you find more relevant public content from your broader social circle.

… is information retrieval, way finding tools informed by human judgment.

These definitions are quite broad and varying, and the result is that so many solutions have come under the banner of “social search”. However, one thing common across these diverse set of tools and services is this: they’ve all used collective intelligence (wisdom of the crowds, if you will) in some way to improve what they present to users in the search process.

Here are some that come to my mind:
  • In the early days of the Internet, DirectHit (later acquired by Ask Jeeves) watched which links users clicked through more for a given search and used that data for dynamically ranking search results based on their popularity with the community of users.
  • Amazon has been a pioneer in the space of using social/community data to improve the searches for users on Amazon.com – much has been written about their recommendation engine!
  • Intelliseek’s ProFusion.com engine ( a product I helped design) used an adaptive search mechanism (community usage driven) to determine what are the best sources to pick for a given query in a distributed / federated search environment.
  • Wikia Search used the Wikipedia model of direct, swarm-editing of search result pages for different queries. i.e. Wikia Search users could interactively change the results on any result page, and impact what other users saw directly.
  • In reality, Google has always been a social search engine, in a couple of ways. They’ve always tracked what people have liked through who / what they hyperlink to – a core to their famed PageRank algorithm. In the recent years, they’ve also included user and community contributions (in the form of social media) into their search results, with content from Wikipedia and the blogosphere impacting search results in a noticeable way.
  • Yahoo has tried integration of Delicious (their social bookmarking system) into the search results.
  • Presently, the buzz is all about including social network data and data from popular social tools like Twitter into the search results. Bing did it. Now Google is doing it too!

My company, Zakta, is also a recent entrant in “social search”, and we refer to Zakta as a personal and social Web search engine.  Our aim is to improve informational searches on the Web.What prompted me to write this post was the recent Google announcement on social search.  Our small community of users felt that Google was encroaching on Zakta’s turf, and I thought I should help clarify where Zakta fits.

First, Zakta has no turf – Google dominates all :-) Second, we are trying to add value to the informational search experience of users through a comprehensive solution framework, so we don’t get into feature battles with giants that we don’t have a chance of surviving (as it is, I’ve been called “Nuts!” to start Zakta at this time, and having my tiny company enter into a feature race with the giants should surely bring me the label “Stupid” too – something I’d very much like to avoid!).

Here’s a personal framework that I’ve used to understand the social search space myself and to steer the design and development of Zakta.
screenshot-socialsearchlandscape2

On the X-axis, I plot the Personal (focus is on the individual) versus Communal (focus is on the community as a whole) continuum.  On the Y-axis, I plot the nature of information that users interact with, in terms of whether it is Disorganized (focus has been on mere collection of information) versus Organized (focus is on curation of digital information).

Using this framework, I’ve mapped a handful of social search services and tools that I’m somewhat familiar with. So, admittedly, both this framework and my characterization of these services in this framework are based on my personal viewpoint.  I’d welcome comments for improvement, or other viewpoints.  I hope you find this framework a useful tool to make sense of what is happening with this growing space that is simply called “social search”.

Now I can put Zakta into this context. As portrayed in this framework, Zakta is a personal Web search engine because it provides tools to deliver a personal search engine experience that puts the searcher in control.

Zakta is also a social Web search engine in many distinct ways:

  • It enables a searcher to collaborate with people they trust to find, collect, organize and share information on topics of interest
  • It enables a searcher to connect to others they trust and discover information relevant to their interests from the recommendations made by their trust-network
  • It enables a searcher to benefit from the contributions of the community of Web users in the form of published Zakta Guides on topics of interest
  • It enables a searcher to gain from the ongoing relevance ranking improvements that happen behind the scenes that take into account the signals of recommendation expressed by not only the user’s trust-network, but also the community as a whole not just on Zakta, but elsewhere on the Web

As you can see, Zakta is not as much about finding what your social network has been saying.  Rather it is all about empowering you personally and helping you benefit from your trusted network as well as the community at large to improve your own Web search experience and discover useful information on an ongoing basis on topics of your interest.

As always, I’d love to get your feedback!

Search the experts at Trakkrz “We track down the best.”

October 2nd, 2009 by Charles S. Knight
Posted in CEO Views | No Comments »

logoWe have officially launched the trakkrz blogger community and I’d like to congratulate all of those bloggers who made it into our launch and were selected for their high quality.

As we launched we have received some questions about who we are and what we are trying to accomplish. One of the first was Scott Lowe from scottlowe.org – one of the bloggers we’ve selected for our category Virtualization. I took the chance to give Scott a sense of what we are trying to accomplish that I think fits well in an initial description of our service on our blog. So without further ado here’s the body of the email – hope it gives you a good sense of what makes us different from most other blog communities on the web:

Hi Scott –

My marketing manager informed me about ur having questions about trakkrz.com via twitter. Hopefully I can give you a vision of what we’d like trakkrz to be as it grows:

A place where people passionate about a topic can find what is being said by the top bloggers, trakk their posts, comment together and direct people to real people writing real, current articles using their expertise.

Here’s our strategy for doing so:

* Make sure spam doesn’t get into the system – if we only select blogs with true people behind them that really care about their field then spam will never get in – don’t just follow some scraped RSS feed for the topic and post anything that whizzes by – if those bloggers start to spam we’ll simply remove them – we find that almost anything of significant importance on a topic is mentioned fairly quickly by the top bloggers in a topic.
* Highlight the best of what is already good content – give users the ability to “trakk” what they consider to be the best content for each topic – let people see what is the most trakked today, this week and all time – this way people can see the truly exceptional articles at a glance and then go visit them to get the full scoop – this will take time and building the community but we’re committed to doing just that.
* Share revenue – here is a tough one that we’ll continue to refine and improve – because we want people who really care to drive the community, what better place to get them than our currently selected bloggers and what better way to motivate them than sharing any revenues we generate (I know they’re tiny now – but we can only hope ;-) ) – therefore as people sign up you’ll see user’s adsense code showing up in our source code – along with our own ads.
* Be transparent – we give followed bloggers access to trakkrz-analytics.com so they can see just how much traffic we’re getting, how many visitors we’re sending their way, what keywords are “popping” and more – it’s built on piwik, the opensource google analytics alternative
* Make it fun with some interesting photos and videos from flickr and youtube – seeing what is popular via other, non-blog media we think gives a little unique flavor to the web and sometimes adds perspective to the written blogged view.

Really I’d try to say that we’re trying to be like the love-child of Digg and Alltop – a commentable place but that has that added filter of editorially selected blogs.

Thanks everyone for your interest in trakkrz.com. We look forward to your participation in what we hope will be a wonderful community to support bloggers who provide great content for the rest of us.

Todd Hogan, Founder – trakkrz.com

Melody Catcher – the Internet Music Search Engine

September 23rd, 2009 by Guest Author
Posted in CEO Views, Guest Authors, Music, Verticals | No Comments »

mc

The Melody Catcher

“What’s that song stuck in my head?”

Introduction:

Massive and ever increasing volumes of music are available online. As long as titles or names of composers, performers or artists are known, it’s easy to find related information. If only the melody is remembered, however, common search engines fail to provide any results. Many ideas and concepts of melody finders or Music Information Retrieval (MIR) Systems have been described. For a good overview see: http://en.wikipedia.org/wiki/Music_information_retrieval#MIR_applications With the Melody Catcher, we aim to make melody searching accessible for the broad public.

The Concept of the Melody Catcher
The Melody Catcher system (MC) is based on a novel approach to reach its goal, a better disclosure of online audio sources. The matching technique focuses on monophonic and well defined online audio sources rather than the more complex polyphonic audio sources. In addition, the system improves and combines aspects of existing melody search systems, optimizing the user interface, increasing error tolerance and improving the efficiency and accuracy of search results.

The use of monophonic MIDI files as the basis for matching is not new as such. Some of the existing melody search systems use such MIDI files in their database. However, in contrast to the MC, no such system includes arbitrary sources from the Internet.

2009-09-23_1001

Several online music collections use monophonic (a.k.a. single voice) MIDI files. Such collections can be incorporated automatically into the MC’s matching system with its own web crawler. In addition, there are thousands of sites offering collections of polyphonic MIDI files. These files are already accessible by MC by text only search. In the future, we will improve the MC algorithm to include searching for polyphonic MIDI files. We will also propose guidelines for content authors how to create polyphonic MIDI files for an optimal inclusion into MC’s database.

Furthermore, many sites offering ring tones for cell phones provide monophonic MIDI files for pre-listen functionality. We have found that these sites are an immense source of popular melodies. Other ring tone file formats such as RTTTL can easily be converted by the web crawler for inclusion into the database.

Currently there are some 10 melody finders on line. Evaluations learned, however, that their overall results still are disappointing. In most cases the wanted song is not found or at best low in the results.

The Melody Catcher is based upon a different approach.

Its matching technique is focused on monophonic and well defined on-line audio sources, like MIDI audio files. This is leading to far better results than systems focusing on polyphonic audio sources and in most cases showing the wanted song in top of the results. Although partly based upon its own database the system is mainly dased upon on on-line sources of such audio files, found by a web crawler. In the results the user can play the the wanted melody and find its title, but he is also linked back to the source site of that melody. In this way the system can be of help to discloses sources of MIDI files, containing the wanted melody and in most cases also similar songs. On the site under about one can read more about its history.

Best Regards

Dr. Jan L. van Os
The Melody Catcher