Ran Geva – an entrepreneur and the developer of Omgili. Ran has a deep understanding of various Internet technologies and years of experience when it comes to developing and promoting services and software on the Internet. Ran is also a well known P2P developer, and his software is used by millions of users around the world.  You can check them out at http://www.rangeva.com

We welcome Ran Geva, CEO of Omgili, to our CEO Spotlight.

ASE: Hi Ran, thanks for joining us today.
Ran: Hi there! Happy Holidays.

ASE: Ran, please tell us about yourself- where were you born, what did you study, how did you get into the Internet space?
Ran: I was born in Northern Israel in a small town called Karmiel. I left home at the age of 16 to attend military school and served in the army till the age of 22. I was always a creative guy, I loved art and to invent and that’s what drew me to computers.

ASE: What did you do in the army (if it’s not secret)…
Ran: Actually it (the army) IS a secret Sorry!

ASE: Understood!
Ran: It (computers/the Internet) was the easy way to invent and to create.

ASE: Do you have siblings?
Ran: I am the youngest. I have a brother and a sister.

ASE: Are they involved in your business?
Ran: Nope. My best friend Israel, who is also my cousin, is my co-founder at Omgili. But that’s it…

ASE: Please tell us about Omgili- when and how did you come up with the idea; what was the original vision?
Ran: The initial idea was to create an application that is different from what I had been working on, which was freeware software.

ASE: Sorry, what is freeware software?
Ran: Each of my software programs had a support forum/community around it –most of the software I’ve created is listed on www.rangeva.com.The idea for Omgili was to create a system that could analyze UGC (User Generated Content) and offer companies an idea of what people were saying about them.

ASE: Sounds like free market research.
Ran: I found the original idea to be a bit complex so I started from the basics, developing the algorithm that analyzes discussions. That’s where the Omgili search engine came to life. The feedback was awesome. People loved it, so I continued to develop it and then the VCs entered the picture. Pitango (an Israeli VC) invested seed money in Omgili and now we are working on raising Round A.

ASE: How does Omgili work?
Ran: The analytics system is in development. It is a professional system where companies can extract insight from UGC (User Generated Content). The public version lets you draw graphs, search and read. The amount of data users create on the Internet is amazing! There is no way to manually handle this entire amount. This is where we developed Omgili Insight (prototype) that helps companies get a hold on the data. A true, automatic, real-time online system. In addition, we are working to release more features to the consumer. One of them is called My Omgili. We hope to release it in a week or two.

ASE: So would you say that Omgili is primarily B2B, rather than B2C.
Ran: It is both. We released Omgili Search Aid, Omgili Product Reviews, Omgili Buzz, and soon (in coming weeks) we’ll release My Omgili and Omgili Directory.  So for now it’s actually a much more consumer-targeted business. Omgili Reviews for example aggregates opinions from multiple review sites (CNet, epinions, shopping.com, etc.) into one overall buzz gauge. It gives you the overall rating calculated from these sites, the mood over the past three years, and the discussions buzz from millions of discussions.

ASE: Very substantial data! So, how are you getting the word out about Omgili?
Ran: Currently our marketing budget is non-existent. We are relying mostly on users spreading the word and kind bloggers writing about us. We are doing quite well all things considered. We receive about 50,000 daily unique users!

ASE: Good to hear.
Ran: Thanks… It is very hard to succeed when it comes to a vertical search engine. Everybody is using Google (myself included). The users of Omgili are often those who tried Google, failed to find what they wanted, were searching for subjective information (that’s what Omgili is all about), and only then tried Omgili.

ASE: Which brings us to the next question… Are you exploring partnerships with other search engines?
Ran: Actually I am. We are in talks with two search engines- with one of them we are in the implementation process. I am sorry I can’t disclose their names…

ASE: It seems wise to bundle your services with others. Going back to Epinions- would you say it’s a similar model to Omgili?
Ran: Epinions is a service for reviews and asks users to create content. Omgili is a search engine, and our users don’t create content. We cover these UGC based sites and offer a better search experience. 

ASE: Thanks for clarifying. Are there different versions for different countries/languages?
Ran: Not yet… We will be working on that – hopefully very soon… It is just a matter of changing the GUI (Graphic User Interface). Omgili already enables users to limit the search into specific languages (from the advanced search screen).

ASE: Seems Omgili would be a great resource for movie studios, politicians… All about buzz!
Ran: True. Also for consumer-related companies (software, security, hardware etc.).

ASE: Are you primarily focused on Omgili now? What about the other products listed on your website?
Ran: I am afraid that most of my other projects have been abandoned since I founded Omgili… I hope to be able to maintain them though!

ASE: I’m sure you will come back to them. By the way, what does Omgili mean?
Ran: Oh My God I Love It  8-)  That’s the reaction I’m hoping for when people use it. The name is easy to spell and easy to remember.

ASE: I was trying to figure out if it meant anything in Hebrew. I love acronyms. Oh My God I Love It! *(and will never forget it). How many people working with you now at Omgili?
Ran: Four, including me. We’re tiny.

ASE: Sounds about average for a start-up ASE. So you wear a lot of hats.
Ran: Well the hats are aligned this way:
1) Me – CEO, Developer, Marketing and everything else
2) Israel Tsadok – VP R&D
3) Yoav (my partner) VP Marketing
4) Ella – Developer
That’s it. It’s funny to list these titles, because we are a tiny company and it doesn’t really mean anything.

ASE: Do you have a mascot?
Ran: We used to have one. It was a small dwarf (many people said Omgili sounded like the name of a dwarf). But no longer – apparently it confuses people.

ASE: Personally, I love mascots. But I can see that it might take away from the purity of the brand.
Ran: I loved the dwarf, as well.

ASE: The dwarf, putting a friendly face on search, not a bad thing, I think…
Ran: BTW we’ve also got the bee:   for the buzzzzz

ASE: I like these illustrations. They make it more accessible, friendly.
Ran: A friend of mine drew them – she is a very talented girl.

ASE: I spoke to Michael Hussey of PeekYou last week and he said that they were trying to ‘humanize Google’ – I liked that.
Ran: I really like what they did with Google.

ASE: Anything else you’d like to share with our readers before we finish up?
Ran: I would like to say that I love the AltSearchEngines blog – I think they are doing amazing work! 

ASE: Thanks!  In the meanwhile, thank you so much for your time.
Ran: It was my pleasure!

ASE: Shalom, shalom as my grandma says.
Ran: Shalom.

Natalya Murakhver is a freelance writer/PR consultant based in New York City.

This CEO Spotlight weekly feature is sponsored by   PeekYou “the smartest way to find people online.”

Appendix
In which Ran explains the difference between Omgili’s approach and Natural Language Processing (NLP) for Product Reviews:

I guess it does sound from our conversations that these NLP methods aren’t very good – that wasn’t my intention. First I want to clarify something: Omgili’s product review is just a sub-section of the data source we analyze. We took as a mission to collect UGC information, analyze it and make it accessible. We cover forums, news, groups, mailing lists, q&a sites and review sites. When it comes to discussions Omgili takes unstructured data (the text in the HTML page) and automatically adds structure (title, topic, replies, dates, users etc…).

In the case of the reviews, we found that there is an additional parameter we can extract: the overall grade the user gave. It isn’t ambiguous – it is a number from 1-5 (or sometimes 1-10) and it states the overall satisfaction of the user. We rely only on the structure of the data. This is what we give our users when it comes to Omgili product reviews (and only there). The problem with this method is that we cannot extract sentiment from text where the sentiment is explicitly structured (as explained before).

What all these sites are doing is utilizing very sophisticated methods where by applying rules, and lexicons they are structuring the sentence itself and trying to extract the sentiment. Sometimes it works and other times it doesn’t.

There are many methods and algorithms being applied and the problems occur mainly because: a) Tonality – there is a real problem in detecting cynicism, like “This camera is great really great. It is excellent when it comes to taking bad pictures.” b) Bad grammar – usually these systems rely on some grammatical rules but when it comes to UGC they have real problems where people are writing fragmented and misspelled words or slang. We experimented with some very sophisticated products (Clearforest for example) that worked great on articles but failed miserably when it came to UGC. c) Subjective opinion. Your negative could be my positive. When you say don’t like the service or product because it interacts with another service you don’t like (Apple for example) I might find it a positive thing that they interact with that service. d) Phrases that are common knowledge “this product is as good as x singing y song.” Now for knowing if it’s good or not, I will have to know x and know if x sings y good or bad (and here we have the subjective problem again).

There are more problems with NLP, and what I was trying to say is that when you analyze text and even get to accuracy of 80% which is extremely high when it comes to UGC (I actually don’t know if anyone is able to benchmark 80%) you are left with a scale that you should trust but only at max 80%. Which leaves you with a minimum 20% of the times where it’s completely wrong.

Now, as I said, we are not competing in this field. I just tried to bring up the problem when it comes to sentiment extraction on UGC data. I hope that makes it clear!