Archive for April, 2007

Google’s distinguished webspam engineer Matt Cutts has an insightful post about his company’s approach to user privacy. Albeit I agree with most of his arguments and appreciate Google’s pioneering efforts in this field, I don’t think 2 years is a short period of time to anonymize user queries and I don’t believe your private data can be well isolated from 3rd parties. Google’s personalization technology requires them to keep an eye on you, follow your attitudes and behaviors continuously. They are constantly working on this technology, otherwise it would be impossible for them to bring you the most relevant results related to your search terms. And also, it’s my personal opinion that even if they do their best, they can never hide your personal data from some big big big organizations.

In contrast, privacy is not an issue using hakia, and we will explain soon why that is.

According to Chris Sherman’s recent post Google will start to mix news with Web search results. We, at hakia, looked at each other and asked “aren’t we already doing that?” The answer is Yes, since last August. Not only did we introduce this freshness booster last year, we have also doubled the capacity of news volume to be mixed with Web search results in our latest BETA update.

The bigger news is that searching “news” is nothing but a long tail experience. Since there is literally no time to collect statistics to determine popularity, most search engines are using the short-cut method: a long list of “trigger” words. If your query is “Britney,” for example, then you are likely to see some news at a designated spot on the results page. However, you may not see the same news result if you enter “singer with post-partum depression.” The news article for the first query will not show up for the second query because it did not contain the right trigger. You have to try this experiment with a news article that is only few hours old. When the news get older, it may show up from the archives, and then that is besides the point.

Google says it will give up the practice of designating a space for news, and start mixing them with Web search results using a new “no-trigger” algorithm. It means search for better search is on its way, and we wish them good luck in this new endeavor.

The only proper way of handling long tail searches, whether it is for news or anything else, is to deploy a full-scale, uncompromised semantic search capability. If your algorithm can understand the text and the query, and if it can find the matching concepts (not only the words) in a split second, then you don’t have to collect statistics while the news rot. Nor would you need a trigger list. And that’s the news from hakia.

It sounds good, doesn’t it? Among the many speeches I heard at the Web 2.0. Expo, this one did beat all in originality! Tim Ferriss talked about his book, Four Hour Work Week, where he outlines it for us, the overworked and over interrupted souls, how we can manage our time better. Do you check your email at least 100 times a day? You live in my world! He takes time management to new levels: he told the audience that he outsources everything, even his dating life. I am not sure how one can do that but the book’s premise is worth exploring. Personally, I will implement some of the suggestions in moderation.

While Amitava, Emre and I were representing hakia at the Web 2.0. Expo this week, see our picture below, it was great to leave the virtual world for the real world of exchanging information and cards as one does in such event. We even got caught on camera doing that.

Speaking of social networking….On the search front, I fully agree with ebay’s view and presentation that there is a need to “build truly social experience engines.” What does that mean? We think a searcher should be able to connect with other Web searchers who are looking for the same information for a richer search experience AND social networking. We are already working on a product with this objective in mind. Stay tuned…

As the conversation grows on hakia’s search-for-better-search page, I am tempted to shed light on one particular issue. Where was NLP all these years? And why will it happen now? I will start with an analogy.

Everyone knows how to change a flat tire. It is supposed to take 5 to 10 minutes. In Formula-1 race, however, you are supposed to change a tire in about 3 seconds. Accordingly, a whole new technology and industry derived just to be able to change a tire in Formula-1 race in such a short time.

Natural language processing (NLP) is as old and established as changing a flat tire in 5 to 10 minutes. Well, a search engine environment is like a Formula-1 race. Using NLP in a search engine is very much like changing a tire in 3 seconds. Therefore, an entirely new technology is needed to be able to accomplish this. And that is what has been missing all these years.

Several attempts to bring semantic treatment in Web search have failed in the past that included a variety of short-cut approaches ranging from statistical sampling, brute human effort, tracking user behavior, and many more. A long list of trials, all avoiding the middle layer technology to enable a full-scale, uncompromised, semantic NLP.

Conclusion: NLP and Web search do not go hand-in-hand unless the middleware is invented. Once this obstacle is out of the way, then the possibilities become endless.

Despite the history failed the idea, if someone finds a way to teach an old dog a new trick, we will rest our case. Until then, I say let’s look for new enabling infrastructures to appear in the horizon, … let’s talk about the middleware. If NLP will happen now, it needs this new vision.

Greetings SES NY attendees! Come and meet our team in the Expo in booth #2510 (on the second floor of the exibit halls). We can give you a tour of our search engine, hakia Galleries and hakia Labs. We will give away our own Search Music CD and some surprise gifts (focus on the “i”).

Stop by our booth to say hello, meet our team or pick up search music and gifts – at the very least you can pick our brains on suggestions on what to do in the Big Apple as we are natives based in the Financial District. Enjoy the show!

Not everyone is lucky enough to work for something that he or she truly believes is going to change the world. Well, I’m proud to say that I’m one of those lucky ones. This post is to announce that I’ve started working for hakia as a Search Evangelist. In my non-hakia time, I contribute to the Read/WriteWeb as a writer and grow my company, Groups.

I’m here with a team of motivated people, who are well aware that they are building something new. As you know, hakia’s purpose is to extract bits of meaning from zillions of pages on the internet and give you control over them for search. Shifting a technology and flowing in and around the boundaries of science is a very exciting job and I’ll take part in this quest as an Evangelist.

I invite you all to share our excitement and give us continuous feedback. Come, check, report back our progress and iterate this process. This is exactly what I was doing, as a hakia user, before joining the team. Be part of something new!

Today, we have kicked off a conversation with some of the top bloggers entitled “The Search For Better Search™” where the participants share their views on where search needs to go to get better.

As you read the thoughts of bloggers including ResourceShelf, TechCrunch, VentureBeat, Ars Technica, GigOm, ReadWrite Web, John Battelle, and our own Chief Scientific Officer, Dr. Christian Hempelmann; we want to hear from you as well. You can vote on what you think a search engine should do and join the conversation at: www.hakia.com/searchforbettersearch.html

We are in the humble beginnings of the Web search industry. Now it challenges itself to get better. One thing we all know, “The Search For Better Search™” will never end.

As the new search player around the block and the champion of the “search for better search” conversation, we thought it is only fair to show our technology instead of just talking about it. hakia-Lab pages welcome everyone interested in some deep technology illustrations. Some pages are accessible by invitation only. hakia-Lab keeps the pulse of our technology and will be updated continuously.

The interactive pages for Ontological Semantics implementation show how a sentence (in English) is analyzed by hakia, word-by-word. Although the complete implementation is our trade secret, these first steps shed light on how matching words are replaced by matching concepts. The dots can be connected by careful eyes with a linguistics background. In addition, some new information is posted about our proprietary QDEX and SemanticRank technologies. Access to their interactive pages are by invitation only.

hakia search engine is on its way analyzing Web pages and our development will reach maturation later this year. When the critical implementations are online, the benefits will become fully visible at every search, most significantly at searches of the long-tail. Until then, keep trying hakia to experience the very first meaning-based technology grow. Enjoy your experiments in the hakia-Lab!