Archive for May, 2007

In our first “Search for Better Search” conversation, we asked you what you expect from a better search experience. You said you want a search engine that understands your query the way human does.

Yet, a funny thing happened as Web search became part of our everyday lives. You and I have started to provide more and more information about our search habits and history to our search engines. Gradually, users’ trust has become a hot topic and an important issue.

Many of you are following the active privacy/trust conversation in the press with Google primarily in the spotlight as the largest player, and on the verge of the DoubleClick deal. We have participated in the discussion by talking about hakia’s different approach: We don’t need your personal information to improve our search algorithms. We will communicate with you clearly if some data must be captured for a particular reason and your privacy line could be seemingly crossed. You can read more about this in my prior blog post and in an Investor Business Daily article

With user trust being such a key issue and hot topic, we have dedicated the second “Search for Better Search” conversation to trust. This time we ask your opinion on how the search industry can “win users’ trust” and invite you to blog about it. You can see the comments on trust from some of the top bloggers. Again, we ask for your opinion and input: Do you trust your search engine? What is the single most important issue search engines can address to win users’ trust? Please visit www.searchforbettersearch.com and let us know what you think.

As the search industry evolves, it will naturally have issues and growing pains, like search engines winning users’ trust, just like a friend must win your trust. How does your friend win your trust? By communicating with you so that you get to know and understand each other.

We think there could be a better way for a search engine to build trust with Web searchers. Search engines can engage users in a conversation about their privacy in a transparent way. Will it take more effort on the part of search engines? Definitely. Is it worth it? Absolutely!

What do you think?

We hope you have stumbled upon Riza’s article, “Semantic Search: An Antidote for Poor Relevancy,” in ReadWriteWeb. We invite all to join the conversation!

The latest conversations around search innovation are focused on semantic search, our favorite topic! Mor Naaman, a research scientist with Yahoo Research Berkeley, has declared that “Semantic Web is dead”. ReadWrite/Web had consecutive posts on who does what in the space and posted questions on what semantic search is. Google’s Universal Search presentation has examples on how they will improve the algorithm by understand users’ intent- a “semantic” enrichment.

The more we talk to people in the tech space, the more we understood that the word “semantic search” means different things to different people. So, we decided to join the conversation to tell you what we mean with semantic search. Watch this space for our hakia Semantic Search 101 Blog Series where our technologists and academics will chime in to talk about: What is Semantic Search?;What is Ontological Semantics?; How Can You Test a Truly Semantic Search Engine?;What are the Challenges of Building a Semantic Search Engine?;and more.

Looking back, we have kept you abreast of our journey building hakia.com, a new, meaning-based search engine, for about a year. Since Day 1 of our introduction, we have received a lot of feedback from hakia.com users and Web searchers. Our recent “Search for Better Search” poll confirmed our starting hypothesis one more time: Web searchers want a search engine that understands text the way a human does. Users want to benefit from semantic applications in their search experience.

To recap it, hakia Beta search engine has been online for about a year. We have not used keyword search, rather we have employed sentence analysis powered by our semantic capabilities. Our philosophy has been to offer the Web searcher the best experience from one box, the search box, in quest for “Search for Better Search” by:

1) Mixing news with search results http://hakia.com/search.aspx?q=Christie%27s
2) Mixing weather with search results http://hakia.com/search.aspx?q=weather+Paris
3) Mixing famous quotes with search results http://hakia.com/search.aspx?q=democracy (see the bottom of the page)
4) Mixing meaning equivalences (multiple query results) in each search result http://hakia.com/search.aspx?q=did+Bush+kill+the+last+bill+in+the+senate (kill=veto)
5) Mixing categorical search results (equivalent to 10 simultaneous queries) http://hakia.com/search.aspx?q=piano

We take pride in the fact that we are the pioneers to introduce 1, 4 and 5. Recently, a number of new companies joined us in our quest to bring meaning to search. Some established search companies have announced their enrichments to the search experience. It looks like 2007 will be an interesting year in search space with lots of innovations in the pipeline. In the end, the Web searcher will have a big smile on her face and will enjoy a better experience with more relevant search results.

Meanwhile, please drop us discussion requests and questions and stay tuned to join our semantic search conversation.

Next: hakia Semantic Search 101: Definitions.

According to Elinor Mills’ recent blog post , our band is one of the things hakia has that Google doesn’t! Joke aside, we do not know whether Google has an official band or not. What we do know is we enjoy great food, music and company when our band plays in the Knitting Factory . And guess what, they play search music ! Yes, you heard it right, the creators of hakia Search Music composed music around hakia Web searches.

Our upcoming gig is on May 29th and we would like to invite the New York-based search community to join us! Come on and come along: fellow search engine colleagues, SEO experts, other tech innovators and hakia users. Please RSVP to searchmusic@hakia.com if you would like to join us (RSVP required per space restrictions).

We are looking forward to making your acquaintance! To listen or download the album “Cogito Ergo Search,” please visit www.hakiasearchmusic.com

The human brain takes about 5 years of constant training to have basic cognitive skills. Most of those initial years are spent on developing sound and speech processing capabilities along with visual input coordination. Then it takes another 10 years to build up vocabularies and concepts. At the age of 15, an educated human brain might have up to 5,000 words in its active memory. Yet, the development of “understanding” and “logic” continues for a long time.

So, what is BETA-14 in the context of the development of the human brain? Well,… hakia is a meaning-based, language processing system. Therefore, there are unavoidable similarities between the two. I felt like it is time to give a new perspective to what BETA-14 means.

hakia.com has built semantic resources (with over 100,000 lexical entries in English) to match a super human. You can see the depth of our resources in hakia lab. No ordinary human would be able to match that capacity.

However, “understanding” is a different animal. It is the capability of tracing the conceptual relationships among the words in a given sentence to come to an “aha!” conclusion. Therefore, developing this “understanding” capability is considered an entirely different process.

hakia’s BETA-14 is a step toward developing its “understanding” skills by means of improving its Ontological parser along with more page analysis. We expect to reach the “aha!” moment in the upcoming months and that will mark a new era in Web search. Although the users of hakia can see some of such differences now on the BETA site, the upcoming “aha!” moment will make Web search drastically better for almost every query.

One important clarification is the needed here for the term “better”. Better means enriched. For example, the word “kill” in the context of “passing a bill in the senate” also means “veto”, thus you will be seeing search results including the association “veto” and “kill” to a query “did Bush kill the last bill in the senate?”. Zillions of such associations are missing in the conventional search engines today, and breaking this barrier definitely deserves the reconition of “the next generation” search technology.

There were many other enhancements to hakia.com in this development cycle. We will announce those one at a time in the near future. Until then, keep trying hakia.com, you never know what you are missing otherwise.

Search and privacy… The heated debates on this topic are countless. Emre started the dialogue on our blog with his post. Thank you all who waited up for part II of this conversation.

Let me ask you a question: How do you want your privacy to be handled? Do you want to opt-in or opt-out. Right now, there is no opt-in option when its comes to search and we are here to change that.

What do I mean? We take a different approach to privacy. We think we should ask you when we are to cross the privacy border and get your permission to do so. hakia.com does NOT place a cookie on your PC when you search at our site. We are the very first search engine not to do so. We will always ask your permission if and when we will have to put a cookie in your PC or cross the privacy line. Before I dive into the specifics, let me put this in perspective for you.

For starters, let’s outline users’ concerns about the privacy approach of most traditional search engines – including the most popular ones: Google, Yahoo, and MSN Live. For the most part, the concerns are around these 3 practices:

- Cookies and personal trackage with unique IDs

- Survey of which results you pick, which pages you go to, via embedded Javascript code

- Saving of your IP and search terms

Ok, let’s translate these practices into plain English and explain you a little about the process and its results.

1. The first thing a traditional search engine does is insert a cookie into your computer – which may sound innocent, as it is now common practice on the Web. But the scary thing is this cookie is used, not only save some language or to detect your browser, but also to give you a unique ID and save your search history. Moreover, this cookie is immortal in most cases! For example, Google’s cookies expire in 2035! Search history is noisy, but it still provides useful information to measure a person’s likes and dislikes.

2. The traditional search engines run java scripts to analyze the relationship between your click and what you searched for. You may be able to disable cookies on your computer but there is very little you can do to stop search engines watching your search behavior this way.

3. Last but not least, the big issue of the keeping search archives… We are all aware about various viewpoints on that.

Let me repeat myself. Our approach on user privacy is different. While you surf along at hakia.com, we do NOT place cookies on your computer and we don’t keep track of your clicks for link analysis. This may sound odd as it may lead you to think this would decrease the quality of the search results. Well, we can afford to provide you with cookieless search because of our technology advantage- our linguistic, meaning-based approach to improve search relevancy. We don’t need to know you personally to deduce what you mean. You already communicate it in your query, in plain English, and we are able to understand you thanks to our semantic capabilities. We are committed to maintain the “no cookie” policy at hakia Web search. If we need to use cookies one day, we will ask your explicit permission and you will decide to opt-in or not. You will know and you will have the choice.

That’s the hakia.com difference when it comes to your privacy…

We are humbled and flattered by the interest in hakia and meaning-based search by the tech community. We are working hard to push the envelope of today’s conventions by replacing indexing with QDEXing, keyword search with sentence analysis and more. There are many nuances to this change that take time to understand, and for us to communicate to bloggers-writers who are the conduit reporting to the public what’s new and what’s next in Web search. Please know that we are always willing to spend time to give you a demo and show you what is new at hakia.com plus explain its differences. Drop us an email to showme@hakia.com to schedule a demo. If you are a writer constantly executing Web searches, give us your toughest queries and let us know how we perform. In any case, let’s chat.

We polled you and confirmed that search is still an unsolved problem. A number of innovative companies are attacking this problem from different angles. In our “search for better search” campaign, we have results on top tech blog writers’ thoughts about the search landscape and their propositions. We also organized a poll to ask blog readers what they think. As of May 2nd,the results are as follows:

Poll question: What do you think a search engine should do? (504 votes)

As we push the search technology envelope to innovate, it is great to see the poll’s confirmation that we are responding to the demand! We attack the relevancy problem by delivering more meaningful results, powered by the QDEX and SemanticRank, which is summed up in the top two choices. We are working to bring a fresh perspective on other improvements as well, such as providing a clean and strong user interface. Did you check out our uninterrupted, highlighted sentence presentation, as opposed to broken keywords, that let’s you evaluate your answers directly from the search page, even without visiting the source of information? Michael Arrington’s “give me more control” argument is in the 3rd rank with a significant 10% vote – it’s hard for me to understand this because all search engines already give you a lot of control with the advanced search settings. It is interesting. Personalization is surprisingly unpopular with a total vote of 3%.

And now let’s flashback to a similar Read/WriteWeb poll as an independent source to confirm our findings:

Poll question: What ’search 2.0′ concepts you think stand the best chance of beating Google. (625 votes)

Even though ReadWriteWeb results are more evenly distributed, the rankings are almost the same. Both polls share the same top finding: users want more relevancy.

Web searchers want their search engine to become more intelligent; they want to get exact results to answer their questions. Interestingly, personalization is not that important for them, perhaps because keywords that require personalization are not that many and you can easily solve that problem with adding one more relevant keyword to your query – for instance if you’re looking for java, the programming language and not java, the island, you can look it up with 2 keywords java and programming and you’re done in keywordese!

Another striking outcome is that ReadWriteWeb readers want to see effects of power of masses in their results. Just like del.icio.us does with tag based bookmarks integrated with their search functionality.

Last but not least, the user interface of course…Even if you have the best technology, you can do nothing with a boring and slow interface. On the same token, you can do nothing with the best UI, if your technology does not deliver. At hakia, we try to offer you both, a simple interface and higher search relevancy, to reduce your search time.

Yes these are my findings. Let us know what you think…Thanks to everyone who participated in our campaign, blogged about us and spread the word. Expect to see more in the following months!