AltSearchEngines welcomes Dr. Hong Liang Qiao, CEO and founder of Lexxe. Dr. Qiao holds a Master’s in Linguistics from the University of Leeds and a Ph.D. in Computational Linguistics from the University of Queensland. He has taught and served as a researcher at Shanghai International Studies University, Shanghai Jiao Tong University, University of Queensland, Uppsala University, Bergen University, Microsoft Research Institute, and University of Sydney. His research has spanned a wide range of work in Natural Language Processing: Part-of-Speech Tagging, Parsing, Computational Lexicography, Word Sense Disambiguation and Corpus Linguistics.

Since 2001, Dr. Qiao has been engaged in commercial high tech development and startups. While Chief Scientist at Monkey Pty Ltd, Australia in 2002, Dr. Qiao successfully developed a search engine prototype using modern Natural Language technology.

In 2003, he founded General Language Machines. Dr. Qiao founded Lexxe in Sydney, Australia and launched Lexxe’s Alpha version in 2005. Lexxe is working towards releasing its Beta version this year.

ASE: Welcome, Dr. Qiao. Please describe what Lexxe does and how it is special/different from other search engines?

Hong Liang Qiao: Lexxe is different from search engines like Google and Yahoo in that it treats search as a Natural Language Processing problem in the first place and Information Retrieval in the second place. By saying so, I mean Lexxe puts considerable effort into trying to understand a query and the text or pieces of information it may match from a linguistic perspective, because what users input in a query and what they get from a search engine is basically language material or those objects (like images or video) via language description.

No one can deny the fact that search is dealing with language almost 100%. But the reality of search technology development today is that it has been largely ignored by computer scientists, who dominated R&D work in search technology.

Unlike many new search engines that try to focus on one aspect of search improvement, such as “News Search” or other vertical search domains like “Health Search” and “Job Search”, or image and video search, etc, Lexxe is directly dealing with search quality improvement, i.e. “keyword search” in a generic search engine. Going forward, “keyword search” will be a major search method. Whoever can do it best will get a large slice of the market.

Keyword-based search fails to deliver in many cases, for example, key member of the ring will return irrelevant results on Google (most are about key rings). Lexxe tries to solve such problems by conducting a “phrase recognition” on the query, so that it is able to deliver better results in most cases. For more differences in keyword search between Google and Lexxe, please visit: http://www.lexxe.com/difference_to_google.cfm.

Lexxe, just like Powerset, was among the few that tried to open up a new window for this trend. Users are at best using what a search engine can provide. So far they use key words to make queries to a search engine. Now the question is- are key words the best way to retrieve information? If yes, why don’t human beings communicate with each other using just keywords? In fact, today’s search technology has proven that key words are just a way of making queries, only because we do not have sophisticated enough language technology. But that does not mean we cannot deliver a practical system capable of communicating with human beings in natural language to get the information they need.

Another strength of Lexxe is Short Question Answering. Lexxe allows users to ask short questions and it is trying to find an answer. At the moment it is able to answer the 6W’s and 1H (who, what, which, when, why, where and how). It is also able to answer affirmative questions, which require either “yes” or “no” as answer.

Question answering is the most “humanized” way of communication with human users. Generally speaking, answering a question for a definition is not a difficult job. The most difficult one is answering a general interrogative question with the answer as either an object or subject in a target sentence, etc. Lexxe is probably the only open search engine in the world right now that tries to answer with exact answers. Such ability is rated the hardest among a few tough Natural Language Processing tasks.

To summarize, Lexxe is good at extracting answers in the web pages rather than retuning it from a pre-made database. This requires very high NLP power to implement it. Lexxe is building a linguistic brain for a search engine, so that no matter what changes take place in the web pages, the results can be dig out.

ASE: What types of searches does Lexxe perform best? Please share some sample searches.

Hong Liang Qiao: We developed a syntactic parser to analyze the structure of the question and tried to understand what a user really wants. Take the following questions as examples and see how many you can answer (these are real questions recorded on Lexxe search engine.)

• When did East and West Germany unite?
• Who is the CEO of amazon?
• What is the moon made of? (joke)
• How many people in a football team?
• What is the highest mountain?
• When did Columbus discover America?
• What is the longest river in Britain?
• Who was the first female space shuttle commander?
• When was the first pyramid built?
• When was president Kennedy shot?
• How many grams of fat are in a cheeseburger?
• How does the Atkins diet work?
• Who was the fifth prime minister of Canada?
• Who invented dynamite?
• Why is the sky blue?
• Why is Mars red?
• In what year was the transistor invented?
• How does an optical mouse work?
• What is the effective range of Bluetooth?
• When was AOL founded?
• When was Cameron Diaz born?
• Who dies in Half Blood Prince?
• Who directed Gone with the Wind?
• Who was the original voice of Mickey Mouse?
• How much does a 20GB Ipod cost?
• Is Melbourne the capital of Australia?
• Is Canberra the capital of Australia?

How many answers did you get right? You can check to see how Lexxe handles them. Some of them are not easy questions, I must say. Of course, there are heaps of questions Lexxe still can’t get right, but we believe there will be ways we could improve Lexxe’s question answering ability in the future with enough resources.

For more such questions, one can visit http://www.lexxe.com/help.cfm.

ASE: How do you monetize Lexxe?

Hong Liang Qiao: We will do advertising just like all search engines are doing and because of our strength in Short Question Answering, we may open up SMS/Text Question Answering business in the future too.

ASE: Does Lexxe work/partner with any other alternative search engines?

Hong Liang Qiao: We may team up with other ASEs to add image and video search to Lexxe, etc.

ASE: How do you market Lexxe?

Hong Liang Qiao: We have a budget to market Lexxe when we launch our Beta version in mid-2008. We will use media to help promote Lexxe. But I still believe word-of-mouth is the most powerful marketing method of all.

ASE: On the personal side- where were you born? What did you study in college?

Hong Liang Qiao: I was born in Changchun, Jilin, China and I grew up in Shanghai, where I studied English at Shanghai International Studies University 1980-1984.

ASE: What did you do before founding Lexxe?

Hong Liang Qiao: I was mainly an academic working in universities in China, Australia, Sweden and Norway. But I gained some industry experience before founding Lexxe in 2005.

ASE: What was the inspiration for Lexxe?

Hong Liang Qiao: think when I finished my MA in Linguistics at the University of Leeds in 1988, I was asked by the lecturer of the Computational Linguistics course what I wished to do with my knowledge in computing and language. I said I would like to use a computer to access an encyclopedia so that I can ask all sorts of questions to it (there were no websites at that time and Internet was mainly used for sending emails). I would never get bored. That was my dream. Now with the Internet available, you get access to much better sources of information than just an encyclopedia.

In 2001, I wrote a piece of code and tested “who assassinated President Lincoln?” and “John Wilkes Booth” was ranked no. 1 as the answer. This gave me a lot of inspiration and after about three years of research I finally got some funding to start coding Lexxe in 2004. Imagine you can talk to a computer endlessly about so many interesting things that no single person can match, isn’t it amazing? It is like you would have an endless source of knowledge and power.

ASE: Who inspires you in the business sector? Personally? And why?

Hong Liang Qiao: I think Steve Jobs is a great business leader, very creative and has a lot of initiatives. He led Apple with huge success. A business in his hand is definitely not the same as in others’.

Personally, I think Chester Carlson, who invented photocopy machines, inspired me more– as a inventor and as someone who strived to go all the way with his dreams and talent. Search engines just need more and more inventions to make them more intelligent for people to use.

ASE: Which languages do you speak?

Hong Liang Qiao: I speak Mandarin (native speaker), Shanghai dialect, a bit of French, Swedish, Norwegian, and English!

ASE: What are your hobbies?

Hong Liang Qiao: I like cricket and rugby league and soccer. I like music.

ASE: Do you like to travel? Where?

Hong Liang Qiao: I like to travel and I backpacked all the way from England, to about 10 countries in Europe, via Russia, Mongolia to Shanghai, when I was young. I liked adventures. I would like to travel to South America, Africa and the Antarctica in the future.

ASE: Thanks for your time, Dr. Qiao. Best of luck with Lexxe.

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

AltSearchEngines is looking for a sponsor for this feature. Your company’s logo, tag line, and link would appear here after every profile. Please email Charles@ReadWriteWeb.com for more information.

This entry was posted on Friday, February 15th, 2008 at 11:45 am and is filed under CEO Views. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.