A Reply to Users
User = Lexxe Users
Qiao = Dr. Hong Liang Qiao
Qiao: We want to thank so many users for their emails, compliments and offer for help. We really felt humbled and inspired by your support. It is a great pity that Lexxe still couldn't allow more people to use it on a daily basis and we hope things can be improved on our site in the nearest future. Since we cannot reply to our users one by one, we compiled answers to some mostly asked questions below. We apologize if your question was not among the following.
Unfortunately, Lexxe has been shut down most of the time recently, only allowing a limited number of users to search everyday. Lexxe was not designed and intended to handle a very high volumn of traffic in the first place. We have been trying to demonstrate the unique features of a 3rd generation Internet search engine to a samller group of people since July 2005.
User: What is cluster?
Qiao: The cluster in the left-hand column on the search result page is one of the three dimensions of search results, the other being the short answer and the search summaries (with link to webpages). Cluster information is like themes or classification of the documents matched to the query. It offers more focused search on specific topics to users. By clicking on a cluster item, Lexxe will search again for you with the original search words and the cluster words.
User: What if there are more answers to a question?
Qiao: You could possibly find them in the cluster. Sometimes, if the answer does not seem to be right, you can also check the cluster and you might find it there.
User: Why the answers I got were often wrong?
Qiao: Lexxe is not perfect in the first place. But we did find a lot of the queries had spelling errors and grammatical mistakes in them. They will directly cause wrong answers, because Lexxe is very sensitive to grammar. For example, someone asked "when was world trade center built?" and didn't get a sensible answer. The reason is the word "center" should better be spelt the American way - "centre". Here is another example about the same subject. The question "when was twin tower built?" should be re-written as "when were twin towers built?". There you go.
Another reason is that if the answer to your question is statistically weaker (or does not have enough occurrences on the Internet), Lexxe is likely to miss the target. Or if some question may lead to multiple correct answers and the one you are after might not be fortunately selected by Lexxe, please check the cluster and it could be there.
Finally, try to use more formal language, such as “US” vs “the United States”.
In case you get a wrong answer, try ask it another way.
User: What is the difference between Lexxe and other search engines that can answer questions?
Qiao: Lexxe is not the first search engine in the world that attempted to answer questions. However, the main difference is Lexxe’s answers, apart from some definitions that it finds hard to retrieve from the webpages out there, come from real texts (technically we call them “unstructured data” as against “structured data”, such as database data, e.g. timetable, name list, etc.) in the webpages, not question/answer databases that have been prepared before hand.
We train Lexxe’s intelligence in understanding questions and texts in English and with some basic knowledge, like the format for time, weight and size, etc. For example, once we teach Lexxe how to understand and find answers to a question like, “When was Boeing founded?”, it can basically answer similar types of questions, as long as some instances of the answers can be found on the Internet. We never ever have to manually prepare any question/answer pairs in the database to answer a search engine query.
In short, Lexxe finds answers on the fly directly from webpages.
User: Is Lexxe’s keyword search method different from others?
Qiao: Yes, Lexxe currently uses linguistic analysis on the keywords and it tries to recognize phrase(s) in them, before executing the search. That way, keyword-based search accuracy rate is significantly raised.
User: Is there someone behind the search engine answering the questions?
Qiao: No. Lexxe is fully automatic.
User: Why is Lexxe slower than other search engines?
Qiao: Lexxe carries out multiple tasks on the fly for one search, e.g. question understanding, search pattern conversion, phrase recognition, clustering, short answering retrieving, etc, which is a lot more than a conventional search engine. The system Lexxe is currently using also needs to be upgraded with higher speed programs and hardware environment..
User: I found some answers offensive. Can you remove them?
Qiao: Lexxe is a search engine. It only searches for information relevant to your questions. The answers come from contents written by other people, not Lexxe. Therefore, Lexxe would not be responsible for what it finds. Please check our Terms of Use.