Lexxxxxe

By dpnickdpnick (1228149762|%a, %b %e at %I:%M%p)

We covered a lot of different search sites in class this past week, and I found a bunch that I really liked. My favorite is Lexxe. Lexxe is new to the industry (it was formed in 2006 by an Australian techie) and it's still only in its Alpha version.

lexxe_logo.gif

I have had a lot of fun using Lexxe's unique features. It claims it uses "advanced natural language technology." From what I have learned, this means it's a semantic search engine: it is a system that interprets search queries using algorithms that work very similar to the human brain. In other words, it tries to decipher your search queries using human intelligence to understand what you are looking for (something every search engine should do)

Questions

I decided to ask Lexxe some questions relating to my term project on Apple Inc. My first question was "What was Apple's 2007 net income?" The answer it gives is wrong: "apple bank highlights in 2007 include: net income of $75.2 million", which comes from the first result. After seeing this, I was a bit discouraged that it got something so easy wrong. However, while the first result wasn't even about Apple Inc. (it was about Apple Bank), 8 of the top 10 results gave me my answer. This was pretty impressive in my opinion, and indicated that it was just a weird mistake because Lexxe confused Apple and Apple Bank (something I could've prevented by specifying my search further).

My next question was "When was Apple founded?" This brought back the correct answer (April 1, 1976). 6 of the first 10 and 14 of first 20 results provided the right answer. The results that didn't have the right answer was because Lexxe was looking at a different company (Apple Bank, Green Apple, caramel apples, etc.) Therefore, I decided to refine my search a little bit and search for "When was Apple Inc. founded?" and WOW! The first 12 results all have the correct date, 1976, in the description! I don't even have to open up the link to find my answer! This is very awesome.

Check out the Lexxe Help feature to learn more about Lexxe's search language.

The Cluster feature

When you create a search, there is a sweet feature called Cluster. Cluster provides similar key words that Lexxe found in your results. For instance, in my last search "When was Apple Inc. founded?", it offers a bunch of alternatives, such as: 1976 Apple Inc. (provides the answer right there), 1976, AAPL stock, Steve Jobs, Apple Computer Company and more. This is a spectacular feature, because it makes refining your search A LOT easier. If for some reason your search doesn't provide the results you are looking for, you can look at the Cluster options and polish your search. Who knows, you may even find your answer in Cluster like I did.

Yuck

lm596318ba61ae83e02ced5ae96cdff012_1218766395Confused_Smiley_Med.jpg

Lexxe does have some major downfalls. One big problem is that it's slow. A Google search for "When was Apple Inc. founded?" took 0.23 seconds, compared to Lexxe which took about 8 seconds. As well, Lexxe only provided 83 results for this search, whereas Google found over 800,000. However, after analyzing the results, it's amazing to say that Lexxe actually provided better results even though it took a few seconds longer. I don't know about you, but I have a few extra seconds to spare for better results. Who needs 800,000 results anyways?

Also, as discussed above it doesn't always provide the right answer. I searched for "How old is the United States?" and it gave me an answer of 5 years. For this search, Lexxe failed miserably. Researching forums online, I found some interesting stories from Lexxe. For example, asking "How old is Martha Stewart?" provides the answer "old" (haha). This illustrates that while Lexxe can be useful, it is in ALPHA version, and therefore is not perfect. This is especially true for questions asking the age of something (it's much easier to provide a date, than giving the age) as witnessed with my two examples.

Consensus

Even though the answer it provides is not always right, I found that the results are almost always very accurate. My advice: Try using Lexxe (with caution) and keep your eye out for the Beta version!

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-NonCommercial-ShareAlike 3.0 License