Search engines results -- Why the discrepancy?

[xaze]

Hello, does anyone know why searching for a bi-gram such as "信息" produces more search results than searching for an individual character that is part of that bi-gram, such as "息" Some bi-grams will return results in the millions, but individual character searches will return 1/2 that. Is there a special way to search using google when using chinese characters?

Search one

Search two

[roddy]

I'd imagine Google assumes that you aren't searching for a character so much as a word, and as 信息 is more common as a word than 息 you see that reflected in the results.

[jbradfor]

That does seem strange. When I just tried it, it wasn't 2:1, but the single-character hit was slightly higher.

I would assume this is just error in the estimate of the number of page hits. While useful, the page estimate is just that, an estimate, and google has stated that it is subject to error.

[xaze]

I would say it was an estimation error too. However, some of the hits are extremely off or they are about equal, which probably shouldn't be the case. I was going to use the search results as a poor mans word frequency (even though it is not quite a word frequency given that it matches pages and not words) for long-tail chinese words. However, the estimates are so inaccurate it is useless. I did the same for Spanish irregular conjugations and for western languages the estimates are a lot more useful.

[roddy]

There are definitely word factors coming into play. If you do a search for 息 you'll notice that it's not just that character that gets highlighted - 信息, when it appears in the snippet, is also highlighted in red, so there's clearly an awareness of words in the algorithm. Quite how it works I don't know.

[tooironic]

Yes, Google's search methods continue to elude me.

"信息的": 31,600,000

"信息的一": 35,100,000 [??]

"信息的一個": 3,840,000

I don't get it.