Baidu's Search Phonetic association function is roughly on the principle is how to thank you!

In the Baidu input Guangzhou below will be prompted Guangzhou, Guangzhou News. I'm thinking Baidu is not going to be some hot keywords, and then use a field to remember the pinyin of these keywords; search the table directly. If it is pinyin, it will be fuzzy to match this phonetic label column. The result is returned when the match is complete. These are just my imagination, as if there is a keyword weighting mechanism. Google can not find the relevant information, is there any other way to achieve what I did not think of it. Help the Warrior Thank you!

Reply to discussion (solution)

Baidu Search when the drop-down menu principle is the same, and then specific is the search technology, do not understand

Again to Lao Xu Showtime ...

I think your idea should be correct.

A small feature, but it's very complicated to do.
1. Lenovo function requires a database, of course, small write a document is OK
2. Each word Lenovo also wants Ajax

Not hundred poison, but when I was in GG, this feature was almost always slow to respond to Firefox prompt page
Understandably, because of the cause of a giant filter, I use GG to turn off this function.
There is nothing special in writing these words, just to remind you that although it looks beautiful, but still can not rely on hardware support, cautious use of fancy things

There's no problem in principle, it's a little tricky.
If implemented in Ajax, then speed is a problem (there is no problem with local testing)
So Baidu to improve speed, will let you install "Baidu Toolbar" by the control to complete

Why installed Baidu toolbar to improve speed?

front and the company search Department of people hit a lot to understand the search engine work principle. There are many words in the

Search engine:

Stop word list, build Justice Thesaurus, thesaurus, kanji-Pinyin thesaurus, suggest.

When you enter a Chinese phrase on a search engine, the search engine begins with a word breaker, and then the words are searched to find out if there is any information associated with each of the lists mentioned above. As you say, you will find the thesaurus of pinyin-Chinese characters. Encounter Guangzhou = Guangzhou, will be automatically translated. Then take Guangzhou first to search.
When you enter an error word, you may be corrected by the suggest of the search engine and prompt you: Are you looking for XXX?

In fact, it's just one of the branches of the search engine that processes searches, and a search makes many requests in parallel.
For example, you enter a short phrase in the search engine. The

Search engine first determines what to search:
1 full sentence
2 standard participle (can be understood as Chinese grammar participle)
3 natural participle (word breaker by word, space, punctuation)
...

Then the table takes each branch, the list of auxiliary vocabularies mentioned above, to refine what will be searched.
Several branches request at the same time, get multiple result sets.
Next is to deal with the problem of sorting, in general, the whole sentence search to get the most relevant results, so the highest weight, it should be ranked first. But the real-world search engine may also have to consider the promotion bit, as well as the content you want to search has more official results (such as your search Nginx,nginx official website should be ranked first). Or Baidu's Baidu promotion, it may be put in front.

This is roughly the case, in fact the logic of sorting is very complex. It determines the sort results based on several dimensions, which they call "curves." When he adjusts the parameters of each dimension, it has an effect on the sort results.

In the Baidu input Guangzhou below will be prompted Guangzhou, Guangzhou News. I'm thinking Baidu is not going to be some hot keywords, and then use a field to remember the pinyin of these keywords; search the table directly. If it is pinyin, it will be fuzzy to match this phonetic label column. The result is returned when the match is complete. These are just my imagination, as if there is a keyword weighting mechanism. Google can not find the relevant information, is there any other way to achieve what I did not think of it. Help the Warrior Thank you!



Principle Steps
(1) obtain pinyin and convert it into the most likely Chinese characters.
(2) The most likely pre-line in this string of Chinese characters or strings.



As to why which is most likely. This is from the data analysis results, the most likely ranked first. Baidu uses more than 100 million times a day, through data analysis of course use more accurate.

Default7 There are no simple search sorting algorithms to introduce.

If the problem is clear, there will be no problem with the interview with Baidu Technology. It can be disclosed that the relational database is not directly accessed due to large traffic and speed.

Upstairs is not the big God of Baidu Ah can be a little more, I want to do a simple. You're a little too revealing.

The former and the company search Department of people hit a lot to get to know the search engine of the general principle of work.

There are many word lists inside search engines:

Stop word list, build Justice glossary, thesaurus, Chinese character-Pinyin thesaurus, suggest.

When you enter a Chinese phrase on the search engine, the search engine will first make a participle, and then use these words, respectively, to find out if there are any related information in the above mentioned list. As you say, you will find the thesaurus of pinyin-Chinese characters. Meet Guangzhou = guangzhou, just ...
Do you have any relevant papers to recommend?

