The search quality evaluation usually involves several indicators:
- Relevance-ask trained people to assess whether the first few results of each engine are relevant. The source and brand of the results are not referenced in the evaluation.
- Index Scale-each engine knows its own size, that is, the number of webpages (excluding duplicate pages), but does not know the size of the other side. However, you can crawl the index based on the search results of the two sides, I learned how many web pages I have, or I have, and I learned about the index scale of the two engines. One challenge is that a large index may cause a decline in relevance (because some long-tail results are sorted too frequently ).
- Speed: the time when the search term is input and the result is obtained. Many tests tell us that the speed of 0.2 seconds will lead to a gap in user satisfaction and the frequency of future use.
- Freshness-it is the ability to climb to new content, and there must be a certain scale (only news content is not new enough ).
When I first joined Google in-, Google compared with Baidu:
- The relevance is two points ahead of Baidu (about the gap between today's English Google and Microsoft, that is, not big), but I understand that Baidu was ahead of Google in this regard. This is possible because each company's internal evaluation is different, just as Microsoft thinks that English is more relevant than Google's.
- Google is large, but it does not have much important content (such as forums ). (Chinese indexes are compared here, but Google indexes are actually stored in all languages around the world, so any search may search for results in any language or country)
- The speed is far slower than Baidu.
- Freshness lags behind Baidu.
After two years of hard work, Google compared with Baidu in 2008:
- The correlation is far ahead of seven points (about the gap between Google English and Yahoo ).
- The index size is about 10 times that of Baidu (of course, there is no difference in most common search terms), and it makes up for some crawling errors and gaps.
- The speed is about the same as that of Baidu, although many servers are not in China.
- The freshness is within 6 minutes, that is, a web page can be searched 6 minutes after it is launched (if the PR value is high enough ).
After discovering Google's progress, Baidu began to spend more time improving the search quality and improving its relevance and index scale. Of course, Google also launched Google instant, real-time search, and universal search. Today, I believe that Google is still ahead of the market, but the Chinese team has not made any Chinese search in the past year. The gap should be narrowed, and the progress must not be more than 2008.
Finally, in addition to the above scientific evaluation, we also need to consider some other factors:
- The above assessment is for experienced and highly educated people. The higher the educational level, the more I like Google. In the doctoral community, Google is far higher than Baidu, but with the decline in education, the resolution capability also declines, and there will be no difference among the high school students (there is no brand factor here, there is no brand evaluation), while the high education level accounts for a small proportion.
- If a brand is added, users will think that Baidu's accuracy exceeds Google's, even when the quality of search is the most different in 2008. That is to say, if you don't see the brand, if you invest 70% of Google's products accurately, you may only invest 45% of Google's products in addition to the brand.
- The above evaluation did not take into account the influence of posts, knows, MP3, etc. These functions are included in the search results, making Baidu more recognized, thus improving its perceived quality.
- Google has a lot of results today, which is fatal to a search engine, because most users blame Google for this phenomenon, thus affecting Google's "quality" in users' minds ".
Is the quality of Google's simplified Chinese search results catching up with Baidu?
