Comparison of search, advertisement and recommendation
|
Search |
Search Ads |
Show ads |
Recommended |
First Principles |
Correlation |
Return on Investment (ROI) |
User interest |
Other requirements |
Each vertical field Independent definition |
Quality, Safety (Safety) |
Diversity (diversity), freshness (freshness) |
Index size |
~ Level 1 billion |
~ Millions --Tens |
~ Millions |
~ Millions -billion-class |
Personalized |
Fewer individual needs |
-Hundreds of millions of users on the scale of personalization |
Retrieving signals |
More concentrated |
More abundant |
downstream Optimized |
Not applicable ' |
Apply |
advertising is significantly easier than searching in part without the need for sophisticated crawler technology and PageRank. And it is more difficult than the search is that it needs to model the amount of data larger than the search. The main difference between the search, the ad and the recommendation is that their criteria are different, the search is primarily for relevance, and the ads are focused on ROI. For example, for example, search for the United States, then for the search, we must put the United States to the top of the results, otherwise it is unreasonable. But for advertising, if the U.S.-based agency's ad click-through rate is higher than the United States, because of the ROI, it can be the United States to line up the agent in front, and do not need to authority of the United States in the front row.
Some articles refer to recommendations (recommendation) and personalization (personalization) as synonyms, but individuals think that they are different and individuals think that personalization is a recommended guideline, But there are other criteria, such as diversity, freshness and the combination of the three, that will have a very good effect. For example, a user who never cared about the military, but if the war is due to the Diaoyu Islands issue, it is recommended that this news be recommended to the user when it comes to this very important news.
The difference between the recommendation and the advertisement is that the recommendation is a homogeneous recommendation, such as recommending a product when buying a product, recommending other news when watching the news. It is also recommended to optimize the flow (downstream) concept, such as the user when watching the news, according to the recommendation to jump to another news page, and on this news page can continue to recommend, optimize the flow is to optimize the whole according to the recommendation to see the news of the process of CTR. For advertising, the recommended ads click, jump to the target page, there is no way to optimize the downstream opportunity.
Basic knowledge of advertising-search, advertising and referral comparison