However, this boring UI is what made the search results so great. This slide is about image search. Images ask similar questions. Which one is your favorite. let me introduce a slide about knowledge cards. For example, some knowledge cards require an additional tap to fully open. In the example on This is good UX the left, the additional taps mean the user wants a more detailed breakdown and overview. In the example on the right, it means that the user lacks enough background information. What else is there? How is tapping here different from scrolling down? Users cannot make appropriate decisions. Therefore, taps and clicks are recorded as distinct events. And we need to give meaning to each one. Reference: Google presentation: Google is magical. (October 30, 2017) 5. Log and ranking This presentation discusses the important role logs play in ranking and search. This familiar slide once again reminds us that the source of Google’s magic is two-way dialogue. Search is like a dinner party where everyone brings their own food to share.
This is a great initiative that allows everyone to enjoy
A variety of cuisines. However, this dinner party will only be successful if everyone cooperates little by little. Similarly, search is supported by a huge body of knowledge.. People who come to India Telegram Number Data search are gradually contributing their knowledge to a system that everyone can use. This slide discusses Google’s interpretation of user behavior. This slide contains the following: The log does not contain a clear value judgment such as whether this search result is good or this search result is bad. Therefore, we must somehow convert logged user behavior into value judgments. This conversion process is extremely difficult and has been studied steadily for over 15 years. The reason we’re conducting this kind of research is because value judgments are the foundation of Google search.
If we can squeeze even a little bit of meaning
Out of one session, we can get a billion times more value out of it the next day. The basic mechanism is that it starts with a small amount of “ground verification data” that tells you which search results are good, which are bad, and which are better than that. And then you look at all the relevant user behavior and say, “This is what users think is good, this is what users think is bad, and this is what users like.” Of course, people China Telegram Number are all different and irregular. So all we can get are statistical correlations, nothing really reliable. for example. If a user clicks on three search results, which one is bad? If you clicked on three search results, the query would be difficult, so all of them would be applicable. The challenge here is figuring out which ones are the most promising. Finally, this slide discusses how logs support ranking and search. This slide has the following description: And this is where I warned you. I’m selling something. They sell the idea of log terminology with the
