"Is there anything else you would like to ask?" Meng Fanqi grinned, he had already discovered that Xi Xiaohu was deliberately examining himself.

It's just that he knows that Xi Xiaohu is the leader of Google Shanghai advertising, not to mention the first, at least the second place.

If his doubts are not completely dispelled, it will be difficult to carry out the work in the future, and it will be difficult for the follow-up promotion and testing to go smoothly.

You must know that Google tests various methods thousands of times a year, and in the end, if the algorithm wants to go online, it still needs to rely on others.

Moreover, Xi Xiaohu is now only an understandable suspicion, not a malicious target.

Xi Xiaohu's expression froze when he heard that he was called Mr. Hu, and he felt that this kid was deliberate. How does this title sound stupid? Tiger, huh.

Meng Fanqi looked at him innocently, what can he do, he can't be called Mr. Xi, wouldn't that rob the old Biden of the rare title in the future.

"It seems that Meng does have a good understanding of our overall strategy, but I don't know what improvements you plan to make." Xi Xiaohu has no opinion on Meng Fanqi himself, if the other party can really make a breakthrough, this will also be a big achievement for him.

In particular, this R&D process seems to be carried out in Google Shanghai, and he is bound to become the person in charge and gatekeeper.

Naturally, the results can get a piece of the pie.

When talking about this, Xi Xiaohu had basically put away his suspicions, at least the young man in front of him was obviously not here to make trouble, and the lower limit of the level had been guaranteed.

The next thing is to see what kind of high opinion he has and where his ceiling is.

"My idea is that the optimization aspect is mainly based on two aspects: mobile devices and artificial intelligence interpretation models." The two algorithms that Meng Fanqi started at the beginning were both methods that Google iterated on in 15 years.

There is a certain distance from the current vintage, and he confirmed that this part of the interior has not yet begun to be developed, but it is not too far away, and the effect will be more predictable.

"The former is optimized for mobile web pages, and it is foreseeable that more and more people will use mobile devices to search online in the coming years. I think a way to evaluate how well a web page looks on a mobile device can greatly improve the search experience for mobile users. ”

At the end of 2013, many web pages are still very uncomfortable to browse on mobile phones, it is very inconvenient to zoom in and out, and many functions are not even supported.

However, there are still very few search engines that have noticed this problem and introduced it into the page ranking.

Just a few years later, the web page on the PC side has been ignored, and the traffic on the mobile side has increased tenfold.

Pushing a website with a poor reading experience to a mobile device is a great waste of search resources.

After listening to Meng Fanqi's detailed technical methods, Xi Xiaohu's pupils shook, completely restraining any doubts.

All that remains is surprise.

"You're pretty clever... By judging the specific visual requirements, it is confirmed that the scale of the web page is indeed conducive to the user's current screen ratio. ”

As an insider, Xi Xiaohu instantly realized the value of this method.

Although it does not directly generate revenue, this kind of care and attention to user experience is more appreciated by him, as a technical leader, than revenue.

"There is also a variant that can be added to this direction, that is, websites with fast loading pages on mobile are preferred." Meng Fanqi's patch is a new judgment added by Google in 18 years.

The mobile terminal is not as stable as the desktop network, and the loading speed of the web page is also a major factor affecting the experience.

Xi Xiaohu smiled after listening to the core context of this patch version. The previous suspicions vanished and were never mentioned again.

grabbed Meng Fanqi's shoulders and said, "You have made these two things, and you are worthy of the results of the group for a year!" ”

"Tiger doesn't ask basic questions anymore?" Meng Fanqi also took the opportunity to quip.

"Alas! What is this called? Xi Xiaohu waved his hand, "I'll test you casually!" ”

The two laughed, and the subtle places in front were revealed.

"The other direction is the introduction of a new version of AI." Meng Fanqi naturally will not forget his old profession, the judgment of mobile adaptation has little to do with AI, and this explanation model can introduce more factors to comprehensively judge the relevant results of a search.

This is a very important technological breakthrough, and Meng Fanqi will also develop the Transformer and BERT methods next year, which is the T of ChatGPT.

At that time, there will be further upgrades to this interpretation model.

Even if Meng leaves Google more than a year later, the algorithm will continue to iterate.

It can even be said that it is a very low-profile version of Miss Bing (GPT-4 kernel), which has a certain language understanding ability enough to help search engines better understand the user's search statements.

However, it doesn't have any features that look like intelligence, so you can't really talk to the user.

Xi Xiaohu's understanding of AI at this stage is not so sufficient, and the language models at this time are basically based on temporal models, and they are all old technologies with a certain number of years.

Therefore, he still knows a little: "Temporal language models may be good at understanding sentences, but will they be too slow to use them at such a large scale?" ”

The so-called time series model may need to be related to the context of the model calculation at a time, and there is no way to calculate it independently, so it greatly limits the parallelism of the operation.

Originally, it was really slow for one person to do 100 questions, but when 100 people did 100 questions, that was a hundredfold increase in speed.

But if the second question must be answered to the first question before it can be done, then there is no choice but to wait.

The so-called one-core work, seven-core onlookers, is a similar situation.

In this context, even if Google has enough computing resources, it is still difficult to deal with this problem of computing time.

To know the latency of searching, it's a lot worse than the latency of opening a web page.

"Mr. Tiger has forgotten what my old job is? The problem of parallel language models, of course, I came up with this plan when I had an idea. In terms of recommended advertisements, Meng Fanqi's details are far less solid and stable than Xi Xiaohu.

But if you want to talk about a breakthrough in the direction of AI, a big guy like Xi Xiaohu is far from being Meng Fanqi's opponent.

In addition to the first two, Meng Fanqi focused on mobile optimization and search term understanding optimization. The last Fred's algorithm is a punitive method that has the most immediate results.

"My last algorithmic strategy, specifically looking for low-value content and all sorts of deceptive ads and all sorts of affiliate links in search results. The most severe punishment is used to force the site to favor high-quality original content and a premium user experience. ”

At this point, the basic framework of Google's main recommendation advertising algorithm 19 years ago has been established.

Just wait for Meng Fanqi to complete the development.