Search engine that learns human behavior

Convenience store shelving is famous for being extremely precise.

Carefully selecting hot-selling products based on the data collected by the "POS system". By thoroughly studying the behavior patterns of customers, the height of product shelves and the display location of products are determined.

However, recently, online shopping “EC (e-commerce) sites” are gradually approaching convenience stores.

The sales have stopped increasing just by arranging a lot of products, so we are trying to put the best selling products in the most visible places.

Also, the recommendation engine is actively introduced in place of the clever staff.

We infer customers' intentions and recommend products that are suitable for each customer.

The trend of inferring user intentions is not limited to EC sites.

In the United States, even on major news and information sites, "people who are watching this news are also reading this news," or "people who are reading this blog post are also reading that blog post. Services are increasing.

In a familiar example, the character conversion function of mobile phones is the same.

Simply enter one character and he will recommend (recommend) that "I don't want to enter this word next time."

This is quite smart because my mobile phone is learning my daily activities.

Nowadays, search engines are also eager to infer the intentions of those who search.

For example, in "Google Suggest," which Google provides as a trial, from the moment users start entering keywords in the search window, "Is this the keyword you want to search for?" > And suggest possible keywords and refined keywords.

"Yahoo" is also focusing on research on how to grasp the user's intention from the entered search keyword.

Even if the same keyword is entered, there are cases where the user wants to go to the site that he/she wants to go to, and sometimes he wants to look up information related to the keyword.

So it's trying to detect the user's intent and change the search results accordingly.

Specifically, after hitting the same keyword, a human classifies each of the keywords that are being narrowed down one by one, and the analogy is "why did this person put this keyword?"

"In that case, it would have been better if we could produce such search results," he says, incorporating human ideas into the search engine.

To read the user's intention more easily, it is quick to ask the user directly.

Speaking of Google, if you put "English and Japanese" in front of the search keyword and then enter the English spelling, the meaning of Japanese in the English-Japanese dictionary will be displayed.

Similarly, if you enter the name of a station after entering "route", you will be guided to transfer.

Communication with the computer will be much smoother if the user simply conveys his or her intention.

Search engines are becoming more sophisticated every day.

In the meantime, the distinction between search engines and recommendation engines will disappear, and they must be integrated as a "universal engine."