"Fast" and "slow" search: two different approaches to search the site
Good afternoon friends.
Today I would like to talk about two approaches to finding information that I use on my site. Conventionally, I call them "fast" and "slow" search (although you can come up with other terms, but I like these). In general, the idea is quite simple: a “fast” search is one that gives information “here and now” , and a “slow” search is extended in time , for example, daily, weekly, etc. Each approach has its own target users, and both complement each other very organically.
')
Theory
For understanding, let me give you a vivid example: let us be interested in the work of our president. To do this, we can easily score in Yandex or Google "Medvedev official meetings of the Kremlin" and they will helpfully give us a lot of links. This is a “fast” (instant) search. And now let's imagine ourselves in the role of a political scientist. In theory, every day he needs fresh events about the president - with whom he met, what he said, etc. Walking for this every day in the search engine - a thankless task. It would be great if you could score specific search queries once and for all, and that the search robots themselves (!!) selected fresh materials at regular intervals. This is a “slow” search.
In theory, it looks "tasty", so we will try to implement it on a specific site. I will illustrate on the example of my website , which has a book focus (second-hand books, book exchange).
Practice
"Quick search
The implementation of a quick search in one form or another is on any site. One or several html-forms, according to which the SQL-query is formed, is executed, then the results are processed and presented to the user in a convenient form. After completing such a search, the user sees the current state of the book base (books of a particular author, books with a specific title, prices, exchange / sale conditions), can order something, etc.
"Slow" search
The key question: what to do if a quick search did not return results? Let go of the frustrated user to other sites? No, this will not work ... We must give him the opportunity to leave his wishes, and as new books arrive, produce more and more new searches until the necessary books are found.
As a basis, as you can guess, all the same quick search is used, which iterates through all the search words entered by users and forms a table of results. In order not to strain the hoster, I do such a search on a local copy of the site (on my home computer), and then upload the results back to the site. After that, the mail robot sends an e-mail notification: “Hello, at your request a new book was found ...”
From the user's point of view, everything looks very convenient: just enter your requests once, and then you will receive messages in the mail. As the saying goes, "we work for you to rest." :-)
Some technical details: the launch frequency is once every three days, from about 60 requests being processed (the service has just started) for three or four new books are required (a good percentage). The number of entries involved in the search (i.e., new ones that appeared in 3 days) is about 150. The search is conducted by two separate fields - authors and book titles (in this respect, everything is simple - just search for the occurrence of the substring). In general, it is still quite modest, but, I feel, it will continue to grow.
findings
I suppose the presence of both types of search on the site gives its advantages. For users, as I noted, this is the convenience of finding information. For the owner of the site - maintaining the constant interest of the audience. Of course, searching through books is a fairly simple option, but I think that if you wish, you can also implement more complex ones (in particular, using full-text search).
It would also be interesting to hear if anyone has come across such things? (I personally saw this in Yandex. blogs, when you can subscribe to rss for a specific request)
PS I cannot give a direct link to the service, since it is available only to registered users. Only on the main page.
Today I would like to talk about two approaches to finding information that I use on my site. Conventionally, I call them "fast" and "slow" search (although you can come up with other terms, but I like these). In general, the idea is quite simple: a “fast” search is one that gives information “here and now” , and a “slow” search is extended in time , for example, daily, weekly, etc. Each approach has its own target users, and both complement each other very organically.
')
Theory
For understanding, let me give you a vivid example: let us be interested in the work of our president. To do this, we can easily score in Yandex or Google "Medvedev official meetings of the Kremlin" and they will helpfully give us a lot of links. This is a “fast” (instant) search. And now let's imagine ourselves in the role of a political scientist. In theory, every day he needs fresh events about the president - with whom he met, what he said, etc. Walking for this every day in the search engine - a thankless task. It would be great if you could score specific search queries once and for all, and that the search robots themselves (!!) selected fresh materials at regular intervals. This is a “slow” search.
In theory, it looks "tasty", so we will try to implement it on a specific site. I will illustrate on the example of my website , which has a book focus (second-hand books, book exchange).
Practice
"Quick search
The implementation of a quick search in one form or another is on any site. One or several html-forms, according to which the SQL-query is formed, is executed, then the results are processed and presented to the user in a convenient form. After completing such a search, the user sees the current state of the book base (books of a particular author, books with a specific title, prices, exchange / sale conditions), can order something, etc.
"Slow" search
The key question: what to do if a quick search did not return results? Let go of the frustrated user to other sites? No, this will not work ... We must give him the opportunity to leave his wishes, and as new books arrive, produce more and more new searches until the necessary books are found.
As a basis, as you can guess, all the same quick search is used, which iterates through all the search words entered by users and forms a table of results. In order not to strain the hoster, I do such a search on a local copy of the site (on my home computer), and then upload the results back to the site. After that, the mail robot sends an e-mail notification: “Hello, at your request a new book was found ...”
From the user's point of view, everything looks very convenient: just enter your requests once, and then you will receive messages in the mail. As the saying goes, "we work for you to rest." :-)
Some technical details: the launch frequency is once every three days, from about 60 requests being processed (the service has just started) for three or four new books are required (a good percentage). The number of entries involved in the search (i.e., new ones that appeared in 3 days) is about 150. The search is conducted by two separate fields - authors and book titles (in this respect, everything is simple - just search for the occurrence of the substring). In general, it is still quite modest, but, I feel, it will continue to grow.
findings
I suppose the presence of both types of search on the site gives its advantages. For users, as I noted, this is the convenience of finding information. For the owner of the site - maintaining the constant interest of the audience. Of course, searching through books is a fairly simple option, but I think that if you wish, you can also implement more complex ones (in particular, using full-text search).
It would also be interesting to hear if anyone has come across such things? (I personally saw this in Yandex. blogs, when you can subscribe to rss for a specific request)
PS I cannot give a direct link to the service, since it is available only to registered users. Only on the main page.
Source: https://habr.com/ru/post/70656/
All Articles