Hacker News Reader

Comment by notastartup | Hacker News Reader

can you elaborate on search relevance?

is it mostly natural language processing? matching keyword queries to document corpus

Search relevance loosely refers to whether a query returns relevant results. There are different ways of measuring this, none of which is really bulletproof.

say one types 'sex' into a search engine

from that query how does one know user's true intent? Is he looking for a wikipedia article? is he looking to hook up? is he looking for porn? is he referring to the gender? is he just bored and trolling?

since google's reach is part of our lives, wouldn't they have a very good idea based on your previous history, what you clicked on, what search results you clicked on, what you typed in google chrome.

couldn't microsoft have pulled off the same thing?

You don't know for sure what a user means by a query. You can make informed guesses. Typical tricks include looking at what users in the same area have searched for, looking at previous queries (turns out a lot of searches are repeated), looking for "replaceable results" (results that appear on multiple pages), and so on.

And no MS could not have pulled off the same thing. The data MS has about you is not even close to the data GOogle has on you.

> The data MS has about you is not even close to the data GOogle has on you.

Also, users have adapted and are trained to work with Google nowdays, they memorize how to find particular results on Google (even though Google puts a lot of effort into breaking features like "link:...").

Interestingly enough my topic result (search history turned off, but signed in to a Google account) was www.funnyordie.com/topic/sex‎

I've never cared about funnyordie, didn't even realise it still existed. Wonder how they picked that, or if other users results differ.

that is my top result as well, and I haven't been on funnyordie since it stopped being funny around 2008, possibly earlier.

At Google there is an internal metric that always shows Google search quality leading Bing's. That's fine, but I always wondered if Bing has the same internal search quality metrics and they are always ahead?

Tonight I typed Eagles into Google and it showed me all the information I needed about the game that starts at 8:10 pm. That is search relevance. I've noticed Bing doing a lot of this type of thing for the past year or two but Google is definitely the master of it.

Last night I asked Google how much a driving theory lesson costs. It spoke the answer back to me that it had extracted using NLP from the official website. That is the power of their infrastructure.

Since Google knows a lot about me (for eg) by parsing my emails (I use gmail) and knows about the sites I visit too (Chrome and thru Adsense on 3rd websites), Google is in a much better position to provide me with more relevant results(and ads) than Bing, since I don't use much Microsoft's online services.

Google has a lot of ways of knowing what I want to find, which definitely has disadvantages; I get presented with the same ads & results even if I try to find other stuff. I get presented with the same article or ad in different searches numerous times to the point I go on to anonymous browsing / ddg.

The good thing; when just searched for something about UIView and later type something "populating a table", it'll give me SO articles (and such) about UITableView which is what I meant but forgot to type add ios/or uitable.

The bad thing; when i'm doing research, it'll give me the same stuff over and over again and I have to page quite far and add / remove query info / play with search tools to get to different things which might not be the most popular or might not be what Google thinks I want.

The awful thing; the ads... I use ad blockers, but when I get ads (gmail / some sites), they are always the same. For Facebook I just think their algorithm is broken or maybe they have none (I only get ads about dating, cheating, stocks all of which I have no interest in nor ever shown), but for Google they show things I am interested in but I already know all of them, have been there, knew most of them before those ads and they cannot possibly sell me anything. If I wanted something from them I would go there without those ads, so why are they targeting me like that? Show me ads I don't know and if I can use it, I would click.

I'm technical, but I hear the same annoyance from others; a friend of mine rents out houses; the search results when he wants to research the competition are usually almost only his own sites (presumably because he visits them a lot?) and ads for his own properties. He is very non technical, so explaining this all to him is not really sinking in.

This is a great example of search result subjectivity. I don't care about football. I'm annoyed when it rudely intrudes my consciousness. If I type "eagles" into a search engine, I'm looking for the band. The football team has no relevance to me, even as millions of other people tonight are probably looking for exactly what Google showed you.

Yep, and that's why it's important for the search engine to be able to collect information about you! It's not just for commercial ends like ad targeting.

Reply on news.ycombinator.com