Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Ask HN: Is 'search' a solved problem?

news.ycombinator.com

Post reply on HN
Print view View original

81–90 of 125 posts

Re: Ask HN: Is 'search' a solved problem?

#81

Post by fiedzia »

Far from it. I frequently have issues even with search on Google or Amazon, who invested millions into that. Dealing with people being imprecise in naming and spelling, different contexts and personalized results is still barely touched or non-existent, and definitely not solved.

Lucene provides building blocks, not a solution, and not even all of them.

Search on a small scale may seem working if you have phrases that are mostly unique, like recent movie titles or celebrity names. Go beyond that and its a mess.

Re: Ask HN: Is 'search' a solved problem?

#82

Post by maaark »

tomsmeding wrote: post #77
axaxs wrote: post #64

Earlier quoted context omitted.

Ah! This drives me insane and it's only started recently. I can enter as little as two words in format 'uncommonProperNoun commonNoun', and the entire first page will be results of 'missing uncommonProperNoun'. It's maddening. It's like searching for 'the moon' and getting results back containing 'the'. Does anyone know how/why this started? Is there a way to enforce a query?

I think enclosing the word you really want in "quotes" forces Google to not discard that term.

I thought you were supposed to do "+ImportantWord LessImportantWord". Neither seems entirely reliable.

Re: Ask HN: Is 'search' a solved problem?

#83

Post by Moghammed »

tomsmeding wrote: post #77
axaxs wrote: post #64

Earlier quoted context omitted.

Ah! This drives me insane and it's only started recently. I can enter as little as two words in format 'uncommonProperNoun commonNoun', and the entire first page will be results of 'missing uncommonProperNoun'. It's maddening. It's like searching for 'the moon' and getting results back containing 'the'. Does anyone know how/why this started? Is there a way to enforce a query?

I think enclosing the word you really want in "quotes" forces Google to not discard that term.

It does! And another trick that can be useful in these situations is gaps in quotes: "the * fox" will return (for me): "the main character, Simon Fox", "The Fabulous Fox", etc.

Re: Ask HN: Is 'search' a solved problem?

#84

Post by sebringj »

Elastic and Solr (Lucene) with the help of various other dbs or graph dbs etc get you farther along but you really have to combine machine learning to get things in context which requires more data in certain domains so it requires someone like me that goes things alone to dive into many disciplines. It is not straight forward and results are not yes or no, more like 70% good and the rest is subjective.

The most challenging parts are if you have many dimensions at the same time such as location / full text / user preferences / social filters / permissions etc. These are the problems that make my life suck right now as you cannot simply not have joins for some things or not have graph relations etc etc. Case in point following feeds with many dimensions so you need a pipeline approach in stages.

Re: Ask HN: Is 'search' a solved problem?

#85

Post by elskerpudding »

It's solved, but as many has cited not to a satisfactory degree. Modern search engines have extremely short query deadlines (users won't wait more than a couple of seconds when searching), which gives low precision results.

This book is very informative: https://nlp.stanford.edu/IR-book/

edit: I should say it's slightly outdated because of lack of "big data" and how search companies currently deals with huge amount of data.

Re: Ask HN: Is 'search' a solved problem?

#86

Post by slantyyz »

arthev wrote: post #67

Given that google is the de facto standard ATM and often returns irrelevant results, I'd wager that search isn't solved. "That does not by itself mean there's room for a new search engine, but lately when using Google search I've found myself nostalgic for the old days, when Google was true to its own slightly aspy self. Google used to give me a page of the right answers, fast, with no clutter. Now the results seem i…

>> Given that google is the de facto standard ATM and often returns irrelevant results, I'd wager that search isn't solved.

Search has become a cat and mouse game where every time search is "solved" with an improved algorithm, it is "unsolved" quickly by people looking to game the algorithms.

Re: Ask HN: Is 'search' a solved problem?

#87

Post by mannykannot »

maaark wrote: post #82
tomsmeding wrote: post #77

Earlier quoted context omitted.

I think enclosing the word you really want in "quotes" forces Google to not discard that term.

I thought you were supposed to do "+ImportantWord LessImportantWord". Neither seems entirely reliable.

Google has a habit of silently breaking things that used to work, and as it's silent you don't know what the workaround is, even if you are so lucky as to have one.

Re: Ask HN: Is 'search' a solved problem?

#88

Post by Finnucane »

fiedzia wrote: post #81

Far from it. I frequently have issues even with search on Google or Amazon, who invested millions into that. Dealing with people being imprecise in naming and spelling, different contexts and personalized results is still barely touched or non-existent, and definitely not solved. Lucene provides building blocks, not a solution, and not even all of them. Search on a small scale may seem working if you have phrases tha…

The 'solution' you're looking for also isn't necessarily the 'solution' Google and Amazon are looking for. You're looking for an answer to a question, or to find a thing, they're looking to maximize ad revenue and sales. So the answer to your search is tempered by what gets them the most money. _That_ is what they've spent millions trying to figure out.

Re: Ask HN: Is 'search' a solved problem?

#89

Post by nailer »

diziet wrote: post #30
obelix_ wrote: post #27

We are still in the stone age when it comes to search. Ask Google who played the mens semifinals of Wimbeldon three years ago and Google will tell you it indexed 6 million pages to provide a link that may or may not have the 4 names I am looking for. Why is it doing all this pointless work? And why is it that dumb in 2018? We have got so used to what it does that lot of people have stopped asking questions about how…

I think these sort of queries are solved.. if you know the categories of sites that index information and are able to scroll and process text, images and information quickly. My first query string idea was 'men semifinals Wimbeldon 2015 wiki' and the resulting page contains the list in a nice format. This is because I have the context that wiki pages would contain this sort of information. Google and others are getti…

> if you know the categories of sites that index information and are able to scroll and process text, images and information quickly.

That's a job for computers to do.

Post reply on HN