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

Shirt Without Stripes

github.com

Post reply on HN
Print view View original

541–550 of 639 posts

Re: Shirt Without Stripes

#541

Post by bad_user »

munk-a wrote: post #367
njharman wrote: post #363

Earlier quoted context omitted.

less than lay person, but in your example > "no evidence of cancer" and "evidence of no cancer" are very different things. Why is it not as simple "no belongs to the word it precedes" ? like unary operator, ! (not), in typical computer languages.

You need to actually build up noun comprehension though, because "evidence of no remaining cancer" or other qualifier words can greatly confuse the situation. You'd also have quite a bit of fun trying to parse the phrase "no means no" or other usages where "no" is being used as a noun... And for bonus points, folks talk to search engines in broken english all the time so "shirts no single striped" is a totally reason…

The broken English aspect also happens because we know search engines don't understand English anyway.

Re: Shirt Without Stripes

#543

Post by 6510 »

spunker540 wrote: post #522
jefftk wrote: post #512

Earlier quoted context omitted.

> I was about to say there are no such queries There are very clearly many queries with no advertising revenue, because there are many queries that show no ads. Trying some searches off the top of my head that I expected wouldn't have ads, I don't get any ads on [cabbage], [who is the president], [3+5], or [why is the sky blue]. On the other hand, if I search for a highly commercial query like [mesothelioma] the firs…

Thanks for clearly articulating what many people on HN seem to fail to grasp. It’s not that Google got worse over the years at surfacing the obscure content they used to so easily find. That obscure content has gotten completely buried under the mountain of content being published every day, and the cat and mouse game of SEO has evolved so rapidly that the problem space of generalized search is so much harder these d…

It is the same thing! The mistake is obvious. For-profit content is prioritized. Google is the driving force behind the for-profit internet but sadly for google: you cant do organic ranking on commerce. ahhhh The index is now a [very limited] snapshot of the glory days.

You don't have to bother creating anything new unless you have something to sell and are willing to invest (big).

Facebook is actually a pretty pathetic implementation where we can still find content created by normal people. If people made traditional websites in stead of facebook groups and facebook pages NO ONE would be able to find it.

We've witnessed the great obliteration of what was once a nice place and now we have to hear google was not to blame?? The death by a thousand cuts is actually well documented.

We tell you what your site must look like or we'll gut it:

https://en.wikipedia.org/wiki/Google_Penguin Google Penguin is a codename[1] for a Google algorithm update that was first announced on April 24, 2012. The update was aimed at decreasing search engine rankings of websites that violate Google's Webmaster Guidelines[2]

There, this is what the entire internet must look like. We went from indexing to engineering here.

https://support.google.com/webmasters/answer/96569?hl=en

  rel="ugc"
We recommend marking user-generated content (UGC) links, such as comments and forum posts, as ugc.

If you want to recognize and reward trustworthy contributors, you might remove this attribute from links posted by members or users who have consistently made high-quality contributions over time. Read more about avoiding comment spam.

Before this those elaborately contributing got actual credit for it. Do you think you got a choice in it? Google clearly demands you ban credit for comments. OR ELSE!

  rel="nofollow"
Use the nofollow value when other values don't apply, and you'd rather Google not associate your site with, or crawl the linked page from, your site. (For links within your own site, use robots.txt, as described below.)

Woah association! How did we go from linking-to to association? It was important enough for readers but be careful to hide it from google. Such little unimportant websites simply shouldn't exist in our index. We command you to help keep our index clean of such filth!

Then the magical: We wont actually tell you what is wrong with your website! Ferengi Rule of Acquisition 39 "Don't tell customers more than they need to know." Get a budget and hire someone to do SEO. Deal with it, we don't care. No, you don't have any feedback.

Re: Shirt Without Stripes

#544

Post by rmetzler »

zacksinclair wrote: post #195
exhilaration wrote: post #181

Earlier quoted context omitted.

I disagree, "shirt without stripes" is an unusual word choice, not one that our ML models would be optimized for. Try "solid color shirt" and you'll see how much better the results are - at least on Google.

Every human understands that phrase and yet the AI doesn't. That's the gap that has to be fixed.

You have to realize that search is not AI. It's pattern matching. And the string "shirt without stripes" matches really good with "shirt with stripes". Levenshtein distance is 3.

If vendors would use the term "shirt without stripes" than it would match great, but they call it "plain shirt".

Re: Shirt Without Stripes

#545

Post by hrktb »

js2 wrote: post #299
derivativethrow wrote: post #288

Earlier quoted context omitted.

Consider the query, "non-glass skyscrapers", which suffers from the same problem. What do you call a skyscraper like that if you want to refer to it? They exist, but you can't find them using that search term on Google.

I'd call it windowless: https://www.google.com/search?q=windowless+skyscraper&tbm=is...

It comes down to what is considered "windows". When the whole external material is glass, glass panes aren't just windows anymore.

That's the full glass buildings returned in your windowless query.

Re: Shirt Without Stripes

#546

Post by tsimionescu »

snarfy wrote: post #515
DenisM wrote: post #197

This problem is known as "attribution" - you have a "no" or "without" in the sentence, but you don't know where it belongs. One could (and one does) argue that the problem cannot be solved with statistical methods (ML), especially not in any domain where accuracy is required, such as medical recored analysis: "no evidence of cancer" and "evidence of no cancer" are very different things. Zooming out, the language fiel…

It seems like the attribution problem is an English problem. The query doesn't have to be English. https://www.google.com/search?q=shirt+-"stripes"

You've solved the problem of accuracy, but you've now introduced a huge recall problem. Your query will never find a page advertising 'the best shirts without stripes', as it contains the word 'stripes'.

To be useful, Google must solve natural language problems. You can't solve natural language problems by using formal language in sine bits of the problem, at least not until we have a full Chomsky-style understanding of the whole of human language.

Re: Shirt Without Stripes

#547

Post by ghayes »

zaat wrote: post #494
indecisive_user wrote: post #481

Earlier quoted context omitted.

Agreed, it's not that weird of a search. Other, similar queries like "shirt without buttons/collar/pocket" seem to work just fine. The only difference is that those are all physical properties of a shirt while stripes is a type of pattern.

Shirt without paisley - fails shirt without buttons - preety much fail. shirt without red button - as already expected, shirts with red buttons

To be fair, the first and last query would also confuse me if someone asked me for that item. "Shirt without paisley" feels a bit like "cereal without elephants." You don't usually explicitly exclude an element that is relatively uncommon.

Re: Shirt Without Stripes

#548

Post by tsimionescu »

nailer wrote: post #350
Aardwolf wrote: post #306

Earlier quoted context omitted.

> but you don't know where it belongs Yes, the English grammatical rules make it unambiguous where it belongs. This is solvable.

A shirt is an item of clothing. Items of clothing are made from material. Some materials have patterns. One type of pattern is stripes. Seems like a matter for logical inference. At which point it becomes fairly easy to find shirts made from material where that materials pattern is not stripes. But yes, no AI I have seen works reliably on even basic queries like this.

What you've described is not simple logical inference, it is logical inference from common sense knowledge. This is an extremely hard problem, much harder than solving attribution in such simple cases.

Most likely, common sense reasoning will be required to get full natural language processing, since human communication relies extremely often on such reasoning. But building a knowledge base of common sense facts will be one of the hardest challenges ever attempted in machine learning/artificial intelligence.

Re: Shirt Without Stripes

#549

Post by quickthrower2 »

notRobot wrote: post #529
quickthrower2 wrote: post #524

"Plain shirt" works a charm though. What is a 'shirt without stripes' anyway? That could be a shirt with diamonds? Or a plain one? Or a Hawaiian shirt? What is the expected result, can we agree?

> What is a 'shirt without stripes' anyway It is a shirt with anything except stripes.

Cool but what is the search intent? It's a bit like saying "Google gimme something that ain't a ukulele"

Re: Shirt Without Stripes

#550

Post by belorn »

searchableguy wrote: post #504
belorn wrote: post #443

Earlier quoted context omitted.

One can play this game a lot and most results will return expected cultural biased results. A "kind person" is apparently a white girl. A "good person", a white woman. A "bad person", white men. A "evil person", white men. A "honest person", equal mix of white women and white men. "Dishonest person", white men in suits. "Generous person", hands of white women. "Happy person", women of color. "Unhappy person", old whi…

Did you google all of them?

Yes. I thought about words people use in priming studies, usually in order to trigger a behavior, and just typed the word with space and "person" appended.

I did use images.google.se in order to tell google which country I wanted my bias from since that is the culture and demographics I am most familiar with. I also only looked at photos of a person and ignored emojis.

I have also seen here on HN links to websites that have captured screen shots of word association from google images and published them so you could click a word see the screen shot. They tend to follow the same line as above, but with some subtle differences, and I suspect that is the country culture being just a bit different to mine.

Post reply on HN