Slashdot Mirror
Mobile App Search: So Broken AltaVista Could Do It
waderoush writes "First-generation search engines such as AltaVista — built when the Web had only a few hundred thousand sites — produced notoriously goofy and spam-prone results. Well, when you search the Android Market for 'restaurant guide' and the top result is the U.S. Army Survival Guide, it begins to seem like we haven't come very far. San Francisco-based Chomp is one of the companies trying to fix mobile app search and discovery by leapfrogging Apple, Google, and the other app store providers. Founder and CEO Ben Keighran, creator of the once-hugely-popular Bluepulse text messaging system for Java phones, says the company plumbs the app stores, the Web, Twitter, and other sources to distill accurate keywords ('appwords') for each app. The top apps at Chomp for the search terms 'restaurant guide': Yelp, Urbanspoon, and Zagat, just as you'd expect."
So an engine called chomp finds restaurants. Paint me surprised :)
---- MISSING MISCELLANEOUS DATA SEGMENT --- [sigdash] trolololol
produced notoriously goofy and spam-prone results.
So that's who Google has between copying lately with their own spam-filled results.
I've been wondering for a while about how Google's market search is so terrible. They're supposed to know how to do search, right?
Right, but I think the motivation here is greatly exaggerated. The search is only "broken" (if you can call it that) when doing search from the phone. Fire up the android market website and it gets much better. I recall reading somewhere that the market now uses google's search for providing search results, maybe someone can confirm (or refute) this.
So what will happen to all those companies if google flips a switch tomorrow and all phones provide better search results?
Also, keep in mind that there are not many apps, as in there are 1M apps for any app store. It's not like you have a lot of quality results to show like when you search the web.
Why would one expect anything different in a civilization where knowledge (like search engine algorithms) is legally locked away indefinitely from the rest of mankind?
Modern copyright is theft of culture from everyone and it retards the progress of the useful arts and sciences.
That's all I need on my N900
"When information is power, privacy is freedom" - Jah-Wren Ryel
AltaVista, from 1995 to late 1998, provided perfectly acceptable search results that you would be unlikely to find substantially different than the quality of Google's results today. While the speed of the search was slower than Google's today, and it lacked the contextual searching that Google has integrated in the last several years (UPS tracking numbers, weather, calculator, Google Shopping, etc.) results could hardly have been described as goofy or spam-prone. AltaVista also provided a number of advanced search options (such as the NEAR boolean) that it took Google a decade to catch up with.
That all went away when AltaVista was sold to CMGI, and the quality of search degraded radically. AltaVista became a shadow of its former, Google-like self. While AltaVista at any time in the aughts was a laughable alternative to Google, the first few years of AltaVista - in terms of layout and quality of search - were in many ways more Googly than Google, and was clearly the service that Google was aspiring to be when they started.
http://chompapps.com/ has been slashdotted. Come back tomorrow.
This seems much more useful than searching Google for "Best Android Puzzle Games" because those results tend to be very, very subjective.
I searched for tablet news in the Android category. Neither pulse or newsr are on the first page of results, but the Google reader app which is a joke on the tablet is the second result. Maybe they should have categories for tablet apps and phone apps. It's not like google search is much better, because the results for "android tablet news app" just brings links to a bunch of top 10 stories and no links to individual apps. Google app market online has a pretty good result though for "tablet news."
Why is Altavista mentioned in the title and the summary when this slashvertisement has nothing to do with them?
Page shows me a bunch of irrelevant news pages. Guess what the top story is?. Hmm.
Stop hatin' all you haters! I HAVE the US Army Survival Guide app and it's the best. Talks all about how to watch the locals to see what's good to eat. Way more accurate that Zagats. Sounds to me like Google's app search is working just fine.
How good does Handango work? I haven't paid much attention but they've been doing an "app store" longer than Apple.
"A person is smart. People are dumb, panicky dangerous animals and you know it." - K
"The top apps at Chomp for the search terms 'restaurant guide': Yelp, Urbanspoon, and Zagat, just as you'd expect."
I think you're confusing what we hope for with what we expect. Those are what we hope for, but the army survival guide really is what we expect from a search.
Google is the king of search, and Android can't search for files by filename.
If you have a ringtone named "smw.mp3" on your phone, you can't find it by searching for Mario.mp3.
You can't find it by searching for smw
You can't find it by searching for *.mp3.
You can find it by searching for certain metadata (ID3 tags in this case).
"Super Mario World" might return a hit. "Koji Kondo" might return a hit.
If you want to search for files by something as bizarre as their fucking file name, you have to use a 3rd party application, or just mount the fucking SD card in your computer.
Of course, MS isn't much better with Windows Vista and 7 - shit takes ages to search non-indexed locations even if you have a pair of SSDs in RAID 0 and specifically use the file: filter to search for a specific file name only. And it'll take about 8 years if you're searching a network location.
FTA sounds like a default OR search, which gives you morehits with more search terms. Set the default to AND and much of the noise can be avoided.
Older web developers will remember meta tags for keywords and site description... use them anymore? Nah, they were from a more innocent age, were it was expected that site owners would limit the keywords and description to accurately describe their site so it would only be found by those really intending to find it.
Ah... happy days.
Google really doesn't have that much to search for in an app submission. They rely pretty much on app owner submitted information. Gosh, we better hope they file accurate keywords and description so only people really intending to find that app will find it... and pigs will fly.
The various app markets are rife with spammers and husslers trying to sell apps no-one needs at outragous prices. It would like trying to index that co.cc domain THAT google STOPPED indexing because it became an impossible job.
But they can't stop indexing their own site.
Their famous fix was to not just look at a site but see how it was linked to. That got rid of the meta tag spam BUT spammers worked around it and with apps, there is no in build linking (remember, the web is all about linking).
MMO Quests are like orgasms:
You may solo them, I prefer them in a group.
It was pretty awesome until Yahoo bought it. Also babelfish.
I was looking at the wikipedia page for web search engine and looking at the timeline along the right; i didnt recognize a single one on there after 1999.
We can start measuring internet time in Before Google and After Google i think.
"Sic Semper Tyrannosaurus Rex."
It makes me want to turn off JavaScript.
Have gnu, will travel.
You just hat to do logical expression search. Bit it had things work well that Google does not have, for example the "near" keyword.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
Yeah, so - I installed this app because of this post. I load it up. Right in the center there is some feature of "flight simulators." I like flying games, so I click on it. The category was loaded with spam apps. Really? At first glance the software is crap and actually has the problems highlighted in this post!
San Francisco-based Chomp is one of the companies trying to fix mobile app search and discovery by leapfrogging Apple, Google, and the other app store providers. Founder and CEO Ben Keighran, creator of the once-hugely-popular Bluepulse text messaging system for Java phones, says the company plumbs the app stores, the Web, Twitter, and other sources to distill accurate keywords ('appwords') for each app. The top apps at Chomp for the search terms 'restaurant guide': Yelp, Urbanspoon, and Zagat, just as you'd expect."
so ... /. is free advertising for startups now? anyone excited to hear the Ben Keighran is the author of text messaging system for java phones? are we impressed?
if they can fool anyone into funding them, good for them. they are providing a service that you have to think google can and will solve in milliseconds.
Honestly, we're going backwards insofar as searching goes. For god's fucking sake, almost 15 years ago, we could do things like proximity searches (only count matches if the words are within N words of each other) and brute-force exact matches of things that would otherwise be stopwords (you know, sequences of words that are individually meaningless and background noise, but refer to something interesting when taken as an exact literal string that has to match verbatim).
My first memory of Google was disbelief that it couldn't even handle a search for something like "AT&T" or "C++" without barfing up results that had nothing to do with either one.
The metadata system is effectively a directory structure where the names correspond to the different fields. They compensate for this rigidity with playlists, but that isn't quite enough sometimes.
The big problem with this system is that you don't always have ID3 info. The non-technical user lacks the ability to modify their ID3 tags, or has them set incorrectly by an automated mechanism which misidentifies the piece. This is to say nothing of unusual formats which can be played via an addin, but don't get indexed properly. (e.g. Windows Media Player can't read FLAC tags) The problem is even worse for video, which usually lacks such tags (I don't even know if the major formats support it.)
I had a song I liked which I got off some guy's site about 10 years ago and had absolutely no idea where it came from (and neither did he). (It was instrumental, so no lyrics to search for.) I only found it's origin a few years ago on a geocities archive. For something like that, filename is the only thing you have. While it might be unusual, you can bet that people have their own names for songs. e.g. Axel F is best known as 'the crazy frog'.
What we need is an indexing system which supplements the conventional model. By all means build the indexing right into it so that there's no need to search for new files to add to the database, but don't force the user to effectively dump all their files in the one directory and rely on the ID3 tags.
Most human behaviour can be explained in terms of identity.
Is this like how searching for Macaroni Grill in Abilene, TX brings up a single Macaroni Grill in Seatle, WA on my iPhone? I know there are at least 2 that are within 200 miles but it insists on showing me the one in Northern Seatle, 2000 miles away!
http://apps.search.yahoo.com/
as soon as google spends a bit of time sorting out their marketplace search, which you'd imagine would be quite easy for them.
Altavista was trying to sell computers running altavista. They were succeeding quite a bit for the young industry. However, altavista cleaned up their results. google gave people what they wanted till they crushed the competition in usage, then cleaned up some. It has always been the working search formula and why sites like Baidu are winning of google (all the piracy downloads they directly link to). It is also why google quit china as they couldn't compete. They key to winning search is always giving beter results, it is easier to do without censoring.