GameCafe since revamp

[GreenLego]

I saw the brief discussion on Bargain Chat.

Ever since the revamp, it doesn't seem to work as well as it used to. I'm looking for the best price for "Ghost Recon Future Soldier". Search for "Ghost Recon" (all platforms) doesn't show it, and neither does "Future Soldier" (all platforms). I would have thought for a relatively new and big title (not something obscure) that it would have picked it up? I also tried "Tom Clancy" on the off chance that that may work, but it doesn't bring up Ghost Recon.

[Anthony]

Thanks for putting this in the correct location (instead of filling bargain chat). I dont actually use GameCafe so Im not really here to discuss it but its always nice to hear community feedback on other sites.

[steveau]

Thought I'd quickly chip in, as I think it's a fantastic resource.

A few searches recently have been a bit strange. For example, if you type in Assassin's Creed, it brings up approximately 20 different types of games: http://www.gamecafe.com.au/Prices/Assassins-Creed

A couple of bizarre results too, such as 'Assassins Creed 3Pc'. I've also searched for a few games (like GreenLego) that I thought were popular, but that brought up no results.

Anyway, hope this helps.

[wingman]

Guys,

Thanks for the feedback!  you are right about the wierd games like 'Assassins Creed 3Pc'.  I have fixed that particular issue (I think!).   We use a 'edit distance' algorithm for search, so short names and long names are a long distance apart.  You would need to search for 'Ghost-Recon-Future-Soldier' to get a result.  I wonder if we should include 'substring' searches, so it would match like *text* products.... would that work?

[steveau]

The issue is still around. An Assassin's Creed search, for example, still brings up 'Assassins Creed 2Pc' ''Assassins Creed 2 Only' and 'Assassins Creed 3Xbox'.

It makes duplicate listings for the same item. For example, there is 'Assassin's Creed 3' for Xbox 360 and 'Assassins Creed 3Xbox' for Xbox 360.

The other issue is that if you search for 'Assassin's Creed 3' (for example) then it brings up ALL the Assassin's Creed titles, and not just Assassin's Creed 3. To make it a little worse, the original Assassin's Creed is the first search result. Not a massive problem, but not quite streamlined either.

I hope this helps!

[GreenLego]

Thanks Wingman, your suggested search for 'Ghost-Recon-Future-Soldier' did return more results. What does the dashes between words mean? I thought GameCafe was substring search before? 'Ghost-Recon-Future-Soldier' doesn't return many results however. What happens if it is listed as "Ghost Recon: Future Solider" with a colon, or if it is "Tom Clancy's Ghost Recon: Future Soldier" will your suggest search return the results?

on http://www.getonthegame.com.au and on http://www.savemi.com.au/gr-future-soldier-x360.php return lot more results, like the price from ozgameshop ($44.89) where as GameCafe doesn't.

I never even needed to use these other search engines than GameCafe before the revamp.

[Anthony]

Guys,

Thanks for the feedback!  you are right about the wierd games like 'Assassins Creed 3Pc'.  I have fixed that particular issue (I think!).   We use a 'edit distance' algorithm for search, so short names and long names are a long distance apart.  You would need to search for 'Ghost-Recon-Future-Soldier' to get a result.  I wonder if we should include 'substring' searches, so it would match like *text* products.... would that work?

I have to say I am impressed with what you and all the other programmers of such search engines have done. I considered it briefly for Economical Gamer and ruled it out due to the complexities and unreliable nature of page s****ing.

Anyway, I'm a little confused as to why this is an issue for you? I have to admit I had to Google what edit distance was so I don't know that much. But can't you use a pre-built library for search? Wouldn't a Lucene search on an index of game titles and platforms be enough?

EDIT: lol at censoring out s-C-R-A-P-i-n-g

[wingman]

It would be simple to put in a substring search in addition to edit distance, however, it creates bad search result pages.  It is difficult to ensure that search results correctly correlate with the related product page.  Any ideas on how we could prevent searches like 'war' resulting in 500 million results, it would be appreciated.

Maybe we should re-introduce the intermediate page, where the user can select from a range of products after the initial search, and then click through to the product page with the large price list from there...?

[wingman]

Anthony,

Why don't we merge our two business ideas... perhaps we could create a control on your home page to allow people to quickly search, and get price data from our GameCafe servers.... Or it could be directly integrated into the chat box (like some IRC bots are). 

Could you run a poll on the home page, to see if users would be interested?  I think it would complement your site   Having 80,000 odd game prices at your fingertips... updated every day

Cheers,
Andrew

[Anthony]

Any ideas on how we could prevent searches like 'war' resulting in 500 million results, it would be appreciated.

I wouldn't have a clue. I'm one of those don't reinvent the wheel types of people. A .NET implementation of a Lucene search index seems to me like the best option. It will require duplicating all the relevant data in your existing database into the new Lucene index (name, platform). As long as it is fully featured as the Java/PHP versions you should have a wealth of options. Similar to a search engine best matches at the top, then I believe you can set a minimum score so that you squeeze 500 down to a couple dozen, with the top being what the user wants to see. For the record, I've never used Lucene, I just did some reading for one of my own projects.

Why don't we merge our two business ideas... perhaps we could create a control on your home page to allow people to quickly search, and get price data from our GameCafe servers.... Or it could be directly integrated into the chat box (like some IRC bots are). 

I haven't forgotten your original proposal. I've been considering it for the next version of the site but haven't decided which (if any) price indexing site to go with. Plus one could consider my new site in competition with yours, even though the new Economical Gamer is still based on community posts.

One last thing, may want to remove the link to PALGN reviews.

[wingman]

Thanks guys for the comments...  We agree, we really need to provide a better search function, so we are going to work on it !!!   Thanks for the discussion.  Our Revamp was actually primarily about the back-end engine which s****s prices directly from the game websites, which now keeps approximately 70,000 game prices up-to-date every 10 days... which is 4-fold improvement over the previous version of the engine.  During the switch to a faster database and engine, some of the website was re-written, and many of the functions we used to have just dropped off.

We have been focusing on performing site analysis for gathering data, as most game store don't provide the XML datafeeds that sites like getonthegame and savemi most likely use.  These datafeeds are always out of date and never match to site prices, and only 20 or so stores that ship games to australia provide accurate ones.  Our approach scans all websites in our database that have sound website structure and a rich collection of games on sale.

I will post here, once we have the search functions sorted out, and I would welcome any feedback you have to offer.

Cheers,
Wingman

[GreenLego]

Good work wingman! It's excellent to see that my comments has been taken as a constructive criticism, and I eagerly await the updated GameCafe. If I find any more 'odd' search results from GameCafe, I'll post it here.

[jagregory]

We have been focusing on performing site analysis for gathering data, as most game store don't provide the XML datafeeds that sites like getonthegame and savemi most likely use.

Just FYI, On the Game doesn't use feeds. They're unreliable or just not available, as you say. We've crawled all our own data from day one.

Good luck with the aggregation and search, it's not an easy job!