Super Search Category/Indexing Issues

Forum to discuss the NewsLeecher SuperSearch Service.
• Before posting SuperSearch questions, please read the SuperSearch Usage Guide.
• The current SuperSearch status is listed on the Services Status Page.
Forum rules
IMPORTANT : Be sure to read the NewsLeecher forums policy before posting.
Lips
Forum Moderator
Posts: 3804
Joined: Thu Mar 18, 2004 6:57 pm

Super Search Category/Indexing Issues

Post by Lips » Tue Mar 22, 2005 4:06 am

Mar 20, 2005
We have just re-arranged the way the Super Search engine indexes articles. From now on all article subjects containing one of these substrings, *.jpg, *.jpeg, *.bmp, *.png, *.gif, are put into the "Images" category, articles containing the substring, *.mp3, *.ogg or *.m3u, are put into "Audio" and the rest are put into the "Other category". It is only contents indexed from today that will be indexed the new way. All currently indexed contents are still categorized the old way (by groups).
Spiril, I’m concerned that by indexing based upon the Subject whole posts will be split up into different categories due to the key words in the subjects. I’m starting to find that this is indeed occurring, at least within the Audio category.

I just searched for two different mp3 posts, both in a.b.s.mp3.jazz, and posted on 3/21/05 about 11:30 GMT.

the mp3’s are in Audio.
the sfv and par2 files are in Other.
the jpg files are in Image.

Spiril, I've just sent you a PM with more details about this problem.

--
Lips

Rain
Posts: 33
Joined: Tue Dec 14, 2004 3:25 pm

Post by Rain » Tue Mar 22, 2005 4:47 am

As a supersearch user, i don't really see how the "making different groups" helps in ANY WAY...

I always though that a big category that would put all things together and a "AND" and "OR" statements would be easier to manage and fill everybody's needs...

but hey, that's just me.

coolz
Posts: 4
Joined: Tue Mar 15, 2005 9:47 am

Post by coolz » Tue Mar 22, 2005 10:18 am

Rain wrote:As a supersearch user, i don't really see how the "making different groups" helps in ANY WAY...

I always though that a big category that would put all things together and a "AND" and "OR" statements would be easier to manage and fill everybody's needs...

but hey, that's just me.
I agree with you, maybe you can also create "all" category with specifics categories

Sphere2k
Posts: 9
Joined: Wed Aug 04, 2004 9:53 am

Same here

Post by Sphere2k » Tue Mar 22, 2005 1:07 pm

I thought about this the second I saw the notice on the new page...

the new indexing way isn't working for me.. for the exact reasons you guyz specified..

I was just looking for the new 'Doves' album.. an the mp3 files are under 'Audio', the support files (nfo, sfv, par2 set) categorized in 'Other'.. WTF ? who needs this .. it's to lame ... 2 searches for 1 post ?

I don't want to start downloading headers again... please switch it back to the old way.. and maybe add some wildcard functions.. worked best in my opinion...

Rain
Posts: 33
Joined: Tue Dec 14, 2004 3:25 pm

Re: Same here

Post by Rain » Tue Mar 22, 2005 3:02 pm

Sphere2k wrote: I was just looking for the new 'Doves' album.. an the mp3 files are under 'Audio', the support files (nfo, sfv, par2 set) categorized in 'Other'.. WTF ? who needs this .. it's to lame ... 2 searches for 1 post ?
exactely. If the purpose of having different groups of search is to save resources (smaller groups, smaller machine time to do the search), this actually screws it up real bad, since you'd have to do 2 (and on most cases) even 3 searches to actually get your posts.

Not really that smart, if you think about it.

Driekes
Posts: 61
Joined: Thu Sep 16, 2004 6:58 am

Post by Driekes » Tue Mar 22, 2005 3:54 pm

I hope this 'feature' can be undone. I have the same problem described as above.

User avatar
GekkeKoe
Posts: 201
Joined: Sat Mar 19, 2005 1:29 am
Location: Reg'd NL User

Post by GekkeKoe » Tue Mar 22, 2005 4:43 pm

Same here.. Categories are fine if there's also a way to search all of them simultaneously. Would be cool if you (Spiril) could manage an "ALL" category or reverse the categorizing process.

Maybe checkboxes for Audio/Image/Other with the option of searching one, two or all 3 categories at once would be an idea?

Hoping for a speedy solution, thanks.
NARF!

goost
Posts: 14
Joined: Fri Sep 10, 2004 1:00 pm
Location: I don't know anymore

Post by goost » Tue Mar 22, 2005 5:41 pm

Spiril,

When I'll try to download a movie with a cover, I have to search in Other for the movie and in Image for the cover,which was part of the post. This is not comfortable and takes 2 searches. This is not convenient.

Goost

neo_leecher
Posts: 69
Joined: Sat Jul 17, 2004 12:35 pm

pars and mp3 searches are incomplete due to 3 categories

Post by neo_leecher » Thu Mar 24, 2005 12:45 pm

Hello Spiril,

Why is it that a super search in a.b.sounds.mp3.complete_cd only returns par files and not the files which are uploaded.

search for KAYAK and see what you get. The actual files (the music) are missing in the list

So whats wrong with the super search engine?

i had similar experience in other groups didn't pay attention to it because ss is in beta, but since we are in final mode what version 2 concerns the problem still occurs so i thought lets report it now.

Okeej i found a solution. By modifying the search category to audio, But then the par files are missing again. So the 2 (par files and mp3) are not combined in one search request. This makes the search incomplete. I have to search 2 times. I have to be aware of the fact that the pars are indexed in an other category. Or we have to post the mp3 files as rars to get them together in the super search in the catergory other. This way we circumvent the supersearch gap and make al the files appear in the other category.

Spiril please do something about this. This is not working fine. Actually its a minor in a very nice working program and super search.

Why not returning the results of all categories. Super search is limited to 5000 results. Reduce this amount to 2500, which is also enough and return all hits (with a max of the 2500) in all categories/indexes to make searches complete.

Looking forward to your replies

Thanks
Neo

User avatar
Spiril
Site Admin
Posts: 4278
Joined: Fri Nov 07, 2003 3:11 am

Post by Spiril » Thu Mar 24, 2005 11:04 pm

Hmm, yea I can see the new indexing method isn't working optimal. I'll revert the indexing back to the method within the next couple of days.

Cheers
Spiril

enigmatl
Posts: 37
Joined: Sat May 07, 2005 1:10 pm

Post by enigmatl » Sat May 07, 2005 2:11 pm

(sigh) I had my credit card out and everything. Finally, I was gonna get rid of 1995 looking Newsrover which I reluctantely use because of it's search. While I download tonight and am excited about how cool and flashy it looks and how much better the speeds and other information is delivered to the screen, I come into these foums checking out comments and preparing to buy what looks to be the best usenet program in the world. (sigh) and it IS minus one thing - the searching.

I'm sorry but in the end, the #1 use for me with newsreaders is can I search and FIND without downloading headers and thus, the search is the heart. Everything else is the body. So, while Usenet is the beat up Chevy, it will get me to my destination. So, as much as I want the Ferrari, it needs to get me there too.

So, I respectfully say to the author, this is the year 2005 and in 2005, to be blunt, a search engine is a failure if you don't at least have basic boulean functions and and or. Not is also nice. A literal search is almost useless today so "this string" should also bring up "string this" unless you have it in quotes. This business of category slicing smells more of corner cutting to me (and to other posters) yet your product is not priced in the cheap category. News Rover's search is more powerful and just as fast.

So, am I hear to bust on newsleecher? No way! I'm just here to tell you what a talented author already knows deep down - that if you open up the search engine to it's full power, give us and, or, non-literals unless quoted, and subcategories only if we chose them and I won't walk to newsleecher, I'll RUN. I just don't think that out-of-the-ordinary of a person. Newsrover has many subscribers and I kid you not when I say that a discussion of what was done for the 10.0 release would put you to sleep. Many people on newsrover, I'd guess would love to switch just because we're sick of seeing bytes per second in the year 2005 but I bet I'm not the only one waiting for your rudimentary search engine to become powerful (or for someone else to make one) because the search engine is the only thing making newsrover attractive right now. Would you please enhance your search engine and give us basic bouleans, non-literals, and no more of this grouping? If it can be done cheaper elsewhere, i know you can do it for $29 a year and I for one will be there.

enigmatl
Posts: 37
Joined: Sat May 07, 2005 1:10 pm

Post by enigmatl » Sat May 07, 2005 2:47 pm

After rereading the above, I wanted to make perfectly clear that when I said I was telling you what a talented author already knows" that I was thinking was that I wanted to tell you what you, a talented author already knows." It's only one word but an important one and didn't want you to think for a sec that I was saying someone else was talented and you (the author) are not. and part of the thought process in the comment was that you, as a talented author knows the search needs to be enhanced... Disagre if you want but know I wasn't thinking or meaning when I typed that that you weren't talented as the above sorta looks like I'm implying (and am not) Omitting "you" that second time was purely typo (just like when I used the word "usenet" for newsrover (another oops but that one was minor)

User avatar
Destroyer
Posts: 639
Joined: Wed Feb 16, 2005 3:15 pm

Post by Destroyer » Sat May 07, 2005 3:36 pm

IMHO supersearch is the best thing since... errm.. newsleecher :D

It does need a few better search and filtering options but im sure they will come along sometime soon :)

User avatar
Spiril
Site Admin
Posts: 4278
Joined: Fri Nov 07, 2003 3:11 am

Post by Spiril » Sat May 07, 2005 7:32 pm

enigmatl,

How many days of article retention does the newsrover searchengine have?


Btw,

I was wondering if anyone know how NewsRover performs it searching?

SuperSearch searches through it's own custom index of articles, but it seems a bit like newsrover searches through an index located at a usenetprovider.
bug fixed. no idea how. hate it when that happens. trying to break it again now. will. not. be. defeated.

enigmatl
Posts: 37
Joined: Sat May 07, 2005 1:10 pm

Post by enigmatl » Sat May 07, 2005 11:38 pm

Newsrover's retention is at least 30 days on the searches and it's results come up pretty fast. Newsrover's search is not literal so "this string" will also bring up "string this" and "this.string" and "string.this" and on newsrover, wild card means wild card and not "substring". If I search for thisstring.*, I will get thisstring.r00 through thisstring.r35 for example but I won't get "thisstringrules".. I can also search for "this" and "string" but not "sucks" (though I forget the symbol, I think it's a ^ but not sure)

I suggest the author plays with newsrover and looks at it's search function to see what the competition has. If Newsleecher's search can match or beat newsrover's search (the most important function is the reader is the search) then newsleecher would be the best reader in the universe. I want it to be the best now because newsrover's interphase is dated - and I hope that happens. I can't speak for everyone but I know 4 people who would be newsleecher users in a second if the change happens.

Post Reply