Search needs work.

[ap3c0]

I recently decided to try newsbin again after having migrated to newsleecher for several years. The main reason I switched in the first place was SuperSearch. Due to all the current issues with SuperSearch and the news that the develpoer had decided to abandon his customers and move on to "bigger and better" projects, I came back. I tried the search function of newsbin, all 10 free searches of it, anyway. The search function in its current form needs a lot of improvement. I would like nothing more than to see a true competitor to Super Search and this could be it, but with a lot of work. The search menu needs more functionality. The basic functions that make Super Search what it is, and some added features that us long time users have been requesting for years. The Newsbin search is too expensive. Even if improvements were made, $60 per year for unlimited searches is far too expensive.

[Quade]

So, what you're saying is, don't bother working on it because you're not going to pay for it anyway? I don't consider $0.16 a day to be particularly expensive. At least you don't have to worry about Dex or I bailing on you.

If you have some specific functionality improvement suggestions, we're happy to listen. I was actually adding some interesting things to Newsbin that should gravitate to search. Compacting JPG and other non-RAR sets for example.

[ap3c0]

So, what you're saying is, don't bother working on it because you're not going to pay for it anyway?

Wow, I don't recall saying that at all. In fact, I just reread my post and I'm sure I didn't say that. Quite the opposite actually

I would like nothing more than to see a true competitor to Super Search and this could be it.

At $60 per year, newsbin search is the most expensive search service available. Newsleechers unlimited search is $29.99 per year and that includes use of newsleecher without having to purchase it first. Personally, I'm not going to pay $60 a year for use of the search function. Failing newsleecher, I'll simply continue to use one of the dozens of free search and nzb services available. Of course,that's just me. I'm sorry if I hit a neurve with this post, however, the only reason I took the time to write is because I would like very much to see newsbin search succeed. I'm just not sure it can at its current price point. That's just comming from someone who has spend the last 5 years or so on the newsleecher forum, and more importantly the last few months that Spiral has disappeared. There are a lot of angry newsleechers over there. You can either benefit from that, or you can charge twice what they do and risk antagonising a huge instant customer base. My 5 cents.

[richy99]

well you said it needs improvement and is too expensive for you to pay for, you cant really compare newsleecher and newsbin on the price, you pay a ONE time fee of $35 for newsbin for LIFE with full upgrades, this is not like newsleecher which is $30 per year and you are limited to the latest version your key was available for if you decide not to use it, plus v5 of newsleecher has been beta for over 2 years now, in that time we have had about 5 or 6 major revisions of newsbin

i have used both services and both have their merits and both have their faults, is it worth it tho? yes it is, there is also another tier of $30 per year which you get 300 searches per month

your comment about it antagonising the newsleecher users that might move over, i dont see how that can occur as both services are charged and the support for newsbin and the search are far greater that what it is for newsleecher, you dont have to use the service, its an optional service that the guys run and charge for

[ap3c0]

I can see that you guys don't take critique well here, so I'm just going to say good luck with your program and leave it at that.

[dexter]

I can see that you guys don't take critique well here, so I'm just going to say good luck with your program and leave it at that.

If you have some specific functionality improvement suggestions, we're happy to listen.

How can this be interpreted as not being able to take critique? The author just encouraged you to make some specific suggestions.

[Andreas]

Okay, I'm reviving this thread, because I also have some critique about NewsBin's search function. You asked the thread starter to make some specific suggestions. Here is one:

In NewsBin, when you search for "man 2 man" (with the quotes), a disco band from the 80's, you will get a lot of findings, but none from that band. In contrast to that, when you enter exactly the same search term in NewsLeecher, you will find a lot of songs from that band. I could provide screen shots from both NewsBin and NewsLeecher and post them here, if you want, so that everybody can compare the search results and get a clear picture.

As you already explained, NewsBin drops any search terms that are 2 characters or less. There may be good reasons to do this, but in the above case, this policy clearly leads to poor search results. So there is room for improvement.

Don't get me wrong. There are a lot of things which NewsBin does better than NewsLeecher. For example, with NewsBin I only need 8 connections to reach the same download speed as NewsLeecher with 50 connections. But for me, a good search function is essential, so currently I am not sure which program's license I will renew. If the above weakness gets improved, I think it will be NewsBin, otherwise I'm afraid I will have to stick with NewsLeecher.

I hope this post won't get deleted for mentioning another newsreader and giving a concrete search term, but otherwise it would be really difficult to discuss things, and that's the purpose of this forum, isn't it?

[dexter]

Finally. Thank you. You are the first person to actually make a suggestion and show an example in this thread. The design choices we made for our search engine make your specific request difficult. We use a really fast full text search algorithm that is optimized around keywords. Sending in a quoted phrase doesn't lend itself well to it. Quade and I are discussing what we can do about it though.

[Andreas]

Also here in this thread I would like to say Thank You for implementing the exact matching mechanisim of quoted search terms so quickly. Even now (with the small oddity mentioned in the other thread), NewsBin's search function is already better than NewsLeecher's, due to the longer retention time in the Erotica category. So my decision is clear: I will not renew my NewsLeecher license but stick with NewsBin, even if it's a little more expensive. The better support and the better overall functionality are worth it.

[dexter]

Cool, good to hear.

[Andreas]

Hi Dexter,

here are two more small issues with NewsBin's search function which could/should be improved.


1. NewsBin sometimes returns results of the following form:

something.part20.rar
something.part01.rar
something.nzb
something.sfv

In contrast to that, NewsLeecher merges these four lines to one line (something), which is better, because they belong together.


2. For some reason, some postings on the Usenet do not contain spaces where there should be spaces. When you enter the search term with spaces, NewsBin does not find these postings, but NewsLeecher does.

For example, when you search for

autom obile

NewsLeecher finds all postings containing automobile, but NewsBin finds nothing.

Of course, the above example by itself does not make much sense; it's just to illustrate the problem. But there were real, relevant instances where this behaviour of NewsBin has bothered me, i.e. where NewsBin didn't find postings I was interested in.


Again: These are only small problems, but nevertheless it would be nice if you could fix them.


Best regards,
Andreas

[dexter]

For the first one, I'd need the specific post to take a look and see why they didn't combine properly. This should be the exception and not the norm. If you can PM me a couple of examples where this happens, I can look into it further.

For the second one, you can now search for the exact term by quoting it. Try something like "autom obile". I understand that you would like these things to show up automatically but the boolean search algorithm we are using is very literal and would see those as two different keywords. Quoting is the best alternative I can provide.

[Andreas]

So my decision is clear: I will not renew my NewsLeecher license but stick with NewsBin

I hope that I didn't spoke too soon. At the current point in time, the search function of NewsBin has some problems which lead to postings not being found. Postings not being found is a serious issue in my opinion. In order to discuss these problems here, I have chosen christmas songs which should not be copyrighted bla bla


Problem 1:

Search term to demonstrate the problem:
bla bla

NewsBin finds nothing. In contrast to that, another newsreader does find the desired mp3 file with the very same search term.

The problem here is that NewsBin doesn't handle underscores well. When you change the search term as follows, NewsBin does find the desired mp3 file too:
bla bla

However, it is not a practical solution to try all combinations with and without underscores just to make sure that you don't miss something.



Problem 2:

Search term to demonstrate the problem:
bla bla

NewsBin finds nothing. In contrast to that, another newsreader does find the desired mp3 file with the very same search term.

The problem here is that NewsBin doesn't handle subjects/file names without spaces well. When you change the search term as follows, NewsBin does find the desired mp3 file too (note the missing space between bla and bla):
bla bla

However, it is not a practical solution to try all combinations with and without spaces just to make sure that you don't miss something.



Problem 3:

Search term to demonstrate the problem:
bla bla

NewsBin finds nothing. In contrast to that, another newsreader does find the desired posting with the very same search term.

The problem here is that under certain circumstances NewsBin compacts all titles of a music album into one single posting. In the above case, you can get the desired mp3 file as follows: Search for bla bla, then right-click on the search result, Uncompact Search Results, then search for bla bla in the search results. It is obvious that this is not a practical solution.



Dexter already knows at least two of these problems. I am posting this summary here as a reference, hoping that these problems will be solved.

[Quade]

Copyrighted or not, it's specific and I don't allow specific content here. You were discussing this in private with Dex with specific stuff. Examples that mean nothing or descriptions of the problems are fine.

In the US, copyright is pretty much perpetual. We'll all be dead before a current song goes out of copyright.

[dexter]

Yeah, we're working on it. I don't know that we can address all your issues but we are actively looking at the more serious ones you pointed out.

[Andreas]

Copyrighted or not, it's specific and I don't allow specific content here. You were discussing this in private with Dex with specific stuff.

Sorry, I did't want to make trouble.

Note that I was not discussing Problem 1 with Dexter in private, so this problem may be new to him. Since you deleted my example search term in the above posting, it may be unclear now what the actual problem exactly is. If you want, I can send it to you by pm.

[mho]

It would be great if there were some boolean filters (or if there already are, they'd be more obvious:-))
If I'm looking for "foo" and "foobar" is an extremely common subject, I have yet to find a way to pare down the false hits so the "foo" hits (except by pure luck!) are not crowded off the returned results by "foobar" hits. Ideally, I'd like two search boxes and a NOT/AND selector (don't think OR would be needed), but I have no idea what's possible with your full text db. (Alternatively, allow some && syntax in one box; the drawback being that there is no obvious NOT-syntax as "!" and "-" are overloaded enough and "^" should probably be reserved for anchoring possible future regexps.)

- mho

[Andreas]

If I'm looking for "foo" and "foobar" is an extremely common subject, I have yet to find a way to pare down the false hits so the "foo" hits (except by pure luck!) are not crowded off the returned results by "foobar" hits.

Do you know this page? http://isearch.newsbin.com/iSearchHelp.htm

The search term

foo -foobar

should do what you want. In practice, you will still see some results containing foobar, but this is just due to some shortcoming in the current search engine.

[mho]

Do you know this page? http://isearch.newsbin.com/iSearchHelp.htm

The search term

foo -foobar

should do what you want.

I tried that, but in my tests I didn't see any difference in the number of false matches returned when I added "-foobar". (Try searching for "gir" and "gir -girl" - the difference is hard to tell:-))

- mho

[Andreas]

Try searching for "gir" and "gir -girl" - the difference is hard to tell:-)

Hmm, you are right, gir -girl really works bad. I had tried a few other terms, and they worked better. For example, auto -automobile works quite well.

You can improve the results of your girl example with the search term gir -girl -girls -girlfriend -girl.rar -girl.zip -girl.mp3 -girls.zip, but it is still far away from perfect.

[Andreas]

January 16, 2011, 11:20am Internet Search Updates...

Thank you! Muuuch better now!

Problem 2 of my above posting and the problem with the "gir -girl" type of search still exist, but as far as I've understood, this is due to design of NewsBin's Internet Search and cannot be changed easily, right? But this type of search term isn't handled well by other search engines either. For example, binsearch.info, which is normally very good, fails here too. The only exception I know is NewsLeecher. Somehow they managed to solve this problem. But NewsLeecher has other problems, as you can see in their forums. So the perfect search engine just doesn't exist.

[dexter]

Awesome, good to hear.

Had a little hiccup after turning on the new database, forgot to switch the results database over so some people will have seen "no results" errors when trying to download from search results. This has been resolved.

[mho]

January 16, 2011, 11:20am Internet Search Updates...
the problem with the "gir -girl" type of search still exist, but as far as I've understood, this is due to design of NewsBin's Internet Search and cannot be changed easily, right? But this type of search term isn't handled well by other search engines either.

Not sure there is an easy way to fix it; maybe post processing is the only way (pipe the results from the search engine through "grep -v" (pardon the unixese :-)) before returning to the client.)

- mho