Results 1 to 7 of 7
  1. #1
    3 Star Lounger
    Join Date
    Jan 2001
    Location
    Boulder, Colorado, USA
    Posts
    231
    Thanks
    0
    Thanked 1 Time in 1 Post
    We can't use Search for A and B. Can't specify whether, when we have two search terms,
    we want both or either or just the phrase. I entered the two words java quickstart,
    and found that many of the results did not have the substring "quick" in them.
    And, there was no way of specifying that kind of thing.
    Am I missing something that's right there in front of me? I know we used to have
    some of these abilities

  2. #2
    5 Star Lounger
    Join Date
    Aug 2004
    Location
    Connecticut, USA
    Posts
    816
    Thanks
    0
    Thanked 0 Times in 0 Posts
    If the new lounge search is as the previous incarnation's search, a search for multiple terms uses a "+" is used instead of the word "and".

    Search for +java +quick instead of java and quick

  3. #3
    Plutonium Lounger Leif's Avatar
    Join Date
    Dec 2000
    Location
    U.K.
    Posts
    14,010
    Thanks
    0
    Thanked 0 Times in 0 Posts
    Quote Originally Posted by yerubal View Post
    We can't use Search for A and B.
    There used to be 'Help' on searching which (from memory) had explicit instructions. There have also been a couple of threads with pointers in them which I can't immediately locate.

    Meanwhile, try:
    +java +quickstart

    to get answers with both words.

  4. #4
    Silver Lounger
    Join Date
    Oct 2002
    Posts
    1,993
    Thanks
    0
    Thanked 0 Times in 0 Posts
    Quote Originally Posted by Leif View Post
    There have also been a couple of threads with pointers in them which I can't immediately locate.
    Here is one example: [topic=763369]Search facility is really rather poor[/topic].

  5. #5
    Brian Livingston
    Guest
    I just looked at the Help page, and clicked the "Searching" link to reveal IP.Board's explanation of its search features. Unfortunately, like many other things about IPB, the help text appears to have been written by some Invision employee who merely looked at the search form and described what he or she saw. There's no explanation at all in the help text of the syntax to be used.

    Years ago, WindowsSecrets.com wrote its own search engine in PHP. A year or two later, WS added a search of "Windows authority sites": a customized implementation of the Google API. It could be argued that WS puts "too much" help on its search page explaining how the syntax works, but I'll let you be the judge of that:

    http://WindowsSecrets.com/search/

    I'd like to investigate the IPB search engine further. For one thing, I want to make sure we're not expecting too much of the AND operator. In Google, the AND is always implied. For example, a query on boot disc is the same as boot AND disc except that Google has no AND operator. (The word and is a noise word that is ignored.) Google has an OR operator (which must be all caps) to defeat the engine's implied AND, but OR is a wholly different matter. Check Google's "basic search help page" and then its "more search help" page; there's no mention of any AND operator:

    http://www.google.com/support/websea...?answer=134479

    You can't use Google today to find threads in the Lounge, because the Admins decided several years ago that search-engine spiders brought the old, weak server to its knees. As a result, the Admins banned all search engines via Robots.txt, and Claude Almer wrote a special routine that defeated even the ways that Google ignored Robots.txt.

    Windows Secrets has invested in a new server for the Lounge, which will be upgraded to a scalable virtual server cluster by the end of November 2009. At that time, we will lift the Robots.txt ban. About 30 days are usually required for the Google index to include a huge number of new pages at a particular site. (The Lounge holds about 125,000 threads, each of which represents one or more Web pages.)

    We can finally use Google for this kind of search then. But, in the meantime, I'd like to see whether the IPB search algo can be improved. It's ridiculous in this day and age to expect people to understand the plus-sign operator to search for +word1 +word2. Plus signs are fine for specialized searches, but any modern search engine should default to AND. This may be controlled by a switch in IPB. The PHP search code used by WS, for example, defaulted to OR unless the coder specified AND. I know we had to change the default behavior when the WS search engine first went live.

    Thanks for reporting this!

  6. #6
    3 Star Lounger
    Join Date
    Jan 2001
    Location
    Boulder, Colorado, USA
    Posts
    231
    Thanks
    0
    Thanked 1 Time in 1 Post
    Thanks everyone! We've got both clarity and direction, and improvements on the horizon to boot.

  7. #7
    Brian Livingston
    Guest
    Admin Kurt Naber and I have delved into the search engine that Invision provides with IP.Board 3.0. The technology is truly awful, and we'll replace it with a real search engine within one to two weeks, as I describe below.

    In the meantime, we've implemented a temporary workaround on the Advanced Search page. Anyone who clicks the "Advanced" link will now see three lines of text at the top of the page:

    Syntax: In the "Find words" field, precede each word with a plus sign (+) to find posts that contain all words.
    Example: +microsoft +excel
    (This requirement will be eliminated when the Lounge installs new search engine technology in December 2009.)


    We're aware that this does not solve the underlying problem. Nor is it a trivial task to explain Invision's klutzy syntax in the tiny "basic search" area at the top of the page. We'll investigate those problems this week.

    Rather than spend a lot of time explaining the fact that Invision's build-in search engine is weak, however, we'd rather work on installing a new search engine immediately. The only other search technology Invision permits and supports is open-source software known as Sphinx (please see SphinxSearch.com). As soon as we install Sphinx, IP.Board's existing search input forms will automatically post the user's query to Sphinx instead of the default search engine.

    Sphinx will be a great improvement. But soon we will also be able to support queries from Google, Yahoo, Bing, and other search engines. These search engines currently contain few or no hits on posts in the Lounge, because Admins for the past several years used Robots.txt to ban all crawlers. (The old servers crashed or became unusable when a search spider requested every page, which is why the Robots.txt ban was imposed.)

    The Robots.txt restrictions will be lifted in a gradual process. Google bans sites that suddenly have 100,000 more pages than the day before, because the search giant's algorithm thinks the increase denotes a spammy site. The gradual lifting of our ban should allow every page to be searchable in Google and every other search engine no later than Dec. 31, 2009. I'll post a Request for Comment thread within the next 10 days on the process we plan to use to gradually alter the Robots.txt file.

    The "syntax" text shown above is not ideal, but it should at least clarify the problem and its workaround for those who use the Advanced Search function. (Personally, I always use the advanced input page of every search engine, but we can't expect most users to do so.)

    If you have any experience with Sphinx or Robots.txt, please post a reply in a new thread in the Suggestion Box forum. Thanks!

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •