Categories?

I'm really sorry if this is a stoopit question, but what are the "categories" for and how do you use them? I mean, i make a category and Yay! I have a category... now what? I have this creepy feeling it's supposed to be self-evident...

And while i'm here, i'm wondering if Sphider or mod thereto will display contextual results like Google? You know, like you get the bit of text you were searching for, in its surrounding context. I have thirty pages of ads but all the pages have the same title, so the Sphider results are sorta ambiguous and unhelpful when they only show the page title and URL.

Thanks,
-jb

Categories become useful if you index a large amount of sites with most different content. Building categories will force Sphider to deliver the search results more accurate, pre-selected or call it limited, as it will present only results of those pages that are associated to the actual category. But be aware that the Suggest Framework does not care about categories. It will present all possible results, also outside the actual category. If selecting one of the suggestions that is not part of the actual category pages, it will cause the message "The search \query\ did not match any documents".

Result listing does not only present 'Title' and 'URL', but also the query surrounding text (250 character fixed for original Sphider, selectable length for Sphider-plus) or meta info of the pages presented in result listing. You may select in Admin Settings.

Tec

I'm sorry, still not getting it. How do i associate pages to categories then? And i only index one site; is it not useful then?

I don't see anything in my settings remotely resembling "query surrounding text" so i don't know what i should be doing to get said surrounding text in my search results. No matter what settings or values i choose my results have three lines:
1.) Page title as hyperlink, (visually identical),
2.) Page description, pulled from meta-tag, (also identical), and,
3.) URL and page size (almost identical)

As noted, the main concern is having my results mean something to a user, or at minimum look different from one another so a user can tell them apart. But since they /are/ classified ads, the very best thing would be if i could make this bit work, or somehow associate results with categories.

Based on my complaints, is it possible my installation is broken, and the things i need are not present or not working? I'm testing on a local Windoze machine before putting in production. It'd be a drag if my questions didn't make sense becuase something expected isn't happening...

<<< i only index one site; is it not useful then? >>>

No, if you only index one site, categories make no sense for you. Because only different sites could be associated to the categories. To be defined in Admin / Sites / Options / Edit


<<< 2.) Page description, pulled from meta-tag, >>>

This you may influence by Admin settings. For original Sphider in
Admin / Settings / Search settings
uncheck the checkbox:
'Show meta description in results page if it exists, otherwise show an extract from the page text.'

In Sphider-pus this item is called:
'Show Description Meta Tags, if they exist, on results page.'

Sorry for this (stupid) info, but sometimes . . .
Please do not forget to confirm your changes in Admin settings by pressing the 'Save settings' button.

If unchecked, you should see the 'surrounding text' in your result listing. If you still do not get the surrounding text, your installation might be corrupted.

Tec



Edited 1 time(s). Last edit at 07/22/2008 08:57PM by Tec.

Thanks very much for the reply. I actually tried what you suggest before i came back but all my results display the very top of each page, i.e., no context with the actual search result.

Could this be because it's all actually the same page with different includes? Google results work as i would wish for, so it didn't occur to me this might be a problem.

Thanks again
-jb

As I don't know the content of that site / pages it is difficult to answer. Perhaps you could let me know the url of the site you indexed, so perhaps I will be able follow your problem.

Tec

I'd much appreciate that.

[mywebthings.com]

I indexed that site with an index depth of 1 (in order to save time).


Afterwards I searched for "Original"
The result listing showed to me:
1 Wise Buys Ads & More ~ Home
10x 1987 FORD PICKUP 4x4 F250, Camper Special. Super Nice, Clean, <bold>Original</bold> Upholstery, Pioneer CD, Custom Wheels, Rear Air Bags, Extra Set Of Wheels & Tires New Condition. $2995 OBO. Call 536-3889.
[100.0 %] | [...] - 13.2 kb


Afterwards I searched for "Madras"
The result listing showed to me:
1 Wise Buys Ads & More ~ Home
Sisters Chamber of Commerce Prineville Chamber of Commerce <bold>Madras</bold> Chamber of Commerce Page: 1 2 3 4 5 6
[100.0 %] | [...] - 14.9 kb


Then I searched for "Oregon"
The result listing showed me:
1 Wise Buys Ads & More ~ Home
of a pet suffering from heat exhaustion or stroke. The Humane Society of Central <bold>Oregon</bold> is located one mile south of Reed Market Road on 27th Street in Bend. The shelter is open Monday through Friday from 10:00am to 5:30pm and Saturday from
[100.0 %] | [...] - 16.7 kb
2 Wise Buys Ads & More ~ Home
Serving All of Central <bold>Oregon</bold> Published Every Tuesday by Wingfoot Design Brad & Cheryl Samuelson PO Box 739, La Pine, Oregon Fax:
[28.6 %] | [...] - 24.5 kb
3 Wise Buys Ads & More ~ Home
Serving All of Central <bold>Oregon</bold> Published Every Tuesday by Wingfoot Design Brad & Cheryl Samuelson PO Box 739, La Pine, Oregon Fax:
[28.6 %] | [...] - 13.2 kb
4 Wise Buys Ads & More ~ About Us
Serving All of Central <bold>Oregon</bold> Published Every Tuesday by Wingfoot Design Brad & Cheryl Samuelson PO Box 739, La Pine, Oregon Fax:
[28.6 %] | [...] - 4.6 kb
5 Wise Buys Ads & More ~ Home
Serving All of Central <bold>Oregon</bold> Published Every Tuesday by Wingfoot Design Brad & Cheryl Samuelson PO Box 739, La Pine, Oregon Fax:
[28.6 %] | [...] - 24.5 kb

etc.


Well, that is what I could hope for. The result listing showed me surrounding text (250 character) around the query word. Of course, if on all pages the surrounding text is the same, also the result listing will show this (as to be seen at the last example)

If you are expecting something different in your result listing, please explain.

Tec

p.s. I never understood how <bold>Oregon</bold> will work in this stupid forum software. I don't have enough time to read the whole PHORUM documentation



Edited 2 time(s). Last edit at 07/25/2008 07:15PM by Tec.

If that's what your Sphider does then i must still be unclear on a concept or two because the onsite search only gives me the page headings in returns for "Original", or anything else for that matter.

I redirected that for a reason, reason being don't want client's young site in a less-than-flattering context like e.g. his Web guy doesnt know what the hell he's doing. Now it's probably more relevance on this page than anywhere else on the web. #-o

No thing, i probably should have made that clear but couldya pls delete that or at least the linkage?

And thanks for your help. I'll just keep plugging at it because like i said obviously if your Sphider works as expected and mine doesn't then it must be my fault.

-jb

I believe this has to do with the settings in your control panel.

Look here, what number do you have in the input box in front of Settings => Search Settings => "The length of the description string queried from the database when displaying search results. Can significantly speed up searching on very slow machines, if set to a lower value (eg 250 or 1000; 0 is unlimited), otherwise doesn't have an effect".

If set to zero no description will be shown, if set to for instance 200 the first 200 characters will be shown.

Willy:
<<< If set to zero no description will be shown, if set to for instance 200 the first 200 characters will be shown. >>>
Sorry, but I'm afraid that your answer is not quite correct. We should not confuse Bluesdog more than necessary.

If the value you are talking about is set to 200, only the first 200 characters of the HTML text will be shown in the result listing. Independend if the query word/phrase is part of these first 200 characters. This is why it speeds up the search procedure.


If this value is set to 0, the text arround the query will be shown in the result listing. The length of the surround text is defined in "Maximum length of page summary displayed in search results"
This version is slower, as the surrounding text has to be found in the whole HTML part and has to be separated around to the query word/phrase.

Tec



Edited 1 time(s). Last edit at 07/25/2008 09:15AM by Tec.

"We should not confuse Bluesdog more than necessary."

Only as sufficient, then?

Tec you came back but didn't honor my request nor advise why not. Is that necessary confusion?

Anyway my previous confusion is at minimum *related* to the setting Willy mentioned. Because if i set it to '0' i get precisely the result i was after. If i set it to '1' returns the very top of the page, again, and no query text.

I'd leave it at that except it doesn't sound like a good thing to do.

Tec, i appreciate your help and don't want to sound ungrateful at all but would you /please/ at least remove the linkage from your post?

Thanks,
-jb

Hi Tec. I have been stack for a while;
I am new to search engine development but I find sphider relevant to my problem. I have followed this thread and realized the topic of discussion is similar to mine Kindly address my direction to this::

Am working on a site in which one of the requirements is to implement a vertical search engine specifically for adverts. I have found Sphider light weight custom API to be useful in this case, however am unable to determine which file should I customise to add a regex so that the new search page only displays results of adverts. E.g from an indexed site, it should only display adverts that contain the posted search term like Pop ups, banners etc.

For instance if I search By Cars; It should return adverts found on the index site like pop up for cars or banners or fash.

Kind regards

No body has an idea??

if i understand you right, look in the /templates/dark/search_results.html

Perhaps you can enter your code there.