UBB.Developers Forums UBB.threads 6 V5.php Modifications V5.php Mod Ideas 'Similar Pages' features!

Since there is no respond at infopop site, I am posting it again ere, I hope someone could implement this hack

===
My problem being an Administrator at my forum is that some new members keep posting topics which in fact the matter is already on the database. They are lazy to do the search and they don't use the advance search. If they use the search feature, they only put 1 or 2 maximum words, and they are shown with hundreds even thousands list of topics.

So, to minimize such problem that I believe we all have, we can add a little feature on the "Extra" like "Similar Page", (same as google, yahoo or other search engine)

I tried to implement this hack, but the problem is is not so easy because I think we have to add more database table for the "Ignore words", you know, we don't want to include word such as "to", "and", etc, right? Because it will be useless afterall

I hope Rick catch this post.
===

How is this different from regular searches? Do you have an example?

Ah.. yes

Ok,
If you go to Google and type i.e. "PHP Installation", you will receive list of website about PHP Installation.

So in order to make the search feature easier to visitors, we can put link in every post something like "Similar Pages" which will give you list of threads with similar interest to the one you are reading.

This is exactly the same with Google feature which have "Similar Pages" link in every result of your search.

Take a look at the Google link above and click on "Similar Links"

I think you are right, it is the same as regular searches, but this "Similar Pages" link will do the following:
- Easier for visitors to find the information they are looking for (it eventually prevents double posting/threads with similar interest because they think there is no information that they are looking for)
- Save bandwidth , you dont have to go into the search page
- The only Bulletin Board which has this feature

Ok, I hope you get the point.

Oh I see what you're talking about.

While you are reading a post about "widgets"... at the bottom it says "similar posts" or "similar pages". Clicking that link sort of does a search for "widgets" for you. Saves a step.
That would be unique. Don't know how tough it would be for it to figure out what the content of the page was about and do a search for it.
But it would be kinda cool.

I think I do

Each post posted would also write to a seperate table the first 100 or so words (minus the obvious words such as "to" "a" "and" "the" etc). The link within the post would be a search link that would search based on the keywords stored in the table related to that post.

posted by JoshPet:

---

Oh I see what you're talking about.

While you are reading a post about "widgets"... at the bottom it says "similar posts" or "similar pages". Clicking that link sort of does a search for "widgets" for you. Saves a step.
That would be unique. Don't know how tough it would be for it to figure out what the content of the page was about and do a search for it.
But it would be kinda cool.

---

Yes! Exactly
And AllenAyres, yes, I was thinking the same thing.

[]
(minus the obvious words such as "to" "a" "and" "the" etc).

---

Does anyone know, how many words the "Obvious Words" are?

let's not forget that there's also obvious words in other languages too... lol

Perhaps a field that was available when making a post for the poster to list what they think their post's key words should be. If I was posting about red cars and speeding tickets I would put (red, cars, speeding, tickets) into the key word field. Then that "related posts" search link could look for other posts containing my key words.

just some thoughts...

That would be the best way to do it.

But you know, we use PhotoPost for our photos... and it's got a keyword field...and no matter how much we beg... it's tough to make them fill in the keyword fields. Everyone leaves it blank.
Maybe a config option to require the field?

ubb.classic does something similar when building its search index, it removes the "common words" and I think others have published their translations for the list

The ones ubb removes are:

a an are as be because but can could for how in is it of should that the there this to was will with would

It's a box in the search settings in the control panel you can edit/add to/ etc.

I was reading the MySQL manual yesterday and found an interesting new feature (since 3.23.23) which is perfect for this (and other searches as well).

It's possible to match results with the function MATCH(), which does some magic internally to decide which rows to select. It orders matches by a sort of score, and each word that is in the match is given a weight based on in how many of the found rows it is found in. If the word is found in more than 50% of the rows that word will give a score of zero.

You can find more info on this here. I didn't read everything, but it seemed very promising and I did a mental note to check on it later.

Basically, you would get the ignoring of common words without the need of a table for every possible language. It will probably be quite fast as well, since it is indexed.

Sounds like it might be the ticket