[RC] Precise Similar Topics II

[MattF]

Title: Precise Similar Topics II

Description: This MOD will display a list of similar/related topics at the bottom of your threads, based on finding the most relevant Topic Titles.

This is based on the Precise Similar Topics MOD by Xabi, which uses a precise and light SQL query to find similar topics. I have taken Xabi's MOD concept, cleaned it up a bit more, gave it a nice looking results layout and packaged it in a proper phpBB MOD format. I have also created an ACP Admin settings page where you can adjust some of its settings to suit your forum's needs.

MOD Download: Precise Similar Topics 1.1.2
MOD Repository: http://github.com/VSEphpbb/precise-similar-topics-II

Credits: To the best of my knowledge, this MOD was originally written by tas2580 (aka Tobi at phpbb-seo.de) as 'Similar Topics'. It was then modified by Xabi into 'Precise Similar Topics' offering a lighter and more precise query. There are also some tweaks inspired by stokerpiller's changes to Xabi's code as well.

Features:

• Uses a precise and light query to search Topic Titles in your forum and return a list of the most similar/relevant matches at the bottom of your threads
• Similar topics will be sorted in order of decreasing relevance.
• You can adjust how many Similar Topics to display
• You can adjust the age of topics to search (ie: search for topics from the past year only)
• You can enable or disable the display of similar topics for each forum
• You can block individual forums from being searched for similar topics
• You can define which forums can share similar topics with other forums
• You can allow similar topic results to use the cache
• Permissions settings for users & groups allow you to set who can view similar topics
• Multi-lingual support for ignore-words
• Forum read-access permissions are respected
• Attractive Prosilver and Subsilver2 layouts
• Very easy to install (only 3 tiny file edits)!
• AutoMOD and UMIL compatible

Notice:

• Requires MySQL 4.0.1 or later.
• Similar topic searches are not case sensitive
• Short words (less than 4 letters) are ignored
• Stopwords are ignored (the, and, some, etc.) full list of stopwords
• If a word is present in more than 50% of all topic titles, it is too common and is treated as a stopword and is ignored.

Demo Board: Test Board

Screenshots:
Prosilver:


Subsilver2:


ACP Admin Page:


ACP Permissions:

[Jorup16]

What this mod looks good, I'll prove I can, thanks for developing it.

[Neuropass]

Great job here... Thanks to the original authors too..

[RMcGirr83]

"ignore forums" could get rather long, no? And does it ignore those that are not true forums?

[MattF]

"ignore forums" could get rather long, no? And does it ignore those that are not true forums?

Yes it only shows forums, not categories or links.

[MattF]

"ignore forums" could get rather long, no?

Well I suppose I could set a max-height for that table and have it scroll any overflow if somebody does have a million forums to prevent the submit/reset buttons from disappearing too far down the page

Download package updated

[Ashley.S.]

"ignore forums" could get rather long, no?

Well I suppose I could set a max-height for that table and have it scroll any overflow if somebody does have a million forums to prevent the submit/reset buttons from disappearing too far down the page

Download package updated

Just a suggestion: But would'nt it be better to have the ignore/include forum in similar topics option listed when you create/edit your forums?

That could save a long list of forums showing on the module as I know nickvergessen did something similar for his NV Recent Topics mod for if you want forums showing in the recent topics list or not.

Otherwise, you could always make it a text box and have users specify what forums they don't want included by adding the forum id to the list or something.

I however am not a coder, so don't even know if its doable, but I thought I might throw in the suggestions to see what you think as it could be another way to solve a long list of ignore forums or scroll box.

[DoYouSpeakWak]

I like this mod. Works fine as i can test.

But with many diffrent types of forums, lets say cars and cooking, in both forums there is a topic named "This is how you do it" (or similar), When the users enters the cooking forum there is a link to a topic in the cars forum names something similar. This may direct a few into the wrong forum.

IMO there should switch for it somewere, Maybe in acp. "Show only similar topcs from this forum" could be another yes or no question in the acp module.

[ac_roma]

nice mod

i will test

[MattF]

Just a suggestion: But would'nt it be better to have the ignore/include forum in similar topics option listed when you create/edit your forums?

That could save a long list of forums showing on the module as I know nickvergessen did something similar for his NV Recent Topics mod for if you want forums showing in the recent topics list or not.

Not really, because if you have a lot of forums, could you imagine the pain it would be to go through each of them, one at a time to make this setting. Not to mention, it would also spread out this MODs settings between various ACP pages instead of centralizing them all in one tidy location

Honestly, this is not an issue. If you have a lot of forums, it is not going to be any more of an issue than in any other ACP page that lists all your forums, like when setting permissions for forums, where they all appear in a scrolling text box.. Now I have made it so if you do have lots of forums, it scrolls here too...

I mocked up an example here for demo:

Otherwise, you could always make it a text box and have users specify what forums they don't want included by adding the forum id to the list or something.

I don't like that option, because that leaves the door open to where people could enter info wrong, which could cause problems... It also forces them to go and figure out the Forum IDs for each forum. With checkboxes, its more fool proof and user-friendly

[MattF]

I like this mod. Works fine as i can test.

But with many diffrent types of forums, lets say cars and cooking, in both forums there is a topic named "This is how you do it" (or similar), When the users enters the cooking forum there is a link to a topic in the cars forum names something similar. This may direct a few into the wrong forum.

IMO there should switch for it somewere, Maybe in acp. "Show only similar topcs from this forum" could be another yes or no question in the acp module.

Hmm, you mean an option like, "SEARCH SELF ONLY"? So that a Forum will only show similar topics from itself?

[DoYouSpeakWak]

I like this mod. Works fine as i can test.

But with many diffrent types of forums, lets say cars and cooking, in both forums there is a topic named "This is how you do it" (or similar), When the users enters the cooking forum there is a link to a topic in the cars forum names something similar. This may direct a few into the wrong forum.

IMO there should switch for it somewere, Maybe in acp. "Show only similar topcs from this forum" could be another yes or no question in the acp module.

Hmm, you mean an option like, "SEARCH SELF ONLY"? So that a Forum will only show similar topics from itself?

Yes sir

[Ashley.S.]

Just a suggestion: But would'nt it be better to have the ignore/include forum in similar topics option listed when you create/edit your forums?

That could save a long list of forums showing on the module as I know nickvergessen did something similar for his NV Recent Topics mod for if you want forums showing in the recent topics list or not.

Not really, because if you have a lot of forums, could you imagine the pain it would be to go through each of them, one at a time to make this setting. Not to mention, it would also spread out this MODs settings between various ACP pages instead of centralizing them all in one tidy location

Honestly, this is not an issue. If you have a lot of forums, it is not going to be any more of an issue than in any other ACP page that lists all your forums, like when setting permissions for forums, where they all appear in a scrolling text box.. Now I have made it so if you do have lots of forums, it scrolls here too...

I mocked up an example here for demo:

Otherwise, you could always make it a text box and have users specify what forums they don't want included by adding the forum id to the list or something.

I don't like that option, because that leaves the door open to where people could enter info wrong, which could cause problems... It also forces them to go and figure out the Forum IDs for each forum. With checkboxes, its more fool proof and user-friendly

ok then, fair enough VSE...I will try this out sometime soon because I do have a few forums that I don't want similar topics to be searched in.

[MattF]

But with many diffrent types of forums, lets say cars and cooking, in both forums there is a topic named "This is how you do it" (or similar), When the users enters the cooking forum there is a link to a topic in the cars forum names something similar. This may direct a few into the wrong forum.

IMO there should switch for it somewere, Maybe in acp. "Show only similar topcs from this forum" could be another yes or no question in the acp module.

It turns out this is not an issue

I added some more info to the first post about how Similar Topic searches are made.
Key points:

• Similar topic searches are not case sensitive
• Short words (less than 4 letters) are ignored
• Stopwords are ignored (the, and, some, etc.) full list of stopwords
• If a word is present in more than 50% of all topic titles, it is too common and is treated as a stopword and is ignored.

As you can see, threads like "This is how you do it" would not factor in to Similar Topics as these words are all stopwords. So "This is how you fix a truck" and "This is how you cook fish" would not show up as similar topics to each other... The keywords in those topics are truck, cook, fish

EDIT: Download package updated in 1st post. No new edits or installer file to run, just copy the included files over the old ones. Just contains some minor language, code and ACP style tweaks

[DoYouSpeakWak]

But with many diffrent types of forums, lets say cars and cooking, in both forums there is a topic named "This is how you do it" (or similar), When the users enters the cooking forum there is a link to a topic in the cars forum names something similar. This may direct a few into the wrong forum.

IMO there should switch for it somewere, Maybe in acp. "Show only similar topcs from this forum" could be another yes or no question in the acp module.

It turns out this is not an issue

I added some more info to the first post about how Similar Topic searches are made.
Key points:

• Similar topic searches are not case sensitive
• Short words (less than 4 letters) are ignored
• Stopwords are ignored (the, and, some, etc.) full list of stopwords
• If a word is present in more than 50% of all topic titles, it is too common and is treated as a stopword and is ignored.

As you can see, threads like "This is how you do it" would not factor in to Similar Topics as these words are all stopwords. So "This is how you fix a truck" and "This is how you cook fish" would not show up as similar topics to each other... The keywords in those topics are truck, cook, fish

EDIT: Download package updated in 1st post. No new edits or installer file to run, just copy the included files over the old ones. Just contains some minor language, code and ACP style tweaks

That does not solve the problem. Topics with non stopwords in them listed in both forums will still be listed together. Like "I need help" can be in both forums, but the topics content can be totally diffrent.

We must have 4-5000 topics with the words "i" "need" "help" in the topic tittle, additional to that there is many many other topics with the same problem. Only way to solve it would be a switch not to seach other forums than the one your in.

I hope you can see the problems this missing feature can give in 8-10 yrs old forums with millions of topics. Like here.