The function _autotag_fields_to_search is used to determine which $node fields should be searched for terms to be autotagged. There is a variable $ignore_fields, which supposedly contains all the non-relevant fields to be ignored. The problem is that there are a large number of non-relevant fields getting searched because they are not contained in the $ignore_fields variable. For example, I added a line to print out the $fields variable being returned on creating a node, and found that the following 29(!) fields are being searched:

language
title
teaser_include
body
revision
date | (Callback) date();
submit
preview
form_build_id
form_token
form_id
book
menu
pathauto_perform_alias
upload
attach
print_display
print_display_comment
print_display_urllist
nodewords
teaser
validated
icon
tid
is_new
timestamp
moderate
tnid
translate

Most of these are irrelevant, resulting in wasted resources, and the potential for matches to be found where they shouldn't be.

Rather than try to maintain a list of irrelevant term in the $ignore_fields variable, it might be better to to use an algorithm that builds a list of relevant fields to be searched. For example:
1) Add the Title and Body fields to the list of fields to search
2) Add any CCK text fields to the list of fields to search
Parts of this could be configurable (e.g. checkboxes to include cck text fields, or to include comments in the text to be searched)

Comments

Marco Palmero’s picture

subscribe

iantresman’s picture

It could be useful to optional search the title field (a) as well as the body text (b) only the title text.

sdrycroft’s picture

Version: 6.x-1.27 » 6.x-2.0
Status: Active » Fixed

This should be fixed in version two of the module. Please feel free to delve into the code to check though.

Status: Fixed » Closed (fixed)

Automatically closed -- issue fixed for 2 weeks with no activity.

TimelessDomain’s picture

Title: Irrelevant fields are searched for terms » Limit Autotagging to Title, Body, or Specific Fields
Category: bug » feature
Status: Closed (fixed) » Needs work

This is not working for version two yet. At least there is no UI to input these settings. This would make this module much more valuable.

sachbearbeiter’s picture

subscribe

iantresman’s picture

It looks like v2 automatically searches at least the node title and body, but I guess it could be useful to search only specified fields.

kyen99’s picture

Subscribe!

sdrycroft’s picture

Status: Needs work » Closed (won't fix)

This will be a feature in the 7.x version of the module, but definitely won't make it into 6.x.

tranquille’s picture

subscribing