Data From Field to parse dates

josh104
Posts: 24
Joined: Mon Oct 09, 2006 5:39 pm

Data From Field to parse dates

Post by josh104 »

I was hoping to use a regex for the Data From Field setting to parse news articles searched from our profiles.

The ideal solution would seem to be translating the parsed information to the Modified Date field, for incorporation into the Search results ordering by date, but there doesn't seem to be any clear way of translating a parsed date such as "August 20th, 2006" to the Midified Date field.

Any suggestions/thoughts?
User avatar
mark
Site Admin
Posts: 5519
Joined: Tue Apr 25, 2000 6:56 pm

Data From Field to parse dates

Post by mark »

search: >>\alpha+ =\digit{1,2}[^,]{0,2}, =\digit{4}
replace: \1\2\3\5\6

(untested)
josh104
Posts: 24
Joined: Mon Oct 09, 2006 5:39 pm

Data From Field to parse dates

Post by josh104 »

Thank you for the suggestion, but the thing I'm having trouble with specifically is the following:

From all appearances, the date modified seems to be stored in the database as "yyyy-mm-dd hh:mm:ss". While the day and year can be pulled from the document and put into the date modified, there doesn't seem to be a way to translate the month field from a semantic representation to a numerical one.

One thought I'm having is possibly trying to implement the "Additional Fields" feature that was added with the last patch, but there isn't much documentation explaining it's implementation.
User avatar
mark
Site Admin
Posts: 5519
Joined: Tue Apr 25, 2000 6:56 pm

Data From Field to parse dates

Post by mark »

The date parser can handle the format you mentioned with the exception of the "th" after the day number. The search and replace I suggested removes that so the date should parse correctly into the database.
josh104
Posts: 24
Joined: Mon Oct 09, 2006 5:39 pm

Data From Field to parse dates

Post by josh104 »

Mark,

The regex you suggested worked great, and it did indeed translate over to date modified perfectly.

While this is a huge step forward in things for indexing, sorting the results by date seems to throw out a lot of the search logic. The additional fields option added with the last patch has a cryptic "Sortable" option next to them.

Is it possible to pull the dates and instead of putting them into date modified, put them in a new field and weight it's relevancy into the existing search metrics?

I tried to play around with this, but when I directed the regex to the extra field I created, it didn't show up anywhere when I checked what was indexed from the site, and there was no weighting option in the search settings for the extra field.

Any info on this? I appreciate the help.
User avatar
John
Site Admin
Posts: 2623
Joined: Mon Apr 24, 2000 3:18 pm
Location: Cleveland, OH

Data From Field to parse dates

Post by John »

Currently the sort by date or additional fields takes priority over the relevancy rank, it isn't blended or weighted in the way you can with Depth.

We anticipate have more fields that can be blended into the relevancy rank in the release at the end of this month.
John Turnbull
Thunderstone Software
josh104
Posts: 24
Joined: Mon Oct 09, 2006 5:39 pm

Data From Field to parse dates

Post by josh104 »

great, thanks for your help!

I look forward to the next patch.
josh104
Posts: 24
Joined: Mon Oct 09, 2006 5:39 pm

Data From Field to parse dates

Post by josh104 »

I'm just wondering about the status of the feature discussed in response 6 in this thread. Was this released? Is it still in development? As far as I'm aware I have the most updated scripts and I don't believe this functionality has been added.
User avatar
John
Site Admin
Posts: 2623
Joined: Mon Apr 24, 2000 3:18 pm
Location: Cleveland, OH

Data From Field to parse dates

Post by John »

That is scheduled for the Search Appliance 7 release. We do not have an exact date yet.
John Turnbull
Thunderstone Software