Invision Community 4.5: Search Insights!

tony45

Enthusiast
Joined
Jul 1, 2014
Messages
190
4.5: Search Insights

Leverage information about searches on your community to focus on areas of importance to your end users.

Every single day, your members are searching your community for answers or interesting conversations to join.
Wouldn't it be great if you could learn what is being searched for to identify hot issues, commonly asked questions and discover trends?
We thought so too, which is why Invision Community 4.5 comes with search statistics.
For the first time, Invision Community gathers anonymized information on what your members are searching for so you can use this to highlight more relevant content and shape strategic decisions with your community's structure.


Searching.jpg


Search statistics help you track searches performed on your community

When a member searches, their identity is converted into a unique key that cannot be reversed to identify the member. This allows us to track a single member's search usage over many search sessions without being able to link it to a specific member account.

The AdminCP now features a dashboard to review the most popular search terms as well as a raw log of recent searches along with the results they returned.
We have a lot of ideas in mind for additional changes down the road with the tracking of popular search terms, but for now, we hope you like the new statistics page and find the information presented useful for your future site plans.

 

Joel R

Habitué
Joined
Nov 24, 2013
Messages
1,035
This is one of the more interesting features of the potentiality of what can be done.

There's been various "search tracker" apps in the Marketplace for many years now, but this is the first time it'll be in the core. This is the newest feature that excites me the most.
 

haqzore

Devotee
Joined
Dec 6, 2012
Messages
2,654
I'm all for privacy, but what's the point in anonymizing the search history?

Wouldn't care either way... Just curious what the point is.
 

Kevin

Oooh, something shiny!
Joined
Jul 13, 2004
Messages
3,451
I'm all for privacy, but what's the point in anonymizing the search history?

Wouldn't care either way... Just curious what the point is.
Some communities may be of a sensitive nature so members likely wouldn't appreciate having it tracked precisely what they're searching for.
 

haqzore

Devotee
Joined
Dec 6, 2012
Messages
2,654
Some communities may be of a sensitive nature so members likely wouldn't appreciate having it tracked precisely what they're searching for.
That was my first assumption, but..

If you're already a member of a "sensitive content" community...

Aren't you already disclosing your participation by registering? Isn't it already assumed you're "participating" in said content just by being a member? Even if you don't reply/etc?

I don't know... Doesn't matter to me/my IPS site anyways... Just curious. Appreciate the reply.

If nothing else, it's better to err on the side of too much privacy.
 
D

Deleted member 3691

Guest
It's to ensure the privacy of your members.

We don't really need to know the exact user who searched for what. We do want to know what a single person searches for over the weeks and months, which is why we assign a one way hash to represent them.
 

zappaDPJ

Moderator
Joined
Aug 26, 2010
Messages
8,450
We do want to know what a single person searches for over the weeks and months, which is why we assign a one way hash to represent them.

So can you foresee at some point in the future this data being used to return results tailored to an individual user's search terms?
 
D

Deleted member 3691

Guest
So can you foresee at some point in the future this data being used to return results tailored to an individual user's search terms?

Not really. We currently use an anonymised hash to represent a person, so we can't then apply that data to a user when they search as we have no idea who it is.

We could do an opt-in but still, gathering data is one thing, machine learning is quite another and probably above the processing power of the average hosted forum. Facebook and Google have massive processing power to get insights from the data collected, and we don't have that.

That said, there are some exciting services being launched from AWS that can process the data for you, but everything comes at a cost.

Finally, I'm not sure how useful it would be anyway. If I knew you searched for "dogs" a lot, then we would need to analyse all the data in the post tables to add a relevance score along with other ranking factors (time since posted, users you follow, topics you have replied in, etc) just for you to make it useful beyond something superficial.

People tend to search for things they can't normally find and aren't generally interested in. So for example, you may have recalled seeing a CSS tweak for the nav bar and think "I need that, where is it.." so you go to search for 'css' and hopefully find it.

Then to be told "HEY! You searched for CSS! Here's all the latest topics with CSS in the title" wouldn't be overly useful when you come back a week later.
 

zappaDPJ

Moderator
Joined
Aug 26, 2010
Messages
8,450
Not really. We currently use an anonymised hash to represent a person, so we can't then apply that data to a user when they search as we have no idea who it is.

We could do an opt-in but still, gathering data is one thing, machine learning is quite another and probably above the processing power of the average hosted forum. Facebook and Google have massive processing power to get insights from the data collected, and we don't have that.

That said, there are some exciting services being launched from AWS that can process the data for you, but everything comes at a cost.

Finally, I'm not sure how useful it would be anyway. If I knew you searched for "dogs" a lot, then we would need to analyse all the data in the post tables to add a relevance score along with other ranking factors (time since posted, users you follow, topics you have replied in, etc) just for you to make it useful beyond something superficial.

People tend to search for things they can't normally find and aren't generally interested in. So for example, you may have recalled seeing a CSS tweak for the nav bar and think "I need that, where is it.." so you go to search for 'css' and hopefully find it.

Then to be told "HEY! You searched for CSS! Here's all the latest topics with CSS in the title" wouldn't be overly useful when you come back a week later.

Thanks for your detailed reply, that's given me something to think about :)
 
Top