Advantages of using Matrix + Funnelback vs another search tool


(Birdg) #1

Hi,

 

I'm researching enterprise search engines for our website and intranet. It seems that Funnelback has a roughly comparable feature set to other tools such as Google Search Appliance and Elasticsearch.

 

However, features aside, I'd like to understand whether the integration between Squiz Matrix and Funnelback allows functionality that is not possible with other products...or just not as straightforward?

 

Are there things that can only be done by a combined Matrix/Funnelback implementation?

 

Many thanks,

 

Graham


(Bart Banda) #2

Hi Graham,

not only is Funnelback a superior search engine appliance, but Squiz Matrix and Funnelback are both designed to make integration between the two a lot easier and smoother. I would say that almost about 80 or even 90% of our Squiz Matrix clients have some sort of Funnelback integration happening. Most sites have backend integration with an external Funnelback server so the user's experience on your site is seamless and they never leaves your website's domain.

 

You can also do very powerful imports and indexes and present that information within Squiz Matrix. So with Funnelback, you can index several external repositories or databases, bring it all together in 1 index, and present it all together for the user. For example, QAG index their social media profiles and present those on their main website search as facets: http://www.qagoma.qld.gov.au/fb-search?query=art&collection=qag-meta&x=-1351&y=-163

 

Furthermore, you can use it as an import tool and not only as a search. So it can index an external database on a daily basis, analyse the content, and then either create, modify, or delete assets in Matrix based on the records found in the index. 

 

I could go on all day and I'm sure someone from Funnelback will reply with even more information. But pretty much, anything that is not possible with Squiz Matrix, is possible if you bring Funnelback into the mix. 


(Birdg) #3

Hi Bart,

Thanks for your detailed reply. I agree with you but it seems to me that much if not all of these things can be achieved with other search engines. We are very interested in Funnelback but what I want to understand is whether using it in combination with Matrix enables functionality that is impossible or very difficult to achieve with other products - and if so what they are.


Hope that makes sense.


Thanks again,


Graham


(Aleks Bochniak) #4

Graham

 

'Integration' via REST resources can be achieved with any web enabled technology really. Most funnelback integrations I have seen involve funnelback returning JSON arrays or HTML back to a REST JS/REST asset which then can be easily parsed by Matrix. 

 

As long as your 3rd party web technology/system/search is returning something that can be easily parsed with a REST JS asset, or displayed within content/paint layout via a REST asset - then that's all you need to worry about.

 

The power of funnelback comes when you need to do complex queries on your integration data - as Bart has mentioned above.


(Benjamin Pearson) #5

Graham

 

'Integration' via REST resources can be achieved with any web enabled technology really. Most funnelback integrations I have seen involve funnelback returning JSON arrays or HTML back to a REST JS/REST asset which then can be easily parsed by Matrix. 

 

As long as your 3rd party web technology/system/search is returning something that can be easily parsed with a REST JS asset, or displayed within content/paint layout via a REST asset - then that's all you need to worry about.

 

The power of funnelback comes when you need to do complex queries on your integration data - as Bart has mentioned above.

I believe the newer integration model (FB >= 12) is a REST resource inside the Funnelback assets, but the traditional ways were using Funnelback APIs. Also, I think the newer integration model uses extra methods like HTTP headers to help the integration process.


(Birdg) #6

Hi Aleks/Benjamin,

 

Many thanks - that's really helpful.

 

And thanks again Bart for the detail on the advanced use of Funnelback.

 

:)

 

Graham


(Gordon Grace) #7

Hi Graham -

 

Funnelback v12.4+ also includes a native 'Matrix' collection type - perfect if you're applying document-level security to your Matrix-managed content, and also need Funnelback to respect those permissions at query time.

 

Matrix+Funnelback implementations are also starting to standardise on the use of an asset listing for publishing external metadata for ingestion and consumption by Funnelback - any of the binary files you're uploading and managing in Matrix (including Matrix-managed metadata) can also be made visible to Funnelback.

 

In theory, the same approach could be used to allow Matrix to manage the data source behind Funnelback's rich query completion CSV files.

 

A lot of the Matrix content templates I'm seeing lately also seem to have support for commonly-used metadata schemas (indexed by Funnelback by default) and demarcation of indexable regions within a page - useful for keeping search indexes clean and well-structured.

 

If you want to get closer to the bleeding edge, Funnelback's Recommender API can be used to automatically augment existing Matrix-rendered content page with 'related' or 'recommended' content that Funnelback also knows about (scoped by the template author), irrespective of whether or not the recommended items live in Matrix.

 


(Birdg) #8

Hi Gordon,

Thanks very much. That’s really useful.


Graham


(Byrn2lis) #9

Hi Graham -

 

Funnelback v12.4+ also includes a native 'Matrix' collection type - perfect if you're applying document-level security to your Matrix-managed content, and also need Funnelback to respect those permissions at query time.

 

Matrix+Funnelback implementations are also starting to standardise on the use of an asset listing for publishing external metadata for ingestion and consumption by Funnelback - any of the binary files you're uploading and managing in Matrix (including Matrix-managed metadata) can also be made visible to Funnelback.

 

In theory, the same approach could be used to allow Matrix to manage the data source behind Funnelback's rich query completion CSV files.

 

A lot of the Matrix content templates I'm seeing lately also seem to have support for commonly-used metadata schemas (indexed by Funnelback by default) and demarcation of indexable regions within a page - useful for keeping search indexes clean and well-structured.

 

If you want to get closer to the bleeding edge, Funnelback's Recommender API can be used to automatically augment existing Matrix-rendered content page with 'related' or 'recommended' content that Funnelback also knows about (scoped by the template author), irrespective of whether or not the recommended items live in Matrix.

 

 

Hi Gordon -

 

I am looking to have matrix-defined metadata associatied with file type assets within Matrix indexed to funnelback.

 

Is a Matrix collection type required to index matrix-defined metadata?

 

My second question is how to you use/push the asset listing to the external_metadata.cfg file?

 

Thanks in advance


(Gordon Grace) #10

Hi, Beans -

 

An asset listing in Matrix generating external metadata for binary files are generally published at a URL like:

http://MATRIX-SERVER/funnelback/external_metadata.html

A Matrix collection is NOT required for this approach - a standard web collection will suffice.

 

Once you've got Matrix publishing this file as expected (and conforming to the Funnelback External Metadata syntax), you can add it to your Funnelback collection's update workflow:

 

For every collection update:

  1. Pre-gather: Retrieve the external metadata from Matrix, saving it to $SEARCH_HOME/conf/COLLECTION/external_metadata.cfg
  2. Pre-index: Validate the external metadata

If  $SEARCH_HOME/conf/COLLECTION/external_metadata.cfg exists, a collection update will take it into consideration automatically during the index phase.

 

See also: