Option not to concatenate metadata

Developers Gathering & indexing
1

There are multiple ways listed in our metamap.cfg to get the meta value c:

c,1,dc.description
c,1,DCTERMS.description
c,1,og:description
c,1,description
c,1,twitter:description

The comment at the top of the file says that:

If more than one entry in the table matches a tag, the most specific match applies

However, in our search results, we’re getting several results joined together with a |, e.g.

From the medieval period to the 21st century: cover the full breadth of English literature|Discover a world of literature at York|Discover a world of literature at York

This is picking up the meta description, the twitter:description and the og:description.

In the list of options available to pass to query_processor_options, there’s -nomdsfconcat listed, but when I add it to my collection.cfg it breaks the search completely. I’ve tried

-nomdsfconcat
-nomdsfconcat=on
-nomdsfconcat=true
-nomdsfconcat=1

All of them make the search break. Can anyone help to let me know why it’s not working right?

We’re on FunnelBack Version 15.10.0.0.

Thanks!

Chris

2

Solved it!

I was adding the option to query_processor_options instead of indexer_options.

I’ve added it and it hasn’t made the search break, so we’ll see if it’s had any affect once it’s re-indexed the content tomorrow…

3

Hi Chris,

Another way we handle scenarios like this is to map only the metadata you need.

i.e. If you only want dc.description in the summary of your search results, you would unmap all the other fields. Or alternatively, map dc.description to some other metadata class and use that in your search summary instead.

In regards to the following:

If more than one entry in the table matches a tag, the most specific match applies

This refers to how Funnelback only matches a metadata field found in the document to one metadata class.

e.g. given the following:

<meta name="one.two" content="test">

and the config

a,1,one
b,1,one.two
c,1,one.two.three

Funnelback will use b,1,one.two and not match the other two configs.

Hope this helps.

4

Thanks, that’s really useful.

Some of our pages don’t have a meta description (I know, I know…) but do have Twitter/OpenGraph descriptions, so we’d want to be able to get at least one of them. Hopefully the indexer_options answer will work!

Thanks for the reply,

Chris

5

Just to follow up on this - it’s worked an now only indexes one of meta description, OG description or Twitter description.

:+1:

1 Like