MetaTagger is forcing suggested metadata to lower case

MetaTagger 4.1.2
TeamSite 6.7.1
Solaris 9

Hi there,

MetaTagger is forcing suggested metadata to lower case (i.e., it’s not preserving the original case). We have a relatively simple summarization-based field called "Description" with the following rack script:

parse -ignorenumbers -mode sentence; summarize -keysentences 1

For example, if a PDF has a description of...

The Quick Brown Fox

...it gets suggested as...

the quick brown fox

Note: I’ve looked into the "-case" flag for the "parse" command. Unfortunately this forces MetaTagger to treat the same sentence with different cases as unique, which is NOT desirable in our case. We simply want to PRESERVE case.

MetaTagger 3.x didn’t have this problem.

Has anyone else had to work around this issue?

Thanks,

Find more posts tagged with

Comments

Migrateduser

Dan,

Could you elaborate on the concern with using -case in parse? It will preserve case on the input as well as doing a case-sensitive weighting. I think there's an existing FR for "preserve case but weight by nocase/stem".

Thanks
Clark

Migrateduser

Just wanted to follow up to close the loop on this.

Thanks for the reply, Clark...and sorry for such a late reply. After further reading and tinkering, I ended up using the -case flag afterall. I think I just misunderstood how it works. For what it's worth, here's the new rack script for the "Description" field. It seems to be producing the results we want:

parse -case -ignorenumbers -mode sentence; summarize -generic -numsentences 1