https://github.com/GlobalNamesArchitecture/gnrd/blob/master/models/name_finder.rb#L309 indicates that all PDF files get OCR'd. While this is good for image-based PDFs, this is not good for text-based PDFs; the result is poorer name-finding than otherwise could be. We need a mechanism to determine if a PDF is text-based vs image-based and set the ocr: true/false flag appropriately.

To some degree, I'm assuming these problems are caused by the name being in a table. I can't be sure.

Hi again,
Thanks for looking into my issues, even though they seem to have disappeared as quickly as they reappeared! This isn't really an issue, but I figured it's the best way to get in contact: I am wondering how you would like GNRD to be cited if I use it in a procedure for a scientific paper.
Thanks!

I am adding gRPC server to gnfinder that we can access via ruby interface to GNRD

Looks like a problem with nginx setting

Example:
There is a spider called Pardosa moesta
vs:
There is a spider called Pardosa moesta.

Note the period in the second & not in the first. Only the second results in name "Pardosa moesta" being recognized.

GNRD is finding a correct taxon name, but grabbing a following word that it shouldn't. Words like "clade", "complex", and "genome" are very common

American Seashell book is an example of that

I would expect GNRD to find "Pycnoclavella aff. detorta" and reconcile it to Pycnoclavella detorta. Instead it finds Pycnoclavella only.

Names that are part of a string such as Rhypholophus dufouri sp. n. get interpreted as "Rhypholophus dufouri sp". The "sp" is clearly extraneous, can this be pruned off? At the moment I have to do this internally so that I can match names to pages, e.g. http://bionames.org/names/cluster/1049906

To see a live example of this bug, see http://gnrd.globalnames.org/name_finder.html?token=3rkep8uekg where I've looked for names in the PDF http://retro.seals.ch/cntmng?pid=seg-001:1993:66::693

I'm not sure this is in scope for GNRD

Not sure if this is in scope

Technically, Bacteria is the name of a Domain of life, like Eukaryota and Archaea, but it is often used as a common name, bacteria. I'm not sure how you want to handle this. When calculating performance, I take all lower case bacteria as a common name and all uppercase Bacteria as a taxon name. As a result "Bacteria" is labeled as a false negative. Would you like to say that all instances of Bacteria are also common names?

I am trying to install the docker container to run GNRD locally. I ran docker pull gnames/gnrd, then started the container with this command to forward port 80 in the container to 5678 on my machine:

docker run -p 5678:80 gnames/gnrd

It seems to load up no problems, and I can load static pages like 0.0.0.0:5678/api, but when I try to connect to the API using a command like the following, then I just get

Internal Server Error

curl -L http://0.0.0.0:5678/name_finder.json?url=http://en.wikipedia.org/wiki/Araneae

Any suggestions on what I might be doing wrong?

From a comment by @stralion in Disqus comment on website:
Look at this paper: http://rsos.royalsocietypublishing.org/content/2/9/150156
The name "Hippopotamus amphibius" occurs twice, and "H. amphibius" occurs twice.
When given the URL, GNRD finds "H. amphibius" but not "Hippopotamus amphibius". When given the PDF, GNRD does not find either one.

Not sure if this is in scope

Refer to GNRD Dryad project:

More names (I can provide the data package number if needed)

• Talavera
• Crosa
• Pinon
• Graus
• Caterino
• Casas
• Ades
• Narum
• Ikeda
• Camara
• Abila
• Simo
• Fraga
• Verma
• Moreno
• Vitalis
• Imber
• Luca
• Ziminia
• Lara
• Darda canica
• Luca
• Kukalova
• Leis
• Mossman
• Vitalis
• Fontanella
• Nason
• Ferna
• Beleza
• Mona
• Pereira
• Talavera
• Luzuriaga
• Vila
• Mona
• Tella
• Ferna
• Gaeta
• Civetta
• Ojeda
• Piras
• Oyama
• Narum
• Vitalis
• Colla
• Moreno
• Narita
• Roche
• Gaeta
• Alfaro
• Barbosa
• Tanada
• Sasa
• Carmona
• Momot
• Quesada
• Moya
• Tarka
• Tobias
• Pereira
• Narum
• Rubini
• Nason
• Tella
• Narum
• Talavera
• Egea
• Vila
• Gregorius
• Moreno
• Vila
• Vitalis
• Tsukada
• Gregorius

GNRD finds the text string, but doesn't know what to do with it. It can't see that it is two names hyphenated together.

When a poll the token URL that I receive after performing a search, I get the same result as the search URL.

For example, using this token URL: http://gnrd.globalnames.org/name_finder.xml?token=6jgv9mqdq8

just returns:

but I need a result that contains a scientific name, like so:

I haven't had this issue in the past. Any idea why its not working now? Thanks!

Below are false positives that are place names. I have included the actual text string, if that helps. We might be able to use that to differentiate places from taxa.

For example, in the JSON for data package 1044, GNRD finds the string verbatim: "N. giraulti-Nasonia longicornis" but lists name: "N. giraulti?nasonia longicornis". How can we get the hyphen in verbatim to show up in name?

Followed the steps mentioned, but eventually getting the following error:
Also, config.example.json is missing from the production branch, hence copied the content form 17-gnfinder-only

Starting gnrd_redis_1 ... done Starting gnrd_gnf_1 ... done Starting gnrd_pg_1 ... done rake aborted! PG::ConnectionBad: could not translate host name "pg" to address: No address associated with hostname /var/lib/gems/2.6.0/gems/pg-1.1.4/lib/pg.rb:56:in initialize'
/var/lib/gems/2.6.0/gems/pg-1.1.4/lib/pg.rb:56:in new' /var/lib/gems/2.6.0/gems/pg-1.1.4/lib/pg.rb:56:in connect'
/var/lib/gems/2.6.0/gems/activerecord-5.2.4/lib/active_record/connection_adapters/postgresql_adapter.rb:692:in connect' /var/lib/gems/2.6.0/gems/activerecord-5.2.4/lib/active_record/connection_adapters/postgresql_adapter.rb:223:in initialize'
/var/lib/gems/2.6.0/gems/activerecord-5.2.4/lib/active_record/connection_adapters/postgresql_adapter.rb:48:in new' /var/lib/gems/2.6.0/gems/activerecord-5.2.4/lib/active_record/connection_adapters/postgresql_adapter.rb:48:in postgresql_connection'
/var/lib/gems/2.6.0/gems/activerecord-5.2.4/lib/active_record/connection_adapters/abstract/connection_pool.rb:830:in new_connection' /var/lib/gems/2.6.0/gems/activerecord-5.2.4/lib/active_record/connection_adapters/abstract/connection_pool.rb:874:in checkout_new_connection'
/var/lib/gems/2.6.0/gems/activerecord-5.2.4/lib/active_record/connection_adapters/abstract/connection_pool.rb:853:in try_to_checkout_new_connection' /var/lib/gems/2.6.0/gems/activerecord-5.2.4/lib/active_record/connection_adapters/abstract/connection_pool.rb:814:in acquire_connection'
/var/lib/gems/2.6.0/gems/activerecord-5.2.4/lib/active_record/connection_adapters/abstract/connection_pool.rb:538:in checkout' /var/lib/gems/2.6.0/gems/activerecord-5.2.4/lib/active_record/connection_adapters/abstract/connection_pool.rb:382:in connection'
/var/lib/gems/2.6.0/gems/activerecord-5.2.4/lib/active_record/connection_adapters/abstract/connection_pool.rb:1033:in retrieve_connection' /var/lib/gems/2.6.0/gems/activerecord-5.2.4/lib/active_record/connection_handling.rb:118:in retrieve_connection'
/var/lib/gems/2.6.0/gems/activerecord-5.2.4/lib/active_record/connection_handling.rb:90:in connection' /var/lib/gems/2.6.0/gems/activerecord-5.2.4/lib/active_record/tasks/database_tasks.rb:57:in check_protected_environments!'
/var/lib/gems/2.6.0/gems/activerecord-5.2.4/lib/active_record/railties/databases.rake:13:in block (2 levels) in <top (required)>' /var/lib/gems/2.6.0/gems/rake-12.3.3/exe/rake:27:in <top (required)>'
Tasks: TOP => db:reset => db:drop => db:check_protected_environments
`

I have not done an exhaustive search, but via spot checking, I'm having a hard time finding an example of an abbreviated name being matched. Examples are below. All have "noMatch" matchType - even if the full name is also given in the text.