EPO - Discussion forums

Hello again,
I'm sorry to bombard you with a lot of questions but this is the first time I use PATSTAT for research purposes and I'm not quite sure how to go along.
I've been trying to figure out how to compute backwards and forward citations. I understand that on a family level there is NB_CITING_DOCDB_FAM for forward citations. Is there a key for backwards too or those should be computed on an application level?
I don't understand fully the subject discussed in relation to this in PATSTAT data catalog v.5 06 on page in the following paragraph I do not understand the last sentence re the difference between appl. and pub.:
" PAT_PUBLN_ID refers to the citing publication.
 CITED_PAT_PUBLN_ID refers to the publication being cited.
 CITED_APPLN_ID refers to an application being cited. This cited application is not related to a cited publication, but is a valid citation on its own."
Basically, what I want to know for each application is the no. of forward citations, backward citations, and type of forward citations (i.e. who added the citations). Can you help me with the search query, I'm a bit lost.
Thank you,
Talya

Based on what I could find in the data catalogue at http://documents.epo.org/projects/babyl ... v_5_06.pdf I think there is no variable for backward citations similar to NB_CITING_DOCDB_FAM. However, you can calculate this yourself by using the TLS228_DOCDB_FAM_CITN.

I have no clue as to what "This cited application is not related to a cited publication, but is a valid citation on its own." means. Probably they are pointing out something obvious in a vague way.

You can look in TLS215 for the type of citation.

ok, I agree the formulation is a bit vague and we will adapt it.

The majority and most obvious of citations refer to publications and non-patent literature documents. This is ALWAYS the case when the citations originate from the examiner.
In a minority of cases, APPLICANTS have made references to earlier APPLICATIONS instead of publications. A reason could be that the application was not published yet, maybe even withdrawn before publication, or whatever other reason.

Many PATSTAT users thought (think) that in tls212_citation table, the attribute "CITED_APPLN_ID" is the equivalent to the "CITED_PAT_PUBNL_ID", and therefore assumed that using the CITED_APPLN_ID to join with tls201_appln would give them directly the cited applications from the publications referred to in the "CITED_PAT_PUBNL_ID" field. This is NOT the case. The "CITED_PAT_PUBNL_ID" is not the equivalent of the "CITED_APPLN_ID".

Learn something new every day. What Geert means is that there are two kinds of citations. Either a publication is cited and this is captured by cited_pat_publn_id, OR an application is cited -didn't know this was possible, hence the confusion- which is captured by cited_appln_id. In the latter case, cited_appln_id has a value, but cited_pat_publn_id is 0.
So to count forward citations for let's say appln_id=1 you'd have to run two queries:
1.) SELECT * FROM tls212_citation WHERE cited_appln_id=1;
2.) SELECT * FROM tls212_citation WHERE cited_pat_publn_id IN (SELECT pat_publn_id FROM tls211_pat_publn WHERE appln_id=1);

Well, this is my latest assumption anyway.

I also just noticed that this description wasn't part of data catalogue v5.02 (PATSTAT October 2014), so either this wasn't in PATSTAT before, or it wasn't explained before.