backward citations

[Joana]

I am using the patstat 2023 autumn version. There I have identified a group of patents (my core dataset; the associated appln_ids are stored in table 'cc_core_dataset'). I am now interested in getting the appln_ids of those patents that have been cited by the patents in the core dataset (i.e. my focus is on backward citations). The idea is to get second dataset based on the core dataset.

I am not sure if my query is correct and I would be happy to get advice. Also, any hint on mistakes or ways to improve my query are very much appreciated. I am also not sure if it makes sense to add a time restriction regarding backward citations (in case of forward citations I find that reasonable, though). Thanks a lot.

This is my query (inspired from here://forums.epo.org/patent-citation-backward-citation-7788):

Code: Select all

SELECT DISTINCT citing.appln_id  citingPatentApplnID
, cited.appln_id  citedPatentApplnID
, citing.publn_auth + citing.publn_nr AS citingPublNumber
, cited.publn_auth +cited.publn_nr AS citedPublNumber
, citeda.appln_filing_year
, j.citn_origin
FROM patstat2023.tls211_pat_publn   AS citing 
JOIN patstat2023.tls212_citation        AS j        ON citing.pat_publn_id = j.pat_publn_id
JOIN patstat2023.tls211_pat_publn     AS cited    ON j.cited_pat_publn_id = cited.pat_publn_id
JOIN patstat2023.tls201_appln       AS citinga  ON citing.appln_id=citinga.appln_id
JOIN patstat2023.tls201_appln       AS citeda   ON cited.appln_id=citeda.appln_id
JOIN patstat2023.tls202_appln_title t           ON citinga.appln_id = t.appln_id
JOIN patstat2023.tls203_appln_abstr abstr       ON citinga.appln_id = abstr.appln_id
JOIN cc_core_dataset ON  cc_core_dataset.appln_id = citinga.appln_id
WHERE cc_core_dataset.appln_id = citinga.appln_id

[EPO / PATSTAT Support]

Your query is correct, the best way to check results is to look at one single application (all publications),
and see whether the result is the same as what you can see in ESPACENET or the EPO register. (small differences are possible, but it should be sufficient to confirm that all your JOINS in SQL are correct.

Instead of using your list, change your query: WHERE citinga.appln_id = 17397285, to have a look at the citations in EP1000000.
See espacenet : https://worldwide.espacenet.com/patent/ ... P1000000A1
See EP regsiter: https://register.epo.org/application?nu ... =citations

Results are the same --> query is ok (sometimes you want to text a couple to be sure.)
Whether or not this is the best depends entirely on what you will use the results for. And I am not sure whether a time frame limitation would provide extra insights or filter out any noise. Examiners are free to cite any document they want, whether it's 10 years old or 1 month, doesn't change anything. (at least not for novelty / patentability searches). But cited documents during opposition procedures or third part observations can of cours give a different dimension. (depends on the nature of your research...)

[sarausa]

In your query you join only patents from the core dataset (cc_core_dataset) with itself, this will return only patents from the core dataset without any back citations. What you want is to get the patents that are cited in the core dataset. This can be done by joining the core dataset with the tls212_citation citation Suika game table.

[alicebobby]

You are joining only the core dataset's (cc_core_dataset) patents with itself in your query, which will only return core dataset patents devoid of back citations. Obtaining the patents that are cited in the core dataset is what you desire. The tls212_citation citation table can be joined with the core dataset to achieve this.tunnel rush 

[Joana]

Thanks for your feedback!

Can you provide me with the relevant code for joining the tls212_citation with my core dataset? Shall I replace the last JOIN? Or do I have to add an additional JOIN?

Put differntly: how do I need to modify my code?

Thanks in advance.

[Gambino]

When constructing your query, it's essential to focus beyond just the primary dataset, known as cc_core_dataset, which, by default, merges only with itself, yielding patents solely from this core group without their respective backward citations. To enrich your analysis by including patents referenced within the core collection, you should aim to merge the core dataset with the tls212_citation citation cool math games table. This approach ensures a comprehensive view by capturing both direct and indirect patents cited within your primary dataset.

[turcotte2]

You noted that there may be small differences between the results shown on the EPO register versus what is available in ESPACENET. What types of differences might one encounter, and why is it important to cross-reference the information across these different patent databases?