Issue Number | 349 |
---|---|
Summary | [Citations] Add ability to link related citations |
Created | 2015-12-09 17:49:16 |
Issue Type | New Feature |
Submitted By | Shields, Victoria (NIH/NCI) [E] |
Assigned To | Kline, Bob (NIH/NCI) [C] |
Status | Closed |
Resolved | 2016-01-18 18:05:26 |
Resolution | Fixed |
Path | /home/bkline/backups/jira/oceebms/issue.175366 |
We would like to explore the possibility of linking (in some way) related citations. An example would be linking an article to an editorial/comment/correspondence/letter to the editor.
To keep this from ballooning into a big-ticket item, I propose that we support this request with a button on the "full citation" page for the article from which the user wants to create a link. The button will bring up a form in which the user enters the EBMS ID of the article to which the link will point, selects from a list of relationships, and (optionally) enters a comment. When the link request is submitted, the link is inserted in the database. Whether the user submits or cancels the request, she is returned to the "full citation" page, which will include display of the relationship links to other articles. Display of the link relationships elsewhere (reports, etc.) would be addressed by separate tickets for each page which needs to be altered. Deletion of links could be handled by a report page for the link relationships (again, separate ticket). If we adopt the approach described here, the LOE for this ticket should be about 13 story points.
This seems like a reasonable approach to us - thanks! We'd like to implement just the linking for now (we'll determine what reports and other pages need modifications down the road, and enter separate tickets for those, as you suggest). We think this solution could work for a few different issues, including the need to associate records for citations that have changed PMIDs.
I clarified a few points below.
1. Please add LINK RELATED CITATION to the CITATION ACTIONS box on the left side of the page. This would call up the form you proposed.
2. On the form, the list of relationships should include "Duplicate", "Editorial/Comment" (ok to use "or" if slash is not allowed), and "Other".
3. On the Full Citation page, please display the related citations near the top of the page, just below REVIEW CYCLE and above the tags. The heading should be RELATED CITATION(S) and only be present if there are related citations linked to the record. We propose the following display:
RELATED CITATION(S)
(indented) DUPLICATE: PMID (with link to related record)
(indented) COMMENT: Free text comment displays here.
(indented) EDITORIAL/COMMENT: PMID (with link to related
record)
(indented) COMMENT: Free text comment displays here.
Please display a comment icon to the left of each comment field, as we do for other fields on the page. Please let us know if anything else needs clarification. Thanks, again!
Estimate adjusted to incorporate requested modifications.
Since the table structure only supports a single comment for a given relationship, the "Add New Comment" icon won't work here. Instead I'll make it possible to edit the relationship (allowing you to change the comment).
Decisions at the status meeting:
1. Please display both the PMID and the EBMS ID on the Full Citation page and link the PMID to PubMed and the EBMS ID (in parentheses) to the EBMS record.
2. Please order the related citations first by type, subsort by PMID.
Changes from the previous comment have been implemented.
... only be present if there are related citations linked to the record.
Just to be clear, you meant "... linked from the record," right?
I'm a little confused by your question. I think the relationship should ideally be reciprocal (i.e., if you add a link from citation A to citation B then you would see the related citation information in both the record for citation A and the record for citation B without having to update citation B as well), but I know we didn't explicitly say that so I'm not sure how this has been implemented. And it looks like you're currently working on the full citation page so I can't check. 🙂 If the links are reciprocal, then I think the answer to your question would be both "to" and "from"... If they are not programmed this way, would it be possible to make the relationships reciprocal? We can do this in the next release if need be (along with deleting relationships, as we discussed today, since we'll have duplicate relationships to contend with once we make them reciprocal.)
At least in some (perhaps most?) of the cases the relationships are not reversible without distortion of the semantics. If I say that article A is a commentary on article B it does not logically follow that article B is a commentary on article A. Similarly if a new Pubmed record D is created, and we discover that it duplicates an original record O for the same journal article, it would be misleading to say that O is a duplicate of D (at least for someone using the original definition of duplicate, meaning "[make] an exact copy of").
What if we show both the related articles to which the current article links (as we do now), as well as the articles which link to the current article, and include an indication of which direction is being displayed ("linked [to|from] 99999999 (123456)")?
If instead you prefer to have purely reciprocal links, for which the semantics normally conveyed by the direction are discarded, we'll want to make that decision up front, making sure that everyone is aware that they will need to repeat the article IDs in the comments and spell out the true nature of the relationship there (and we'll need to change the current descriptions of the relationship types, which imply that we're going to preserve the meaning carried by the direction of the link).
Does all of this make sense?
The page should be working now.
Have any thoughts about how you want relationship directions handled? Do you want me to implement the enhancement I suggested in the previous comment ("What if we show both ...")?
I went ahead and implemented support for deleting relationships on the "full citation" page (as you said you preferred).
Sorry, Bob. I thought I had responded to this on Friday but unfortunately JIRA never took my comment.
I think the first scenario you propose in the previous comment ("What if we show both the related articles...") would work well for us, but we're not sure there's really any value in preserving whether the link was made "to" or "from" the related citation. Do you see a reason to keep that? We think this could become inconsistent over time since we'll likely make the link from whichever page we're on, so it wouldn't be useful. What if we drop the to/from and just list the related citation regardless of the direction of the link? (We propose a change to the editorial relationship type below - to ARTICLE/EDITORIAL - to account for this change in semantics and allow for links in either direction.)
Another question: If we go with this solution, and we link article A to editorial B from the full citation page for article A, what exactly would display on the related citation section on the full citation page for editorial B? I'm expecting that it would display the PMID and EBMS ID for article A, but just wanted to check.
Here are a few additional changes to the relationship types and definitions from what's on DEV. The 'WAY BETTER' category looks great as is. ;-)
DUPLICATE (This article has a duplicate record [e.g., the PMID has changed].)
ARTICLE/EDITORIAL (This article [or editorial] has a related editorial [or article].)
OK, I believe I have it working the way you want (on DEV).
This seems to be working well - thanks!
We noticed a couple of things on the page to link relationships:
1) Please change EDITORIAL/COMMENT to ARTICLE/EDITORIAL.
2) Seems like there are a lot of parentheses. To reduce those, could you please capitalize and add a period after each relationship type and before the description (and remove the outer parentheses). Could you please also place the description in italics? For example:
DUPLICATE. This article has a duplicate record (e.g., the PMID has changed).
Did you mean "Capitalize" or "UPPERCASE" in your previous comment?
UPPERCASE, please.
Modifications implemented.
This looks good, but we have one more tweak to the description for OTHER.
OTHER. This article is related in some other way to another article.
Done.
Verified on QA SG.
Verified on PROD.
Elapsed: 0:00:00.000732