Yes, please provide links to Catalog IDs. I'm flabbergasted that this isn't there already...

This is a long-standing enhancement request and bug, also covered by:

http://feedback.mendeley.com/forums/4941-mendeley-feedback/suggestions/251223-citation-table-view-mix-up-after-search

http://feedback.mendeley.com/forums/4941-mendeley-feedback/suggestions/547812-preserve-view-format-when-searching

Please provide a better (and functioning) view for search results. Ideally, there would be no "search results" view. Instead, what is displayed (search, folder, filtering) would be divorced from how it's displayed (table, citation).

Although this feedback item is a RFE, it is related to a real bug (see above links). Here's a screenshot of the bug in practice:
http://static.inky.ws/image/631/image.jpg

If possible, please merge these related tickets into one request.

Thanks!

Robert- Thanks for the update. Do you have a guess when something might be available? (I don't expect a date, but a month would be nice.)

Hey, Mendeley, can we get an update on this bug please? It's been around a month since any official comment.

Robert- I identified a diabolical test case for Mendeley deduping. See http://harts.net/reece/tmp/mendeley-dupes/abeel-prosom/. There are three files. Two are PDFs of the same manuscript but differ in distiller, PDF version, and indexes (I dunno why). The other PDF is a poster abstract by the same authors. All three PDFs are filed separately with the same metadata by Mendeley.

Mendeley should do two things differently here:
1) at least offer to merge the two legit manuscript entries, ideally showing content and property differences to aid discrimination, and ask which files (one or more) to associate with the merged entry;
2) file the abstract as something else, or perhaps as supplementary material to the merged entries in (1).

This test was on a fresh MD instance, v. 0.9.8.2, on Ubuntu 10.10.

Robert- Given the debilitating nature of this problem for many of us, would you please help set expectations for the nature and timing of a fix? Thanks.

Robert- It's quite possible that my duplicates preexist the implementation of deduping in Mendeley Desktiop. If I understand your reply correctly, there's nothing that would reconcile existing duplicates. Correct? If you're looking for someone to try out deduping an existing library, sign me up. Thanks.

Hi Robert-

I don't have a clear understanding of which actions contributed most, but here is a list of suspicions, roughly in decreasing order of perceived impact. Bear in mind that I am uncertain that any of these actually cause duplication.

• Mendeley Desktop crashing. I'm a long time user, and I have no idea what junk I've accumulated through crashing over the years. Some time ago, crashing was a daily event. Duplication seemed to creep during recovery, but I can't be certain.
• Partly as a result of the above, I have once or twice tried to start from scratch with "Reset Mendeley Desktop". Of course, I have lots of citations that get synced back to MD. Then, I've tried to reattach PDFs by reloading them via a watched folder. This seemed to work fine for some articles and fail for others.
• Many of my documents need manual metadata edits (title, author, etc). Adding a doc, changing the metadata, re-adding the doc may result in duplication.
• I use watched folders as my primary means to upload. I'm sure that I occasionally re-add files via this route.
• I also occasionally add the same citation through the web interface and watched folders, and THEN sync (thus creating a collision of the same citation through two routes).

It would help a lot to know what criteria you guys for deduping. The naive way I'd like this to work is to store everything based on a sha1, make it unique in the database, attach all meta data to the sha1, and refine from there for alternate ids like doi, pubmed ids, etc.

I don't have inside knowledge of the schema, but in a little poking around, I see a bunch of duplicated sha1s and I have a hard time understanding why that's useful. See http://harts.net/reece/tmp/hash-dups.txt for the data.

I've also stopped using Mendeley because of the duplication problem. I'm seeing 2-8x duplication and about 3x on average, which makes searching, citing, browsing, classifying, and annotating nearly unusable.

I'd like the following out of this feature request: 1) deduping files (by sha1 is good, by meta data is better to avoid "trivial" duplicates due to in-PDF timestamps); 2) develop a canonical format for titles and author names and dedup them (I currently have entries like "Last, F", "Last, F.", "Last, First" for the same person, and titles that differ only by a terminal period.), 3) A mechanism to re-retrieve metadata (e.g., from pubmed) and reclassify. Thank you!

And to add to this, I'd like the public notes to appear in the RSS feeds and embedded iframe views of public collections. (Probably should be an option with default off.)

I too would like ranking... 0-3 stars in lieu of the current favorites star.