To identify duplicate references in your database, use the "Select Duplicates" item of the "Database" menu. BibDesk compares the type and all standard fields for that type, and selects all publications for which a duplicate with the same values for those type and fields can be found. You can choose whether you want to select all the duplicates, or to leave out one duplicate for each set (the "original", which will be chosen at random).
Crossref and type are compared case-insensitively, while other fields are compared case-sensitively without removing any TeX-specific commands.
For duplicates in the main database, also the cite key and default fields are included for comparison. For duplicates in external groups, those are ignored.
For a less rigorous comparison, choose "Select Duplicates by <Field>" from the "Database" menu. With this option, similar to the functionality of BibTool, BibDesk compares adjacent values in the currently sorted column of the main publications list; if it finds an item whose value is identical in the succeeding cell, it selects both items. For instance, if you sort by first author and use "Select Duplicates by First Author" and have four publications with the same first author, all of the four publications by that author will be selected. Obviously, author may not be the best choice in this case if you are searching for duplicate publications!
The number of duplicates found will be visible in the status line at the bottom of the window. These are not necessarily identical items, however, so manual review is necessary. For easy review, choose "New Group With Selection" from the "File" menu, which will create a group named "Untitled" with the selected publications (this uses the current group field).