Does anyone else (apart from me) find that a search can end up with a LOT of references to sort and grade for relevance?
I'd like to see a star-rating system (similar to iTunes) accessible by a hotkey and/or context menu so that I can very easily and ergonomically register how worthwhile an article is to chase. I know I can type into one of the user fields by hand but it would be much easier if I could use either exclusively the keyboard or exclusively the mouse because that would result in a significant ease-of-use when sorting out hundreds of refs.
I have brought it up before but I don't think the idea ever really got discussed... so I thought I'd try again - this time with a poll...
Michael
Star ratings.. (again!!)
Hi Michael,
This is not a "relevance" issue. It is about rating (an article may be a 5 in relevance for one search, but a 1 for another).
FWIW, Bookends 9 will have lots of new fields, and it would be simple to assign one to Rating. And using the Command-' combination in a field copies from the same field in the previous reference, which may make adding/editing ratings a bit easier. Finally, you can add the Rating field to the List View window so you can see the them at a glance (and sort them, too, if you like). One user I know actually inputs stars. I found a lot of symbols in the Characater Palette that would work.
Jon
Sonny Software
This is not a "relevance" issue. It is about rating (an article may be a 5 in relevance for one search, but a 1 for another).
FWIW, Bookends 9 will have lots of new fields, and it would be simple to assign one to Rating. And using the Command-' combination in a field copies from the same field in the previous reference, which may make adding/editing ratings a bit easier. Finally, you can add the Rating field to the List View window so you can see the them at a glance (and sort them, too, if you like). One user I know actually inputs stars. I found a lot of symbols in the Characater Palette that would work.
Jon
Sonny Software
-
michaelward
- Posts: 32
- Joined: Thu Jun 09, 2005 9:30 pm
Hi Jon,
yes it's very true that some references will be relevant for some projects and not others.
I suppose I've used "star ratings" as a term because it's a short hand that mac users understand. And if, like me, you have about 6 projects/talks/publications a year to deal with it's generally easier to keep them in separate databases, which DOES make stars relevant. I should also point out that I tend to keep all references found in searching, marking the irrelevant as irrelevant so that if I have to do more searching I don't end up having to regrade the articles again! It's seems illogical to me to throw away the effort that has gone into deciding that a reference is not relevant.
But what I'm really talking about (using stars as shorthand) is a way of grading particular references when sorting through hundreds at a time.
Instead of stars you could have a "status" field ( containing "not relevant", "abstract is enough", "get the article", "article to be read", "article read" etc etc.). The important bit is the one-click marking. It's exactly the same process that people normally used to go through with a pencil and a printout of their medline search: crossing out the articles they didn't want to chase, ringing the references they wanted to and ticking them as they got them.
[In an ideal world I personally would use both a "status" and a "star rating" field - each filled with a hotkey press. That way I can track what I think of a reference AND what I've done with it or still have to do with it.But I don't want to sound greedy].
Groups don't quite do it because an article can be in more than one group, but they come close. I tend to steer clear of the keywords field because it gets so cluttered with stuff from medline.
I suppose some sort of macro using iKey or Automator or sth would be an alternative, but a bit kludgy. And it still wouldn't be easy to overcome the problem of making sure the field concerned could only contain one value..
I await with interest to see what the result of the poll is.
regards (and Happy New Year)
Michael
yes it's very true that some references will be relevant for some projects and not others.
I suppose I've used "star ratings" as a term because it's a short hand that mac users understand. And if, like me, you have about 6 projects/talks/publications a year to deal with it's generally easier to keep them in separate databases, which DOES make stars relevant. I should also point out that I tend to keep all references found in searching, marking the irrelevant as irrelevant so that if I have to do more searching I don't end up having to regrade the articles again! It's seems illogical to me to throw away the effort that has gone into deciding that a reference is not relevant.
But what I'm really talking about (using stars as shorthand) is a way of grading particular references when sorting through hundreds at a time.
Instead of stars you could have a "status" field ( containing "not relevant", "abstract is enough", "get the article", "article to be read", "article read" etc etc.). The important bit is the one-click marking. It's exactly the same process that people normally used to go through with a pencil and a printout of their medline search: crossing out the articles they didn't want to chase, ringing the references they wanted to and ticking them as they got them.
[In an ideal world I personally would use both a "status" and a "star rating" field - each filled with a hotkey press. That way I can track what I think of a reference AND what I've done with it or still have to do with it.But I don't want to sound greedy].
Groups don't quite do it because an article can be in more than one group, but they come close. I tend to steer clear of the keywords field because it gets so cluttered with stuff from medline.
I suppose some sort of macro using iKey or Automator or sth would be an alternative, but a bit kludgy. And it still wouldn't be easy to overcome the problem of making sure the field concerned could only contain one value..
I await with interest to see what the result of the poll is.
regards (and Happy New Year)
Michael
Why not wait and see how Bookends 9 works for you. I would suggest you designate one of the new drawer fields as Rating, and include this field in your List View. I think that smart groups utilizing this field may also be useful. It won't provide you with one-click editing, but it's not so terribly hard to type ***** (or whatever you want for stars) into the field.michaelward wrote:Groups don't quite do it because an article can be in more than one group, but they come close. I tend to steer clear of the keywords field because it gets so cluttered with stuff from medline.
I suppose some sort of macro using iKey or Automator or sth would be an alternative, but a bit kludgy. And it still wouldn't be easy to overcome the problem of making sure the field concerned could only contain one value..
Another approach would be to go through the references and make hits of the ones you really like (or really don't like), the do a Global Change -> Change Field and put
*****
or
*
INTO the Ratings field of the references in the Hits List.
I'm much happier coming up with solutions that conform to the general way Bookends works than creating ad hoc features that are specific to a rather minor use (I also wouldn't add an FM tuner to the iPod, either, despite the fact that some people seem think it's a good idea
).Jon
Sonny Software
-
michaelward
- Posts: 32
- Joined: Thu Jun 09, 2005 9:30 pm
would be okay... if you didn't lose the focus on that field each time you went to the next record! The result is you either have to switch to the mouse or tab about 15 times to get to the field you're using to store the info in. (Unless 9.0 is going to behave differently?) To anyone who touch-types that gets pretty annoying.Why not wait and see how Bookends 9 works for you. I would suggest you designate one of the new drawer fields as Rating, and include this field in your List View. I think that smart groups utilizing this field may also be useful. It won't provide you with one-click editing, but it's not so terribly hard to type ***** (or whatever you want for stars) into the field
I'd not call it a minor use! In fact I think it's a pretty much universal first reaction when faced with a few hundred references from a search to mark the ones to chase - whether working on paper or computer. At least that is how I see pretty much everyone working around me. And I'm also pretty sure that everyone has a way of keeping track of what stage the chase has got to (c.f. discussion of "status"in previous e-mail). All I'd like is an ergonomic way of doing it.I'm much happier coming up with solutions that conform to the general way Bookends works than creating ad hoc features that are specific to a rather minor use
Anyway... I'm sure the poll will give us the answer - if enough people answer. (although I do wish I'd phrased it as "status" rather than star rating).
Looking forward to trying 9 ...
Michael
Yes, you lose the focus. But it's one click to restore it, and then type or press Command-' to copy the rating from the previous record. I'm sorry, Michael, but I'm not going to add the overhead of a special ratings field, special key combinations, special contextual menus (no doubt with their own preferences) for what can be done with little effort now.
Jon
Sonny Software
Jon
Sonny Software
-
michaelward
- Posts: 32
- Joined: Thu Jun 09, 2005 9:30 pm
I don't think I've explained myself very well then because it's really a bit of a no-brainer.
Forget star ratings... call it reprint-status or whatever. It's a single field designed to classify and keep track of all the articles from a search so that you can mark the ones you are chasing and the ones you have chased etc. which seems to be the way everyone works anyway (whether on paper or pc)
And of course you can already do this using the notes or user fields. But it's CUMBERSOME. All I'm suggesting is something that reduces the effort fairly considerably for users. It would be usable for touch typers without reaching for the mouse and mousers without reaching for the keyboard.
Still if it's not to be it's not to be...
happy new year anyway
michael
Forget star ratings... call it reprint-status or whatever. It's a single field designed to classify and keep track of all the articles from a search so that you can mark the ones you are chasing and the ones you have chased etc. which seems to be the way everyone works anyway (whether on paper or pc)
And of course you can already do this using the notes or user fields. But it's CUMBERSOME. All I'm suggesting is something that reduces the effort fairly considerably for users. It would be usable for touch typers without reaching for the mouse and mousers without reaching for the keyboard.
Still if it's not to be it's not to be...
happy new year anyway
michael
OK, another suggestion: why not create static groups: 5 star, 4 star, etc. (or Highly Relevant, Relevant, etc.). Then all you need to do is click and drag a reference to the group you want and drop it. You can instantly see which references are in each group with a single click on the group itself (you can even AND or OR two different groups easily).
If you want keyboard-only, use the arrows keys to move through the references, press Command-M to mark/unmark the references you want, left-arrow over to the Groups list (or tab) when you are ready, arrow down to Hits, arrow to the right, Command-A and then drag them to the group you want (I tried to keep this all on the keyboard, but the last step does require the mouse -- you could use the contextual menu instead of a drag, s well, to add them to the group.
Jon
Sonny Software
If you want keyboard-only, use the arrows keys to move through the references, press Command-M to mark/unmark the references you want, left-arrow over to the Groups list (or tab) when you are ready, arrow down to Hits, arrow to the right, Command-A and then drag them to the group you want (I tried to keep this all on the keyboard, but the last step does require the mouse -- you could use the contextual menu instead of a drag, s well, to add them to the group.
Jon
Sonny Software
-
michaelward
- Posts: 32
- Joined: Thu Jun 09, 2005 9:30 pm
Hi Jon,
(I wish I hadn't mentioned star ratings because they are clouding the issue. I am actually talking about a the article status in terms of whether you are going to drop it, use the abstract, chase the whole article, have the whole article etc.).
Yes what you sugges is actually what I've been doing since groups showed up in the info pane. But it's still a kludge...: for a couple of reasons
- a reference can belong to more than one group, but can only have one reprint status.
- dragging and dropping is never as safe as using the keyboard - it's all too easy to drop in the wrong place.
It is a lot better than it was before the groups showed up in the info panel, and on the whole I do wholeheartedly agree with your approach of keeping things as simple and generic as possible.But you still aren't going to get me to admit that reprint status isn't one of the most basic aspects of managing references and as such should be incorporated seemlessly!
regards
Michael
(I wish I hadn't mentioned star ratings because they are clouding the issue. I am actually talking about a the article status in terms of whether you are going to drop it, use the abstract, chase the whole article, have the whole article etc.).
Yes what you sugges is actually what I've been doing since groups showed up in the info pane. But it's still a kludge...: for a couple of reasons
- a reference can belong to more than one group, but can only have one reprint status.
- dragging and dropping is never as safe as using the keyboard - it's all too easy to drop in the wrong place.
It is a lot better than it was before the groups showed up in the info panel, and on the whole I do wholeheartedly agree with your approach of keeping things as simple and generic as possible.But you still aren't going to get me to admit that reprint status isn't one of the most basic aspects of managing references and as such should be incorporated seemlessly!
regards
Michael
-
michaelward
- Posts: 32
- Joined: Thu Jun 09, 2005 9:30 pm
how we work
Ok Jon,
not to be a bore but I'd like to revisit the question in a less blinkered fashion since I think I've been attacking it from the wrong end. I'd like a suggestion as to the best way to work. This is the situation ( a real live case - one of many):
- I do a search on a subject, say fluid management peri-operatively. It yields several thousand refs, which I need to whittle down (obviously!)
- I refine the search and end up with say 1500 references which I need to scan read and mark as being uninteresting, vaguely interesting or very interesting.
- For the interesting ones I also need to then mark the ones where the abstract alone is enough vs. the ones where I'm going to need to get hold of the original article.
A few points:
1) VERY importantly, I need to do the above scanning/classifying in several sessions so it is important for me to be able to work only on a list of the references I haven't yet marked as important or not.
2) given the large numbers of references I'm going to have to scan and mark I'd very much appreciate an ergonomic way of doing it. In particular I don't really want to keep switching from mouse to keyboard. It may seem trivial but if I waste 10 seconds a reference then I've lost a lot of time.
3) I 'd rather not eliminate the uninteresting ones by deleting them because I've already invested an effort in marking them and if I do any more searches I'll end up drowning in unnecessary refs and having to mark them again.
I think the above is a fairly common way of working. I've tried your suggestions with Bookends as it stands, but so far I seem to be struggling.
- if I use groups as a way of collecting the refs I'm interested in, it's very hard to find a simple way of not going through ones I've already marked if I want to spread the work over more that one session. If I could define a smart group that used membership of a group as a criteria it'd be trivially easy, but I've not found a way of doing that.
- I can use the user defined fields to mark refs and have tried that. However it means switching from the list view to reference windows continuously and using both mouse and keyboard, which makes it very slow. It is also a little difficult to navigate through a list using the keyboard shortcuts because the command+6-9 keys for navigating through the hits list doesn't follow the sort order for the list view necessarily. And of couse using the command+1-4 buttons takes you through the whole database.
- I'm very nervous of using any system based on the global change function because it is not un-doable and could seriously trash a whole database.
At the moment I'm struggling with a system that involves using the groups then using a global change to change a user field then sorting the list view by that user field and using a search to define a new hits list which I then sort into id order so I can go through it using the navigation keyboard shortcuts.. It's rather complicated and cumbersome and I've come unstuck more than once!
Please be patient with me. Although you probably feel you've explained it before can you suggest a way of working with the current functionality of Bookends to accomplish the above?
regards
Michael
not to be a bore but I'd like to revisit the question in a less blinkered fashion since I think I've been attacking it from the wrong end. I'd like a suggestion as to the best way to work. This is the situation ( a real live case - one of many):
- I do a search on a subject, say fluid management peri-operatively. It yields several thousand refs, which I need to whittle down (obviously!)
- I refine the search and end up with say 1500 references which I need to scan read and mark as being uninteresting, vaguely interesting or very interesting.
- For the interesting ones I also need to then mark the ones where the abstract alone is enough vs. the ones where I'm going to need to get hold of the original article.
A few points:
1) VERY importantly, I need to do the above scanning/classifying in several sessions so it is important for me to be able to work only on a list of the references I haven't yet marked as important or not.
2) given the large numbers of references I'm going to have to scan and mark I'd very much appreciate an ergonomic way of doing it. In particular I don't really want to keep switching from mouse to keyboard. It may seem trivial but if I waste 10 seconds a reference then I've lost a lot of time.
3) I 'd rather not eliminate the uninteresting ones by deleting them because I've already invested an effort in marking them and if I do any more searches I'll end up drowning in unnecessary refs and having to mark them again.
I think the above is a fairly common way of working. I've tried your suggestions with Bookends as it stands, but so far I seem to be struggling.
- if I use groups as a way of collecting the refs I'm interested in, it's very hard to find a simple way of not going through ones I've already marked if I want to spread the work over more that one session. If I could define a smart group that used membership of a group as a criteria it'd be trivially easy, but I've not found a way of doing that.
- I can use the user defined fields to mark refs and have tried that. However it means switching from the list view to reference windows continuously and using both mouse and keyboard, which makes it very slow. It is also a little difficult to navigate through a list using the keyboard shortcuts because the command+6-9 keys for navigating through the hits list doesn't follow the sort order for the list view necessarily. And of couse using the command+1-4 buttons takes you through the whole database.
- I'm very nervous of using any system based on the global change function because it is not un-doable and could seriously trash a whole database.
At the moment I'm struggling with a system that involves using the groups then using a global change to change a user field then sorting the list view by that user field and using a search to define a new hits list which I then sort into id order so I can go through it using the navigation keyboard shortcuts.. It's rather complicated and cumbersome and I've come unstuck more than once!
Please be patient with me. Although you probably feel you've explained it before can you suggest a way of working with the current functionality of Bookends to accomplish the above?
regards
Michael
Hi Michael,
Maybe someone who takes the approach you do to dealing with references has a better answer than I, because I've certainly never done anything like that. I'll offer a suggestion that I'm sure you won't particularly like, but I think it will work. And if you really have to manually go through 1500 references, *no* tricks will make it a short job. Anyway...
1. If you get 1500 refs you want to keep around, make them a static group first. These will "live" between sessions.
2. Make two more groups: (a) interesting and (b) get abstract
3. Use the mouse to move references to the group you want (either drag and drop, or right-click and use the contextual menu).
4. If you want to know where you left off, click on the Mark button (it can tell you that you've already considered that ref, or just mark the last one you looked at so when you start another session you know where you left off).
Another possibility is to use the Notes field in the Info Drawer. It is editable, and you can add text to the reference without actually opening the reference window. It could be a single letter code, if you like, that you could enter with the left (non-mouse) hand. Something like 'a' for get abstract, 's' for significant. Or whatever. Then you could use smart groups.
(Parenthetically, you should navigate through the list with up/down arrows, not CommandKey combinations -- the list is sorted by column, unless you tell Bookends not to (in Preferences)).
Jon
Sonny Software
Maybe someone who takes the approach you do to dealing with references has a better answer than I, because I've certainly never done anything like that. I'll offer a suggestion that I'm sure you won't particularly like, but I think it will work. And if you really have to manually go through 1500 references, *no* tricks will make it a short job. Anyway...
1. If you get 1500 refs you want to keep around, make them a static group first. These will "live" between sessions.
2. Make two more groups: (a) interesting and (b) get abstract
3. Use the mouse to move references to the group you want (either drag and drop, or right-click and use the contextual menu).
4. If you want to know where you left off, click on the Mark button (it can tell you that you've already considered that ref, or just mark the last one you looked at so when you start another session you know where you left off).
Another possibility is to use the Notes field in the Info Drawer. It is editable, and you can add text to the reference without actually opening the reference window. It could be a single letter code, if you like, that you could enter with the left (non-mouse) hand. Something like 'a' for get abstract, 's' for significant. Or whatever. Then you could use smart groups.
(Parenthetically, you should navigate through the list with up/down arrows, not CommandKey combinations -- the list is sorted by column, unless you tell Bookends not to (in Preferences)).
Jon
Sonny Software
-
michaelward
- Posts: 32
- Joined: Thu Jun 09, 2005 9:30 pm
Thanks Jon, I'll keep soldiering on with your suggestion.
BTW it's not unusual when trying to define best practice in many medical fields to have to scan a large number of articles because there is no easy way to limit a search on say fluid therapy or preoperative assessment - the subjects are too general. Even a limited field such as cricothyroidotomy yields a good few hundred refs.
Using hotkeys with Sente, or tickboxes on a Palm with a database downloaded from Bookends I find I can scan up to 10 or so a minute - so it's only a couple of hours work to fully survey a very general field (provided the ergonomics are right). And with automatic searching you are then on top of it forever, so can keep your lecture or clinical practice up to date.
Maybe you have considered this before because you are concentrating on thte academic side of things rather than the lowly clinician...
regards
Michael
BTW it's not unusual when trying to define best practice in many medical fields to have to scan a large number of articles because there is no easy way to limit a search on say fluid therapy or preoperative assessment - the subjects are too general. Even a limited field such as cricothyroidotomy yields a good few hundred refs.
Using hotkeys with Sente, or tickboxes on a Palm with a database downloaded from Bookends I find I can scan up to 10 or so a minute - so it's only a couple of hours work to fully survey a very general field (provided the ergonomics are right). And with automatic searching you are then on top of it forever, so can keep your lecture or clinical practice up to date.
Maybe you have considered this before because you are concentrating on thte academic side of things rather than the lowly clinician...
regards
Michael
Hi Michael,
I used to be a lowly clinician myself. I'll give it some more thought. If you want to continue this discussion at a future time, please contact me directly (support@sonnysoftware.com) and we'll talk about it.
Jon
Sonny Software
I used to be a lowly clinician myself
. I'll give it some more thought. If you want to continue this discussion at a future time, please contact me directly (support@sonnysoftware.com) and we'll talk about it.Jon
Sonny Software