I've been using a number of online tools which now allow me to easily tag my data: Flickr and del.icio.us are prime examples. In using these tools I have become increasingly discontent with the way "keywords" are implemented in Bookends, but I think a few simple changes could fix that.
First of all, right now Bookends insists on using the keywords field for attachments as well. That clutters up the keywords with irrelevent data.
Secondly, there doesn't seem to be any way to force Bookends to accept spaces as a deliminator for keywords. As a result, many keyword lists I import from online databases are treated by Bookends as a single keyword in the kewords list. e.g. "Cars Boats Trains" is treated as ONE keyword!
Third, I would like to see adding something to a STATIC group automatically add the name of that group as a keyword. This is good for database stability since "user info" including groups is sometimes lost when the database gets corrupted.
Forth, the Keyword list is a long list. It would be nice to implment the concept of a "Tag Cloud" as seen on the Flickr home page. See here for more info:
http://en.wikipedia.org/wiki/Tag_cloud
It would be nice if one could browse by clicking on keywords in the Tag Cloud. It would also be nice if the Tag Cloud could display ALL keywords, or only RECENTLY USED keywords.
Finally, while it is currently possible to do batch operations with the search and replace function, it is very dangerous as it is all too easy to apply changes to fields other than keywords, or to references not on the hit list, etc. Instead, I think it would be nice to have simple tools to manage one's keywords, such as "replace keyword" (useful if you have a bunch of instances of a typo such as "langauge" or if half your instances of a keyword have a plural ending, while the other half don't), or "group keywords" whereby you could group together keywords that often appear together or are related semantically (apples and oranges would be listed together in the "fruit" bundle - del.icio.us already offers such functionality and I find it quite useful).
I understand that some of these suggestions might be harder to implement than others, but I think revamping the keword functionality in Bookends would bring numerous advantages.
Keywords = Tags ?
Re: Keywords = Tags ?
This is already fixed in the betaLuhmann wrote:First of all, right now Bookends insists on using the keywords field for attachments as well. That clutters up the keywords with irrelevent data.
That would be a nifty feature! (perhaps implemented via a context menu?) All of your suggestions sound very interesting and I think it might be worthwile to ponder ways in which tags (we've got labels in Bookends too - somewhat similar) and keywords could be made useful in new ways.Luhmann wrote:It would be nice if one could browse by clicking on keywords in the Tag Cloud.
Bookends uses Returns. But the import filters let you define what a delimiter is. Tell it that Keywords are delimited by spaces, and Bookends will place each word on a separate line.Luhmann wrote:Just saw the beta. Your right! That's one down ... I'm sure some of the others will be fixed as well. The highest priority for me would be to see user-definable deliminators, if they aren't already implemented some how...
Jon
Sonny Software
Re: Keywords = Tags ?
I think Luhmann is on to something here. This concept of tags and tag clouds really is a revolutionary leap in organizing and interacting with your data. Tags bring 'groups' (as in static and smart groups) to a whole new level. With today's static groups, you have to manually link a citation to a group. And with smart groups, you have to know what kind of characteristics you want to the group have. With tags, the organization happens organically: You add a tag to a citation, and the 'cloud' of tags shows you your universe of organization -- without ever having to impose a 'group' on it. You don't have to know what you're looking for; if you've tagged an item, it is magically relevant.
Compare this to what you'd have to do to achieve this now in Bookends: define smart groups for all of your keywords - you'd have to know what you're searching for in order to even create the first group!
In other words, a tag makes a single record significant -- and no manually-defined, keyword-based smart group is necessary.
Luhmann's suggestions for additional tools to augment tags are very good. They let you tweak tags and refine them through batch processes restricted to the keywords fields.
This would require some UI work - a tag cloud view. But I'd bet that the database underlying Bookends could easily trounce the challenge.
Compare this to what you'd have to do to achieve this now in Bookends: define smart groups for all of your keywords - you'd have to know what you're searching for in order to even create the first group!
In other words, a tag makes a single record significant -- and no manually-defined, keyword-based smart group is necessary.
Luhmann's suggestions for additional tools to augment tags are very good. They let you tweak tags and refine them through batch processes restricted to the keywords fields.
This would require some UI work - a tag cloud view. But I'd bet that the database underlying Bookends could easily trounce the challenge.
Luhmann wrote:It would be nice to implement the concept of a "Tag Cloud" as seen on the Flickr home page. ...
It would be nice if one could browse by clicking on keywords in the Tag Cloud. It would also be nice if the Tag Cloud could display ALL keywords, or only RECENTLY USED keywords.
...
I think it would be nice to have simple tools to manage one's keywords, such as "replace keyword" ... or "group keywords" whereby you could group together keywords that often appear together or are related semantically (apples and oranges would be listed together in the "fruit" bundle - del.icio.us already offers such functionality and I find it quite useful).
Re: Keywords = Tags ?
Without debating the merits of tags, note that Term Lists give you some of this now. Select a keyword and you'll see how many times it is used and a list of all references that use it.joewiz wrote: Compare this to what you'd have to do to achieve this now in Bookends: define smart groups for all of your keywords - you'd have to know what you're searching for in order to even create the first group!
And with multiple selections and the AND/OR boolean selector, you can quickly see all the references that use a set of keywords.
As for group keywords (fruit = apple, orange, etc.), smart groups can fulfill that function.
Jon
Sonny Software
But what about changing the keywords after import? When I download references from libraries, the keywords tend to be long subject phrases, such as "Italy Civilization 1268-1559," which Bookends treats as a single keyword.Jon wrote:Bookends uses Returns. But the import filters let you define what a delimiter is. Tell it that Keywords are delimited by spaces, and Bookends will place each word on a separate line.
To fall out with the other posters in this thread, I think revamping Bookends keyword-functionality in the "tag-cloud" direction is a terrible idea. For me at least, databases are a way of reducing information chaos to manageable dimensions so as to facilitate the further use of the data gathered. To accomplish this, one needs a surprising degree of old-fashioned rigidity - I am all for innovation and the use of new techniques, but it is not by chance that research libraries are organised the way they are. Bookends should largely stick to these same principles, I guess, if it wants to remain a serious research tool.
But perhaps I am overlooking research needs which do not exist in my own field, so I would really appreciate it if the earlier posters could describe in somewhat more detail what could be achieved by the use of tag clouds other than research being led into the direction of the topics best represented in a database rather than in the direction of the topics formulated by the researcher on the basis of scientifically relevant assumptions and hypotheses.
Sorry if this came out a bit harsh, but really I often have the feeling of being choked by the ever-increasing stream of information and therefore feel quite strong about the precious tools which help reduce it to manageable proportions.
But perhaps I am overlooking research needs which do not exist in my own field, so I would really appreciate it if the earlier posters could describe in somewhat more detail what could be achieved by the use of tag clouds other than research being led into the direction of the topics best represented in a database rather than in the direction of the topics formulated by the researcher on the basis of scientifically relevant assumptions and hypotheses.
Sorry if this came out a bit harsh, but really I often have the feeling of being choked by the ever-increasing stream of information and therefore feel quite strong about the precious tools which help reduce it to manageable proportions.
Tag clouds are a different way of representing the SAME information, not a different kind of information. Perhaps GKE is thinking of "social tagging" which is often used by the same web sites that use tag clouds? I suggest reading some Edward Tufte for further clarification as to how the visual display of information can help make that information more useful.
Already Bookends allows users to either use existing keywords, such as those supplied by libraries (although various online services and libraries use different systems), or replacing those keywords with ones that the individual scholar might prefer to use. In both cases there are problems. For instance, the current online databases use many different ways of deliminating keywords, and it is hard to anticipate these problems when importing each different kind of data set - it would therefore be nice to have tools to make it easier to handle this after-the-fact. Secondly, for those of us who like to code our own data rather than letting others do it for us, there could be some improvements as well which would make it easier to both enter and view this information.
In any case, nobody is suggesting removing the existing functionality of Keywords, simply offering additional tools to leverage that information in new ways.
Already Bookends allows users to either use existing keywords, such as those supplied by libraries (although various online services and libraries use different systems), or replacing those keywords with ones that the individual scholar might prefer to use. In both cases there are problems. For instance, the current online databases use many different ways of deliminating keywords, and it is hard to anticipate these problems when importing each different kind of data set - it would therefore be nice to have tools to make it easier to handle this after-the-fact. Secondly, for those of us who like to code our own data rather than letting others do it for us, there could be some improvements as well which would make it easier to both enter and view this information.
In any case, nobody is suggesting removing the existing functionality of Keywords, simply offering additional tools to leverage that information in new ways.
Sure enough the keywords imported from other databases are indeed posing problems - in fact I adapted all BE import filters to skip the keyword field when importing records, preferring as well to tag my data according to individual preferences. What I have came to value over the years, however, are the substantial benefits to be derived from rigid keyword lists, in which one has to choose keywords from an existing list, possibly inserting a new one if the need arises to code something which cannot be coded with the existing categories. In other words, one has to discipline onself in data-entry, with the greater goal of database-consistency in mind. In my experience, only alphabetically ordered term lists impose the transparency necessary to avoid tagging the same type of information with different categories.
As for the benefits of a visual representation of information - well, this is of course a matter of personal preference, and if BE can offer both options, so much the better, but what I wanted to express with my earlier post was the feeling that information increasingly tends to get presented in elusive, interactive and post-modern ways, which reduce its usability in spite of its larger availability.
Finally, a tip, when keywords from online databases foul up your term-lists; after deleting the faulty keywords, click the Update button next to the term list - the keywords which no longer exist disappear from the list.
As for the benefits of a visual representation of information - well, this is of course a matter of personal preference, and if BE can offer both options, so much the better, but what I wanted to express with my earlier post was the feeling that information increasingly tends to get presented in elusive, interactive and post-modern ways, which reduce its usability in spite of its larger availability.
Finally, a tip, when keywords from online databases foul up your term-lists; after deleting the faulty keywords, click the Update button next to the term list - the keywords which no longer exist disappear from the list.
Whether to use "rigid keyword lists" or to go wild with free-association word-salads is a matter of personal preference, of course, but I don't think the software should dictate the answer, it should allow people to pick and choose those tools which work best for them. I used to be a "list" type of guy, but since using other tools I've changed my tune. Thinking about this made me realize something:
One thing that makes word-salads so useful is the ability to see RELATED keywords, something Bookends does not do well. For instance, if I have five items with the words "car boat train" and five items with the words "car train bike" it would be great to see "boat train bike" listed as related when I select something with the word "car." It is hard to explain this, but both Flickr and Del.icio.us do this very well - allowing one to file information more rapidly and find interesting associations you would not have otherwise thought about during the discovery/research phase. I believe that even people using carefully constructed keyword lists would benefit from such functionality.
One thing that makes word-salads so useful is the ability to see RELATED keywords, something Bookends does not do well. For instance, if I have five items with the words "car boat train" and five items with the words "car train bike" it would be great to see "boat train bike" listed as related when I select something with the word "car." It is hard to explain this, but both Flickr and Del.icio.us do this very well - allowing one to file information more rapidly and find interesting associations you would not have otherwise thought about during the discovery/research phase. I believe that even people using carefully constructed keyword lists would benefit from such functionality.
I see what you mean, and I guess that what I am trying to factor out when compiling my rigid lists is precisely these cross-links because they tend to inflate the number of hits, while what I am trying to achieve when doing a database search is get as precise hits as possible.
Of course, if you need the cross-links, I agree with you that a visual presentation is better suited for such purposes, but isn't the danger that in the end any sort of search renders about 70% of all records in the database because of all the cross-links?
Of course, if you need the cross-links, I agree with you that a visual presentation is better suited for such purposes, but isn't the danger that in the end any sort of search renders about 70% of all records in the database because of all the cross-links?
Not at all - because you use that information to narrow your list. For instance, if you see that "cars" is related to "trains" you can then search for "cars + trains" which is more precise than the more general "cars" you would have in a rigid list! In fact, del.icio.us does this automatically, allowing you to add additional related keywords to narrow your search!