Designing Your QuickTag Hierarchy
For many Sente users, the QuickTags feature is the single most useful tool for
organizing their library. For these users, QuickTags make classifying and accessing
references easier and more efficient than any approach. But to get the most benefit
from QuickTags, it is important to think carefully about how you set up your
QuickTag hierarchy. This section introduces the most important concepts to
understand during this process.
There are two key ideas that should be kept in mind when designing your QuickTag
•First, the QuickTag hierarchy should be viewed as a collection of independent
hierarchies, rather than as a single hierarchy. That is, each top level tag is
typically the root of an independent hierarchy.
•Second, each of these independent hierarchies should be either what is
commonly called a containment hierarchy or an is-a hierarchy. Other types of
hierarchies will often lead to confusion and inefficiency.
These two concepts are explained in more detail below.
Multiple, Independent Hierarchies
Modern, hierarchical computer file systems are obviously powerful, and they affect
the way many people think about organizing information, but they are not, in fact,
very good models for organizing complex data.
The main problem is that, without resorting to a complex system of file aliases or
links, each file is stored in a single location in the hierarchy of folders on a
computer. This structure starts to break down as soon as you care about more than
one aspect of a file. For example, if you are classifying material on military history,
do you put folders for topics like battle tactics and political context under folders
for each conflict you are studying, or do you place folders for each conflict under
folders for battle tactics and political context? Neither answer is obviously correct
and both are difficult to maintain and use.
With QuickTags you can treat each of these as separate dimensions on which you
want to classify your material. You would do this by creating several hierarchies
within the QuickTag hierarchy, each starting from a different top-level term, and
assigning one or more tags from each relevant hierarchy to each reference as
For example, in the case of military history, you might create a top-level tag called
Conflict with sub-tags for the World War I, World War II, the American Civil War, etc.
Under each of these you might create tags for particular battles within each war.
Then, you might create another top-level tag called Battle Tactics, with sub-tags for
Offensive Tactics and Defensive Tactics, and so on.
With a design like this, you would assign tags from the Conflict category to any
works that discuss specific wars or battles, and tags from the Battle Tactics category
to works that specifically deal with those topics. Many works would be tagged with
at least one tag from each top level category. With this approach, you would never
have to wrestle with deciding whether you wanted to create a folder for specific
battles inside a folder on tactics, or the other way around. Nor would you have to
decide which aspects of a work were most important -- you would simply assign all
relevant tags to each reference.
One way to think about these multiple hierarchies is to consider that they represent
different dimensions on which you might classify each work. It is common to end
up with QuickTag hierarchies with several different dimensions, or sub-hierarchies.
Types of Hierarchies
Starting with the idea that the QuickTag hierarchy is really a collection of several,
distinct sub-hierarchies, one then needs to decide how to design each of these
hierarchies. Because of the way QuickTags work in Sente, it is often best if each
hierarchy is either a containment hierarchy or an is-a hierarchy. These are
A containment hierarchy is one in which all child tags are parts of, or contained
within, the parent tag. For example, the tag 20th Century might have the sub-tags
Early 20th Century and Late 20th Century (the early 20th century is part of the 20th
century). Or a tag related to the field of study might be Economics, with sub-tags for
Microeconomics and Macroeconomics (Microeconomics is part of the study of
economics). Or here is a larger example related to geography:
A sample containment hierarchy based on geography.
In each case, the child tags represent either a part of, or something contained
within, the parent tag.
Using any containment hierarchy, you would tag a reference with the most specific
applicable tag. All parent tags will be implied by this assignment. Searches within
Sente will behave as though all of the implied tags were also assigned to the
For example, if you tag a reference with Chile from the geography hierarchy
example shown above, and later search for references tagged with South America,
the reference would be returned because the South America tag is implied by the
Furthermore, because the QuickTag hierarchies are evaluated as they are used
(rather than when tags are assigned), if you were to introduce a new level in the
geography hierarchy for Western Hemisphere, and move everything shown above to
be below this new tag, any reference tagged Chile would immediately be returned in
a search for references tagged Western Hemisphere because that would now be
In containment hierarchies, it is usually best to make all child tags represent non-
overlapping concepts or spaces. For example, if one were expanding the geography
example to include specific states within the United States, these might go directly
under the United States tag. If one later wanted to add larger regions (e.g., Western
States, Southern States, etc.) to the hierarchy, the individual state tags should be
moved to be under the new region tags, rather than leaving them at the same level
as the new tags. If this is not done, the implied tags will yield incorrect results.
In containment hierarchies, it is common to make the top-level tag of the hierarchy
be a description of the hierarchy, rather than an actual member of the hierarchy.
For example, in the geography example above, the top-level tag might be
Geography. This lets you collapse all geography-related tags under one parent tag.
The other common type of hierarchy used in Sente is sometimes called an is-a
hierarchy, or a class hierarchy. In this type of hierarchy, each child is an example of,
or a sub-class of, the parent. For example, if one were to classify writing, one might
have a top level tag Literature, with sub-tags for Fiction and Non-Fiction (“fiction is a
type of literature”). Non-fiction might then have sub-tags like History, Biography,
etc. (“a biography is a type of non-fiction literature”). As another example, one might
be studying modes of transportation:
A sample is-a hierarchy based on transportation vehicles.
If you consider each tag in the example above, you could say it is an example of the
parent tag. For example, a Sedan is a Passenger Car, which in turn is a Motorized
As with a containment hierarchy, references should normally be tagged with just the
most appropriate tags from an is-a hierarchy. In our example above, a work that
discusses non-motorized transportation in general should be tagged Non-Motorized
Vehicles but a work that discussed different types of bicycles explicitly should be
tagged with each type of bicycle (and not Non-Motorized Vehicles, which will be
One way to decide which level of tag to apply is to ask yourself whether you would
expect a particular reference to be returned if you were to search on each of the
tags under consideration. For example, a paper describing trends in personal
transportation in China may discuss bicycles extensively. Tagging such a reference
with only the Non-Motorized Vehicles tag would mean that a search on Bicycles
would miss this paper. On the other hand, if it makes little distinction between
various types of bicycles, tagging the paper with Road Bikes and/or Mountain Bikes
would mean that searches on either of these terms would return the paper, even
though it has little to say on either specific subject.
In the case of an is-a hierarchy, the top-level node is typically the most generic term
in the hierarchy (unlike in a containment hierarchy -- see the Containment
Putting It All Together
When you set up your QuickTag palette as described above, with several
independent hierarchies, each either a containment or an is-a hierarchy, you will find
that you can quickly classify new references as you enter them into your library.
You will get used to asking yourself questions like “What region of the world does
this reference discuss?” or “Which modes of transportation are considered?”
Part of the power of this approach becomes apparent when you find yourself looking
for a subset of your references that you may not have explicitly considered before.
For example, though your primary interest might be personal transportation trends
in China, you are likely to have collected many references that cover other forms of
transportation elsewhere in the world. Should you later want to find all references
that discuss air transportation in North America and Europe, it would be easy to
create a new smart collection based on a new collection of your existing tags, as will
be described below.
Finding References Using Tags
The reason one goes to the effort of carefully configuring a QuickTag hierarchy and
then tagging each reference is to be able to use this information to find references
quickly and easily in the future. Sente provides several useful tools for searching
within your library using QuickTags. (There is a fuller description of each of these
methods elsewhere in this manual; this is just a brief introduction to searching
based on QuickTags.)
•Built-In, QuickTag-Based Smart Collections. As you modify the QuickTag
palette, Sente automatically maintains a parallel set of built-in smart collections
in the source list. This makes it easy to select all references tagged (either
explicitly or implicitly) with any of your tags.
The built-in smart collections based on QuickTag hierarchy.
•Custom smart collections. When you want to locate references with a
combination of tags, you can create a custom smart collection. Each such
collection can include tests on tags as well as other criteria (e.g., “all references
tagged ‘Bicycles’ and ‘China’ published before 1970”).
A custom smart collection deﬁnition based on tags.
•The Library Browser. One can do quick, ad hoc searches based on QuickTags
using Sente’s library browser (cmd-B). The library browser will provide feedback
as you refine your search by showing only those tags that are actually present in
the remaining subset of the data.
Using tags in the library browser.
Cha p t er 12
Find, Browse, and Hotwords
Sente includes several tools for locating and viewing references contained in your
•Find, for searching by free text;
•the Library Browser, for locating references based on their attributes, like
author name, journal name, publication year, etc; and,
•Hotwords, for highlighting words and phrases you select to help you quickly
scan references visually.
Each of these is explained in detail below.
The Find command lets you perform free-text searches on your library.
By default, Find will search anywhere within a reference for matches. Alternatively,
it can be told to look only in specific fields. This lets you locate references that
contain the string “white”, but only in an author name, and not in abstracts or titles.
When you perform a search using the Find command, you can save the search as a
smart collection so you can repeat the search at any time in the future by simply
clicking on the saved collection.
How to Use Find
If you have the toolbar visible, and your toolbar configuration includes the Find
field, you can initiate a Find search by simply clicking into that field, typing your
search string, and then pressing Enter.
The Find and Browse items in the toolbar.
You can also initiate a Find by executing the View > Find command in the main
menu, or by pressing the keyboard shortcut for this command, cmd-F. In this case,
if the FInd toolbar item is visible, keyboard focus will be transferred to that field.
Otherwise, the Find tool will appear above the reference list, as shown below.
The Find ﬁeld that is used when the toolbar item is not visible.
In either case, you can enter the search string and press Enter to initiate the search.
As long as a Find operation is active (with the Find panel still open) and a Reference
Editor is open (such as Bibliography Fields or Summary), all occurrences of the
search string will be highlighted in the Reference Editor to help you spot the text
that matched your search.
Documents you may be interested
Documents you may be interested