Chapt 14: Third-Generation Distributed Hypermeida Systems
14.4 NoteCards System and Halasz Seven Issues
NoteCards
-
Developed by a team at Xerox PARC
-
To support the task of transforming a chaotic collection
of unrelated thoughts into an integrated, orderly interpretation of ideas
and their interconnections
-
Designed for use by individuals or small work groups
-
A general purpose hypermedia system, but originally designed
to be used as a tool for idea processing and authoring in a research environment
NoteCards System Description
-
A general hypermedia environment that is fairly typical of
the generation of workstation-based systems.
-
The system provides the user with a network of electronic
NoteCards interconnected by typed links.
-
This network serves as a medium in which the user can represent
collections of related ideas.
-
The network also functions as structure for organizing, storing,
and retrieving information.
-
The system includes facilities for displaying, modifying,
manipulating, and navigating through the network.
How Does NoteCards Look Like?
-
Implemented within the Xerox Lisp programming environment
-
Two primitive constructs: notecards and links
-
Notecards
-
Electronic generalization of the 3X5 paper notecard
-
Links
-
Typed, directional connection between a source card and a
destination card
-
Two types of cards: browsers and filebox
-
Browsers
-
Notecards that contain a structural diagram of a network
of notecards
-
Fileboxes
-
Specialized cards that can be used to organize or categorize
large collections of notecards.
Halasz Seven Issues
Halasz examined the major weaknesses in the design of
NoteCards
Issue 1: Integration of Search and Query Functionality
-
Navigational access by itself is not sufficient
-
Effective access to information stored in a hypermedia network
requires query-based access to complement navigation
-
Content search: all nodes and links of the networks are considered
as independent entities and are examined individually for a match to the
given query. Ignores the structures of a hypermedia network
-
e.g. all the nodes containing the string "hyper*" would be
a content query
-
Structure search: Specifically examines the hypermedia structure
for subnetworks that match a given pattern.
-
e.g. "all subnetworks containing two nodes connected by a
"supports" link, where the destination node contains the word "hypertext."
This query contains a description of node content (i.e., contains the word
"hypertext"). But it also contains a structural description of a network
(i.e., two nodes connected by a "supports" link.
Issue 2: Composite Node Types
-
Need container or collection nodes in addition to content
nodes
-
NoteCards only have two primitive constructs: cards and links
-
Fileboxes and browsers are built up from those constructs
-
Lack of a composition mechanism, i.e., a way of representing
and dealing with groups of nodes and links as unique entities separate
from their entities.
Issue 3: Virtual Structures Over Node Collections
-
Hypermedia: essentially static and fragmentary nature of
the hypermedia model
-
Hypermedia systems have difficulty with rapidly changing
information
-
The network cannot reconfigure itself in response to changes
in the information it contains.
-
Lack of dynamic mechanisms limits the utility of hypermedia
in many domains
-
Nodes and links are extensionally defined: nodes and links
are defined by specifying the exact their components.
-
Virtual structures are defined intentionally: by specifying
a description of their components
-
The exact subcomponents of a virtual structure are determined
by a search procedure whenever the structure is accessed or instantiated
-
e.g. A virtual composite node might be defined by a specification
of the form: a subnetwork containing all nodes created by someone other
than me in the last three days. Each time this composite was accessed,
its structure and content would be recomputed
-
Notion of virtual structures for hypermedia is a direct adaption
of the concept of views (a.k.a. virtual tables) in the world of relational
database systems
-
Notion of virtual structures in hypermedia would be possible
only in a system that support a substantial search/query mechanism over
the hypermedia network
-
A virtual composite allows the user to create nodes that
are dynamically constructed at access time form other nodes, links, and
composites that are stored in the network
Issue 4: Computation over Hypermedia Networks
-
Hypermedia systems are generally passive storage and retrieval
system.
-
Hypermedia systems do not actively direct the creation or
modification of the network or the information contained therein
-
e.g. Unlike expert systems, hypermedia systems don't include
inference engines that actively derive new information and enter it into
the network
-
e.g. Integration of hypermedia and AI technology is an interesting
direction to explore.
-
Hypermedia systems -> heavyweight entities (whole documents),
inter-entity references (links)
-
Object-based systems -> define type (class) hierarchy for
entities, operations (methods) that can be performed on instances of each
type.
-
Frame-based systems -> inheritance, defaulting of slot values,
integration of truth maintenance, inference engine, rule-based reasoning
with frame representations
Issue 5: Versioning of Nodes and Subgraphs
-
A good versioning mechanism will allow users to maintain
and manipulate a history of changes to their network.
-
Lack of versioning has a lesser impact in the authoring,
argumentation, etc.
-
Two levels of versioning
-
Individual entities: nodes, links, composites
-
Each entity has its own version history.
-
e.g. composites have a problem of propagating version changes
from subcomponents to their composites
-
Entire hypermedia system
Issue 6: Support for Collaborative Work
-
Hypermedia is a natural medium for supporting collaborative
work. Creating annotations, maintaining multiple organizations of a single
set of materials, and transferring messages between asynchronous users
are the kind of activities that form the basis of any collaboration
-
NoteCards was originally designed as a single user system.
-
Most idea processing and information management tasks are
inherently collaborative, with groups of two ten people working on a common
project
-
Two important mechanisms to improve support for collaborative
work
-
Mechanism of simultaneous multiuser access to a common network
-> sophisticated , shared database, concurrency control
-
The social interactions involved in collaboratively using
a shared network
-> notify what the other people are doing, update a node,
etc.
-> "mutual intelligibility"-- each participants must
have some degree of understanding of the actions and intentions of any
collaborators
-> "procedural activities" -- decide how to distinguish
additions and deletions as the draft is passed back and forth
Issue 7: Extensibility and Tailorbility
-
Hypermedia model provides a set of very general abstractions:
nodes, links, composites
-
Users have to apply these abstractions to their tasks.
-
Generic future of hypermedia has a blessing and a curse
-
Blessing: allow hypermedia to be useful in a variety of task
domains
-
Curse: generic hypermedia is not particularly well suited
to any specific task or style of use
-
Programmer's interface is needed to extend and tailor the
system.
-
Goal: easy for non-programming users
-
Good example: GNU Emacs, Apple HyperCard/HyperTalk
-
Built around an interpreter for a fully-functional programming
language that is specially designed for the kind of objects and operations
by the systems