[Gzz-commits] journals/plans SA_2003.rst

[Tuomas J. Lukka]

CVSROOT:        /cvsroot/fenfire
Module name:    journals
Changes by:     Tuomas J. Lukka <address@hidden>        03/05/14 04:18:47

Modified files:
        plans          : SA_2003.rst 

Log message:
        Train trip - lots

CVSWeb URLs:
http://savannah.gnu.org/cgi-bin/viewcvs/fenfire/journals/plans/SA_2003.rst.diff?tr1=1.111&tr2=1.112&r1=text&r2=text

Patches:
Index: journals/plans/SA_2003.rst
diff -u journals/plans/SA_2003.rst:1.111 journals/plans/SA_2003.rst:1.112
--- journals/plans/SA_2003.rst:1.111    Wed May 14 04:04:40 2003
+++ journals/plans/SA_2003.rst  Wed May 14 04:18:47 2003
@@ -78,46 +78,77 @@
 
 TJL
 
-The Fenfire project is a free software effort to build 
-a location transparent, hyperstructured desktop 
+The Fenfire project is a free software research effort to build 
+a location transparent, hyperstructured computer operating 
 environment. 
-The work leading to the current Fenfire project (formerly known as GZigZag and 
Gzz) 
-was started in 1999. 
-
-Fenfire consists of three components:
+On an abstract level, the Fenfire differs from conventional desktop 
environments
+by bringing together
+three components:
 
 Hyperstructured data model
-    ... RDF + Xu 
-
-Novel focus+context user interfaces
-    The bidirectionally linked hyperstructural approach to information opens 
new 
-    possibilities to user interfaces. 
+    As opposed to the conventional file system trees and location-based 
pointing
+    within them, the data model in Fenfire is more database-like, allowing 
+    anything to be connected permanently to anything else.
+    The data model consists of two orthogonal pieces,.
+    The Xanalogical media model[ted-xanalogical-structure-needed]_ 
+    endows to the smallest units of media with identity and as a consequence
+    changes
+    the cutting and pasting operation from copying pieces of media to actually
+    creating a structure between the instance locations since the identities 
can be tracked.
+    On a higher level, RDF[w3rdfurl]_ graphs are used to create dynamic 
structure
+    between instances of xanalogical media.
+    In this data model, all connections are fundamentally bidirectional, 
traversable
+    in both directions.
     The user can connect any two items which
     will then be near each other in hyperspace, easily visible and
     accessible from one to the other.
 
-P2P information sharing 
-    To support data mobility, we have developed a software library, named as 
Storm 
-    (STORage Module), for storing and retrieving data as immutable byte 
sequences, which 
-    are identified by cryptographic content hashes. With Storm, we plan to use 
globally 
-    unique, semantic-free references for locating data in Peer-to-Peer 
environment. 
-    Location-independent identifiers allows a peer-to-peer lookup system find 
documents 
-    wherever they are moved. We believe that the properties of Peer-to-Peer 
technology, 
-    such as the ad hoc connectivity and the lack of central authority, are 
prerequisites 
-    while mobilizing our system. Additionally, Storm allows us to study two 
important 
-    issues *of data mobility*: dangling links and version management. 
-    We study how location-independent identifiers can 
-    support *data mobility*. For instance, documents are often transferred 
quite freely 
-    between computers: they are sent as e-mail attachments, carried around on 
disks and 
-    published on the web. We use 'data mobility' as a collective 
-    term for the movement of documents between computers (or folders!),
-    and movement of content between documents (through copy&paste).
-
-
-
-
+Novel focus+context user interfaces
+    The bidirectionally linked hyperstructural approach to information opens 
new 
+    possibilities to user interfaces, since each piece of information has 
+    a richer context available to the computer than in conventional 
filesystems. 
+    The Fenfire project includes libvob, a library for creating flexible, 
animated
+    user interfaces easily, allowing nonlinear fisheye coordinate 
systems[carpendale01presspace]
+    and smooth text rotation and scaling.
+    The user interfaces currently used in fenfire are mostly 
+    focus+context[fc-fisheye]_, showing one location in the hyperstructured 
database
+    as the central item and peripherally the connected locations.
+    Several of the new technologies in fenfire are related to user interfaces, 
+    in particular to showing the context visibly yet nonintrusively.
+
+Permanent data and peer-to-peer (P2P) information sharing 
+    Collaboration with even only 
+    conventional, rigidly structured data leads to versioning problems
+    unless a version control system such as CVS is used.
+    In Fenfire, we have developed a group of technologies collectively known
+    as Storm (STORage Module) to address this problem. 
+    Storm starts from a fairly unconventional idea: 
+    instead of mutable files or databases, all information is stored
+    in immutable byte sequences, blocks[lukka02guids].
+    Blocks are accessed through their permanent globally unique identifiers 
(GUIDs).
+    To allow mutable data, a system of pointer blocks is built on top of the
+    block system.
+    This already helps versioning: unless explicitly requested, old versions
+    are never deleted; new and old and alternate versions are accessible on a 
computer
+    without contacting a central server.
+    Location-independent identifiers allow a P2P lookup system find documents 
+    from any source, as opposed to URLs which specify the explicit location for
+    fetching a document.
+
+There is a strong interaction between these components: for example,
+the data model *enables* the novel user interfaces, but on the other hand,
+without new types of user interfaces, the richer data model would be of 
limited use.
 
+The work leading to the current Fenfire project (formerly known as GZigZag and 
Gzz) 
+was started in 1999 in collaboration with Ted Nelson, who has since 
discontinued
+the collaboration and due to patent and trademark reasons, the Fenfire project
+cannot currently use the name GZigZag nor the ZigZag structure. 
+However, the fundamental ideas embodied in the ZigZag structure were not 
affected
+by the particular implementation and in spring 2003 we were able to convert
+our system to using RDF.
 
+Currently, Fenfire is nearing the public release of its first subgoal: the 
FenPDF
+system for reading, annotating and linking electronic publications.
 
 .. raw:: latex
 
@@ -144,51 +175,66 @@
 
 
 
+Previous activity and publications
+----------------------------------
+
 .. previous activity pertaining to the scope of the topic of the
    research (list of the research team's five most important
    publications in terms of the topic)
 
-In [lukka02guids]_ we discuss the possibility of implementing  
-the xanalogical storage model in Peer-to-Peer environment with globally unique 
identifiers.
-
-In [lukka02fillets]_ we introduced the use of fillets --- 
-smoothing of sharp angles at the joints ---
-for rendering on-screen connections in a way that makes them easier to
-perceive even with several overlapping objects.  
-
-We have developed a perceptually designed hardware-accelerated
-algorithm for generating unique background textures 
-[kujala03paper]_ for distinguishing documents.
-
-Additionally, the methods developed and used in 
[lukka02gomcmc-andalso-kujala02solutions]_ 
-are relevant to the modeling problems in our user interface research.
-
-List of the most important publications
+The most important publications related to this project are:
 
     - Tuomas J. Lukka and Benja Fallenstein: 
       Freenet-like GUIDs for implementing xanalogical hypertext. 
-      *Proc. Hypertext 2002* 194-195, 2002
+      *Proc. Hypertext 2002* 194-195, 2002.
+      This article discusses the possibility of implementing  
+      the xanalogical storage model in Peer-to-Peer environment 
+      using globally unique identifiers.
+
     - Tuomas J. Lukka, Janne V. Kujala, and Marketta Niemelä,
       Fillets: Cues for Connections in Focus+Context Views of Graph-like 
       Diagrams. Proceedings of the 
       *6th International Conference on Information Visualisation*,
       pp. 557--562, London, July 2002.
+      This article introduces the use of fillets --- 
+      smoothing of sharp angles at the joints ---
+      for rendering on-screen connections in a way that makes them easier to
+      perceive even with several overlapping objects.  
+
     - Janne V. Kujala and Tuomas J. Lukka: 
       Rendering recognizably unique textures.
       To be published in the proceedings of the 
       *7th International Conference on Information Visualisation*,
       London, July 2003.
+      This article introduces
+      a perceptually designed hardware-accelerated
+      algorithm for generating unique background textures 
+      for distinguishing documents.
+
     - Tuomas J. Lukka and Janne V. Kujala:
       Using Genetic Operators to Speed up Markov Chain Monte Carlo 
       Integration. 
-      *Monte Carlo Methods \& Applications*, 8(1), 51--71, 2002
+      *Monte Carlo Methods \& Applications*, 8(1), 51--71, 2002.
+
+
     - Janne V. Kujala and Tuomas J. Lukka:
       Solutions for certain number-conserving deterministic cellular automata.
       *Physical Review E* 65, 026115, 2002.
+      The simple cellular automaton
+      rules introduced in [JVKXXXFUCHSBOCCARAREF] produced an unexpected 
result in simulations.
+      We were able to solve the problem exactly starting from first principles,
+      by using generating functions in a rather twisted way.
+
 
 Of these, the two last ones contain some mathematical methods which we hope 
will 
 be applicable to e.g. filtering in the text rendering problem.
 
+Additionally, the methods developed and used 
+are relevant to the modeling problems in our user interface research.
+
+In addition, several manuscripts about various aspects of Fenfire are in 
preparation.
+
+
 - postgraduate studies performed in the research team
 
   Janne Kujala
@@ -200,11 +246,43 @@
 
 TJL
 
-The current proposal 
+The proposed research consists of two parts: continuing research into questions
+opened by techniques invented for Fenfire, and theoretical questions whose
+solutions are relevant to developing algorithms used in Fenfire.
+
+The questions opened by the inventions in Fenfire look at some phenomena in 
new light:
+for example, we do not believe the questions in [kujala03paper]_ on how
+to generate recognizably memorizable textures were asked before this; so far,
+most research has concentrated on texture *discrimination*, i.e. under which
+conditions a human notices that two textures placed right next to each other 
are different.
+Contrasting results obtained in these two different questions may lead to 
surprising
+new insights on the human visual system and memory.  
+This basic research may of course yield new technologies but that is not the 
point;
+the point is to understand and measure the phenomena we have observed in 
constructing Fenfire.
+
+On the other hand, some of the research questions in this proposal are 
+driven by technological requirements. For example, 
+enhancing the readability of the hardware-accelerated rendering of 
+rotated or nonlinearly transformed text is a problem
+whose solution would benefit Fenfire. 
+Here, however, before being able to tackle that problem, a considerable
+amount of mathematical framework is required, related to filtering and 
perception.
+Creating this framework is a complex task, requiring both mathematics
+and user experiments.
+Armed with the framework developed by this research, developing the actual
+hardware-accelerated filtering algorithms becomes a relatively easy task.
+
+All subproblems presented in this proposal are, however, significant in their 
own right:
+even if the Fenfire project ceased to exist, the research results would be 
applicable
+to other programs and problems.
+
 
 Objectives and methods
 ======================
 
+(reorg: "general (user experiments, math. modeling)", then specific projects,
+ stating for each the objectives *and* methods.)
+
 Objectives of the research
 --------------------------
 
@@ -338,6 +416,8 @@
 and smooth animation. 
 Generalizing filleting to three dimensions ...
 
+    - quantitative experimental results from general graphs layout
+
 Text filtering
 ''''''''''''''
 
@@ -378,34 +458,50 @@
 
 TJL
 
-The hardware algorithm for rendering the unique background textures
-currently uses ad hoc distributions in the sampling of the texture
-features. Although the distributions (e.g., for choosing the texture 
+In [kujala03paper]_, we introduced the use of unique background textures
+for distinguishing documents.
+The current hardware-accelerated algorithm for rendering the unique background 
textures
+uses *ad hoc* distributions in the sampling of the texture
+features. 
+Although the distributions (e.g., for choosing the texture 
 colors)  are qualitatively based on 
 general principles for making the texture distinguishable and 
 recognizable, their exact values are not rigorously justified.
-We plan to obtain justification for different aspect of the algorithm
-experimentally and to further evaluate the recognizability of the generated
-textures.
+
+The goals in this area are first to 
+formulate and test (with user experiments) a general way of adjusting a 
distribution
+towards being memorable. This is a nontrivial optimization problem,
+as the recognizability of a texture depends on the distribution it is
+presented as a part of.
+Next, it is important to research how much different aspects of the textures
+may be optimized separately by the methods developed in the previous step, 
i.e., how much
+the different aspects interact. 
+These questions are interesting not only from a practical point of view but 
also
+from the point of view of cognitive psychology: for instance, 
+if any data can be obtained on how fast and how the brain adapts 
+to a particular input distribution, it would be an interesting result.
+On a practical level, these experiments would allow us to get rid of the *ad 
hoc*
+distributions and create a well-founded
+algorithm for generating unique textures.
+
+    - quantitative model of certain aspects of texture perception and 
recognition 
 
 Controllers
 '''''''''''
 
 TJL
 
-Custom controllers can offer new ways to navigate and control the information 
-flow on the screen. 
-They are many usability aspects in the configuration of the degrees of freedom
-provided by the controller that can be theoretically modeled and experimentally
-tested.
-
-We have invented a new type of controller which we cannot describe here
-before the (potential) filing of a patent application.
+In the context of Fenfire,
+we have invented a new type of computer controller which we cannot describe 
here
+before the (potential) filing of a patent application. 
+This controller, however, also opens some interesting related questions for
+basic research.
 
 Storage
 '''''''        
 
-TJL
+JVK, HH
+!!! [tjl] Mikä tässä on KYSYMYS?
 
 Research on how location-independent identifiers can be used to find
 information in a Peer-to-Peer environment. To support our research work, we 
have 
@@ -652,23 +748,15 @@
 
 .. research results to be expected and their significance
 
-The Fenfire project is largely based on novel ideas on both
-user interfaces and hyperstructure. 
-Therefore, the project includes basic research as well as
-applying the developed technologies on 
-the Fenfire distributed hypermedia system.
-  
-The expected results of the research are listed below:
-  
-- paper: procedural background textures for
-  bestowing identity to virtual objects
+As already discussed, this project is a combination of several
+subprojects revolving around specific questions raised by our work in Fenfire.
+The expected results of each subproject are listed in the subsection above
+describing that subsection.
+The results will be published in international conferences and journals.
 
-    - quantitative model of certain aspects of texture perception and 
recognition 
+The expected results of the research are listed below:
 
-- fillets: drawing connections between things in a perceptually
-  motivated way resulting far quicker perception
  
-    - quantitative experimental results from general graphs layout
  
 - experimental justification for the layout geometry and type of animation 
   (buoys, nadir, irregu, etc.)
@@ -676,26 +764,6 @@
 .. - buoys: show connected documents and ideas floating in the margin          
  
 .. - nadir: a natural layout for multiple pieces of data by small rotations
 
-
-- storm 
-
-    - saving data in permanent blocks, requiring
-      less bandwidth, allowing access to earlier
-      versions.    
-    - data uniquess and immutability
-    - distributed storage framework
-
-- urn-5
-
-    - registered URI scheme to our project
-
-- Object/document identity in user interfaces
-- P2P networking
-- Interactive creation of hyperstructure
-- Bidirectional linking in hyperspace
-- Xanalogical hypertext model
-- Relational data/data relationship management, document/data item management
-  
 .. Used techologies, our implementations:
 
        - RDF structure