Usecases

Beginning with December 2010 we are in contact with the developers of freeplane. Wouldn't it be nice to integrate some of the enhancements developed in the project http://jbusse.de/semAuthTool/semantic_freemind_freeplane.html into the freeplane distribution?

The devolpers like to learn more about our prototypes and respective usecases. Here they are.

Work done 2008 - Web site export

What

As of today freeplane is able to html export a mm into a deeply nested list.

Already developed with semAuth: You can tag nodes being section headings, paragraphs, lists, links, arbitrary xml elements and even documents. The tool semAuth exports such a mindmap to a fully fledged linked web site, automatically generating the according navigation (and forthcoming: indexes).

Description:

Needless to say that this very html page and site is produced automatically from one single mm.

If the community request it we are willing to provide an xslt export which provides exactly this feature highly integrated to freeplane.

Who needs that?

Some commercial mindmap tools offer features which allow for slicing predefined mindmap structures into one html page. They generate headings, paragraphs etc. automatically based on the nesting depth of a node. The currect nested html list feature of freemind and freeplane is somewhat behind this state of the art.

To our knowledge there is no other tool around which produces complex xhtml websites from one mm.

Integrating the document structuring semAuth capabilities to freeplane would push this free tool to pole position w.r.t. website generation. As a result freeplane would be an end user capapble tool to maintain static html websites on their own behalf.

We offer to provide an according xslt stylesheet under GPL licence if the community would be interested.

Work done 2009 - Semantic export

What

As of today there is no semantic interpretation of a mm available at the market (be it open or commercial).

Within the project http://www.x-media-project.org/ we developed a mindmap markup language and several prototypes to generate an RDFS or F-Logic-Ontology from a freemind mindmap.

In 2010 J.Busse recoded mm2flo completely and published it on sourceforge under http://semauth.sourceforge.net/

semAuth 2010 differs from mm2flo in that document structuring markup and semantic markup are designed (nearly) orthogonally, such that they can be used (nearly) independendly from each other.

As a result freeplane + semAuth now is one of the early tools which allows for editing RDFa annotated web pages: c.f. the last two screenshots on the page

semauth has become the primary editing tool for J.Busse. Editing larger maps (the site jbusse.de/semAuthTool/ is generated from one mm with > 4500 nodes; jbusse.de/training has > 7000 nodes ) we recognized that one important feature was missing: We missed an auto completion within freeplane.

How

Find a discussion of a modelling example at http://www.jbusse.de/semAuthTool/pdkSandboxOntologies.html

Work done in Nov - Dec 2010 - Semantic auto completion

In autumn 2010 J.Busse set up and managed a programming project at the Fachhochschule Karlsruhe: A group of students volunteered to extend freeplane with semAuth in order to prototype the desired auto completion capabilities.

Auto completion

What: As a minimal step we would like to have an auto completion in freeplane. Sources for auto completion could be:

  • The mindmap we are currently editing
  • All mindmaps we have opened concurrently in a freeplane session
  • Additional 3rd party sources
    • other freemind or freeplane mm
    • concepts (i.e. URIs) which are defined in wikipedia / dbpedia

Background: Many people who use mindmaps for PIM have a mixture between

  • free prose text, i.e. words and sentences with a naive, direct, not overly well defined meaning
  • well defined terms, so called concepts
  • text structure
  • conceptual structure

Benfit of an auto completion: Pretty often we'd like to refer to concepts which are well defined somewhere else. An auto completion

  • helps us to avoid typos
  • enhances editing comfort
  • allows for glancing into existing related portions of text without having to browse away while editing the current node.

Contextually informed auto completion

What

As a next step we'd suggest to have an informed auto completion: only give back text of nodes

  • which have the same style / the same icon like the current node
  • which have the same parent node. This is a nice feature if you are developing taxonomies, part-whole descriptions, project outlines etc.

This informedness can be described purely in a syntactic manner, i.e. by referring to the structure of a mindmap (resp. it's XML serialization).

A purely syntactic context information as strong advantages:

  • Freeplane primiraly is an editor and not a knowledge browser. It is tree syntax which freeplane cares about.
  • freeplane is a lightweight tool. Syntactic auto completion can be performed fast and easy as long as we stick to a set of locally loaded mindmaps.

Background

A syntactic auto completion can restrict itself on all item which can be classified to be items with certain caharacteristics.

Say we want to restict the auto-completion to all persons we already know. Then we'd like to get back all object which are known to be persons.

Now we have some freedom of how to qualify a node (say "Isaac Asimov") syntactically being a person

  • We can attach a person-icon to it
  • We can define a certain node style for persons
  • We can interpret the parent node of the current node containing the qualifying predicate, thus restricting the auto completion to all children (or descendants?) of a node with the same text as the parent node. (This in fact would be a pretty powerful version of a syntactic auto completion.)

Semantically informed auto completion

While a syntactically based auto completion is really nice, it's shortcomings disclose when we have more than a handful categories at and.

Say that you already have classified "Исаак Озимов" being an author of a book, being a friend or a customer at what else. (Perhaps you have attached an author, a friend and/or a customer icon to this node in your map). And of course you already have a subtree in your mindmap which makes clear that authors, friends, customers etc. also are persons.

Say that you now want to refer to a person which name starts with "Исаак". What you'd like to have while typing in "Исаак" is an inferencing step which amends your auto completion with some logic behind the scene. In especially you want to get back "Исаак Озимов" even if this string nowhere else has been classified being a person!

We already have implemented a well defined and not too complex mm markup language which i.e. empowers end users to define and to relate categories (concepts, classes), instances, relationships and much more. Find more at:

While maintaining complex concept networks in the end is called ontology engineering , you don't need to be afraid of that: Many users often already think of (and in fact maintain) class hierarchies, part lists an much more in mindmaps. The semantic auto completion is intended to be usable by end users. As long as you don't want to go really far with it semantic it's pretty simple!

And yes: semauth implements this language.

Auto completion w.r.t. to a shared background repository

What

We already have suggested to rely on more sources than simply the current mm in order to search for (contextually informed) auto completion strings.

Then the question arises: Which knowledge structure is powerful and simple enough to serve as a common representation for shared knowledge about classes, instances, relationships etc?

Possible backend structures are:

  • syntactically
    • We could define a list of local or publicy available mm which could be used as sources for an informed auto completion
  • semantically informed
    • we can query dbpedia for instances of the class "Person"

How

Our answer: We belive that RDF and RDFS is a well suited and commonly available data model to represent conceptual background knowledge which a semantically informed freeplane auto completion could rest on.

The project http://jbusse.de/semAuthTool/semantic_freemind_freeplane.html implemented a freemind prototype which connects to a semantic backend like http://dbpedia.org/About in order to grasp semantically informed auto completions.

Who benefits?

Having an autocompletion w.r.t. to a semantic backend allows group of users to share and organize their knowledge in a common knowledge representation. Each user is able to maintain her own view on the big picture in a private mm, without having to share (or even to know) the full picture at any time.

Freeplane thus becomes a mindmap structured view onto a more (in fact arbitrary complex) knowledge base.

To our knowledge this is a genuine innovative usage of the mindmapping approach in the field of personal information management (PIM). The main fields of application are public and corporate knowledge management.