Text Graphs

Text graphs provide a simple and easy-to-learn diagrammatic notation to represent complex conceptual content. Text graphs can be used for concept mapping and argumentation analysis. The advantages of text graph over previous similar notations is well-defined meaning and accuracy of representation.

A text graph about astronomical tides. Created with TGE (Text Graph Editor).

Motivation and Context

Suitable external representations can greatly improve ones performance in complex cognitive tasks by providing the benefits of distributed cognition:

1. Complex structures need not be kept in ones short-term memory but are stored in an external media for later retrieval and modification. The cognitive capacity can be used for more advanced activities.
2. The structures become open to reflection and can be analyzed at a meta-level to reveal the overall structure, the nature of connections, and the borders of the topic, as well as potential gaps in the understanding. If the understanding improves as a result of reflection, the externalized representation can be modified accordingly.
3. External representations can greatly improve the communication between a group of people. They can be indispensable in developing a shared understanding about the subject matter.

A prime example of the power of external representation is written text. The inventions of writing systems and printing had enormous consequences for the development of societes. Nowadays illiteracy - the inability to read and write - is considered a major impediment for efficient behavior and social or economic development. The question is not about the importance of language as such. Even the illiterate people usually master a language; they can speak and listen. The problems of illiteracy are precisely in the inability to use an external representation for the language.

Text can function as a carrier of knowledge. There are, of course, all kinds of theoretical problems related to ambiguous or vague terms or alternative interpretations due to different cultural background, that may lead us to suspect whether text can really transfer knowledge from one person to another. However, the widespread use of books as learning material or magazines as source of understanding of current affairs is a practical testimony about the possiblity to use text as carried of knowledge.

However, text is opaque: it takes work to find out what is the knowledge contained in a piece of text. Could there be imporved notations that could bring out the important building blocks of the content and their relationships?

Possible candidates are all the diagrammatic representations that can be used to aid generic cognitive tasks such as learning, organization of knowledge, and development of conceptual understanding. Examples are mind maps and concept maps.

Written text, mind maps or concept maps have different properties. Written text can be as detailed and accurate as wished to be while maps more easliy reveal the essential structures described in text. The drawbacks of maps, when compared to written text, are limited representational capabilities and more ambiguous or vague meaning. Text can be a stand-alone carrier of knowledge but a mind map or concept map cannot transfer knowledge without a supporting (verbal or written) explanation of the content. That decreases the value of maps in modeling complex domains and in developing shared models among a group of people.

Text graphs are a new kind of generic diagrammatic notation that can act as a bridge between text and graph-like notations. Text graphs can automatically be translated to text (so-called text expansion of a graph) and consequently a their meaning is well-defined. Text graphs also provide a simple but comprehensible notation for concept mapping. Above is an example text graph about astronomical tides together with the text expansion.

Meaning through text expansion

Below is a text graph of the Peano's postulates that specify the properties of natural numbers:

The text expansion of the graph is:

"(1) Zero is a Natural number
 (2) Every Natural number has a Successor which is a Natural number
 (3) Zero is not the Successor of any Natural number
 (4) Different Natural numbers have different Successors
 (5) If Zero has some Property and
     whenever a Natural number has that Property then its Successor also does
     then all Natural numbers have that Property."

The text expansion tells in natural language what assertions (propositions) the graph makes. It is thus possible to agree or disagree with the graph. Instead of being only an artistic expression of ones conceptual associations (such as mind maps), text graphs can actually state specific facts (right or wrong) and provide a basis for communication, argumentation, and the development of a shared conceptual understanding among people.

It should be stressed, however, that while the text expansion tells the propositional meaning of a text graph in natural language, it is usually not the reason the create a text graph. Its primary role is to provide a sanity check for the person who analyzes a system of concepts.

A text graph can reveal important meta-level structures hidden in text. The graph makes explicit the central concepts of the text and the relations between the concepts. For example, the text graph above shows clearly the three central concepts of Peano's postulates: Natual number, Zero, and Successor, as well as one meta-level concept Property. The number of links shows that the concept Natual number is central in the postulates and that the concept Property only appears in one of the postulates.

• Esko Nuutila and Seppo Törmä, Text Graphs: Accurate Concept Mapping with Well-Defined Meaning (PDF, PS), In Alberto J. Cañas, Joseph D. Novak, Fermin M. Gonzáles, editors, Concept Maps: Theory, Methodology, Technology. Proceedings of the First International Conference on Concept Mapping. Pamplona, Spain, Sep. 14-17, 2004. Volume 1, pages 477-485.
• Presentation slides from CMC2004.
• Esko Nuutila and Seppo Törmä, Introduction to Text Graphs (PDF, PS), in Kolin Kolistelut -- Koli Calling 2003, Proceedings of the Third Finnish/Baltic Sea Conference on Computer Science Education. Oct. 3-5, 2003 in Koli, Finland. University of Helsinki Department of Computer Science, Report B-2003-3, Helsinki 2003. ISBN 952-10-1439-3.
• Presentation slides from Koli Calling 2003.

We have implemented two tools for editing text graphs. The first program is called Conceptual Knowledge Organizing Tool. It was used for creating the pictures in the Koli 2003 conference article and presentation slides. After that, we have implemented a more advanced tool called TGE. The pictures in the CMC2004 conference article and presentation slides were created with TGE.

• Esko Nuutila
• Seppo Törmä