Topic Maps techniques in AToM

Topic Maps is the name of a technique for describing a specific domain of interest. It facilitates knowledge extraction from a set of information. The basic model of Topic Maps elements is:

• Topic - Czech Topic, alias theme, topik,
• Association - English Relationship, or association between topics,
• Occurrence, i.e. relevant information to the topic,

abbreviated TAO [1].

TAO - basic elements of TM

1. theme - relationship - occurrence.

The resulting Topic maps model distinguishes two layers, both based on the basic TAO model:

• knowledge layer - describes the elements of the information layer and their relationships, creates an index
• information layer - specific information in any format, the content itself

In the AToM environment, TM elements are implemented using Classes, Associations and Attributes [2], which represent the design of the knowledge layer. These elements describe the types of groups to be stored, the types of relationships and the types of information. They describe the types of domain specific information of interest, their types and relationships. The elements form data containers that enable the storage of domain of interest data in the expected form.

TM and TAO layers in the AToM system

2. knowledge layer in AToM system
3. information layer in the AToM system

These data containers store specific information about the monitored issue, which form the so-called information layer [3]. It forms the actual content of the ontology of the domain of interest. They are instances of elements from the knowledge layer.

TAO in the AToM system:

T: Topic = Topic = Class

A: Association = Relationship = Association

O: Occurrence = Occurrence = Class Attribute

Example of information processing using Topic Maps technique

You are communicating with someone and they start talking about "something" that was seen yesterday. They begin to describe a situation, a particular scene or a whole story, what it was about, or what emotions it possibly evoked, or what lessons were learned from it. All of this information describes the "something" that could be seen yesterday. That "something" is what we want to know, and what we call "Topic".

If you are really interested in the topic, but it is not yet clear what it was to see, you can ask questions. When was it, where was it on? Who was in it? Who filmed it? Other information may give a better idea of what it might have been about. Like the director. You may know something similar from him.

Actors, directors and other people around the film may be other "objects of interest" that we call "Topic". And interconnections like "who directed what", "who was in what", etc., or relationships, we call "Associations".

In the course of the conversation we will learn a lot of information, probably even the name of the film, and the "something" that was seen will be determined, or identified. Next time it may be what is referred to in the sense of: knowing something similar...

All the other information that came out of the conversation that characterizes that "something" is called "Occurrences". Some of the information is more essential to the identification, some is less essential. When someone adds: "I'll send you a link to a website", it is likely that this "occurrence" will be essential, compared to the other information, for identifying what the "something" actually is.

Knowing information in relationships and contexts is advantageous in navigating and searching through the heap of such information [1]. It enables filtering of information. In determining "something" by signs, manifestations, contexts, i.e. by those "occurrences". Similarly, as the poet James Whitcomb Riley puts it in his duck test, "If I see a bird that walks like a duck, swims like a duck, and quacks like a duck, I say of that bird that it is a duck."

TM application for information

1. heap of general information - concepts
2. categorization of information in the AToM system - TAO information and knowledge layers TM

The example lists several Topics (two films, some people, one of them a director, some actors), then associations (he directed this and the other one) and occurrences (I'll send a link, great film, or vice versa :). At this point we can make the following evaluation and determine what belongs to the information layer and what belongs to the knowledge layer [2].

Two movies (we already call it movies:) form a group of the same type (Topic - something to see), video. The name will be "video".

Similarly, the actors and the director also show similarity in the sense that they are people from the movie, and form a group of the same type of information named "person". A group of the same type is implemented in the system as Class.

The fact that this film was made by this director is a relation whose type, i.e., the making of a film by someone, determines a group of the same type of relations, which is named director. The fact that there are actors in the film is no surprise to anyone, and that is another type of relationship. The relationship between the group "video" and the group "person". This type of relationship is implemented in the system as Association.

The type of information that is heard, such as a description or where there is some info about it, is the type of occurrence of the information. The occurrence type "web link" may contain a specific web address to a site describing a specific movie. There may be more than one such web address. The type of this occurrence is implemented in the system as an Attribute. Attribute Class, Property Group. As a database entry into which the specific value is then written.

This is where the AToM system has its place.