Specific Instructional Objectives are that students are able to interpret knowledge representations in the form of Semantic Networks and that students are able to interpret knowledge representations in the form of Frames.
1. Semantic Networks
Semantic (associative) networks are a form of knowledge base representation in diagram form.
The diagram consists of nodes and arcs. A node represents a concept, while an arc represents a relationship. For example, consider the nodes below:
Figure 5.1 - Examples of Semantic Networks
The example above is a semantic network that illustrates a family relationship. If it is known that Rudi is currently 12 years old, Joko is 40 years old, Andri is 64 years old and Ben is 66 years old. How old is Leni now?
The diagram above can be converted into the form of predicate calculus as follows: father (joko, rudi) father (andri, joko) brother (ben, andri) mother (rini, rudi) mother (susi, rini) brother (yulia, susi) mother (yulia, leni).
Semantic Networks can also be used to describe a single relationship of several concepts. Consider the example in Figure 5.2. The figure represents the sibling relationship between Jim, Joko and Joni.
Figure 5.2 - Semantic Networks depict sibling relationships
The relationship between nodes that is not binary can be represented by changing the relationship into an object. Semantic networks with this form are called reification. Consider the example in the Figure below: The semantic networks can be represented in the form of predicate calculus consisting of 3 things, namely: owner, borrower and object becomes: borrow(maria, jim, car).
Figure 5.3 - Semantic Networks with reification
In most semantic networks there is a common difficulty, namely how to distinguish between individuals and classes? To illustrate this, nodes can be represented in two ways: nodes as individuals and nodes as classes. Consider the example below:
In the image, it can be seen that the animal node and the birds node are classes, while the wing node is an individual.
Knowledge representation using Semantic Networks still has several weaknesses, including:
Figure 5.4 - Semantic Networks with subclasses
This allows for different interpretations of semantic networks which will lead to errors in the conclusion-drawing process.
The relationship connecting nodes cannot contain all the information, for example in Figure 5.1, the father or mother relationship does not describe whether the relationship is a sub-class or a member.
2. Frame
Note: In the 70s and 80s, semantic networks changed form to become a frame representation model.
A frame has a set of slots. For example, consider the family structure diagram below:
Figure 5.5 - Frame of the Adam family
Figure 5.5 can be represented in frame form as follows:
Frame Adam:
sex: Laki-laki teman-hidup:
Ana anak: (Jeremy Jordan Ellen)In the frame, the Adam frame has 3 slots, namely, sex, life partner and child. While the child slot has three values, namely: Jeremy, Jordan and Ellen. As seen in the picture, we can also create other frames which are sub-classes of the Adam Frame. There are at least 7 sub-class frames that describe each individual, namely: Adam, Ana, Jeremy, Jordan, Ellen, Male, and Female.
The Adam frame can be converted into predicate calculus form as:
sex(Adam,Laki-laki)
teman-hidup(Adam,Ana)
anak(Adam,Jeremy) anak(Adam,Jordan) anak(Adam,Ellen)A slot is a relationship to another frame or to a value.
A slot has one or more facets. Consider the example of the Animal frame in Figure 5.6 which illustrates the relationship between slots to form a frame, and each slot has several facets.
Facets have values associated with each slot. Values are not limited to a facet, in some cases a slot in a frame may have multiple values.
An example of an Animal Frame formed by slots and facets is shown in Figure 5.6.
Figure 5.6 - Structure of the Animal Frame
Frames can also be used to represent an action and its consequences. To illustrate this, two frames are used, namely the Action Frame and the State-Change Frame.
For example, consider the example frame representing an action in Figure 5.8. In sentence form, this frame means: "Getting a prize of 500 thousand makes Toni happy."
Figure 5.7 - Example of an Action Frame
As seen in the picture, the state-change frame is the result of the action of the first frame. Primitive Slots can consist of one of 15 types of primitives, namely:
In more detail, the Action Frame can also be divided into several sub-actions as shown in Figure 5.7.
As seen in the image, the Action Frame has 3 sub-action frames, each of which explains the action details of the action frame.
Figure 5.8 - Example of a Sub-Action Frame
From the examples above, knowledge representation with Action Frame must comply with the following rules:
- An Action Frame must have a primitive slot by taking one of the 15 primitives that have been defined.
- defined.
- A State-Change Frame contains slot objects that contain values according to the application's needs.
- An Action Frame can be connected to one or more other action frames via subaction slots.
- Action Frames and State-Change Frames can be connected using result slots.
- The other slots and their values depend on the application requirements.
EXERCISE
Given a set of facts as follows:
- father (Suryo, Arman)
- mother (Susi, Lusi)
- wife (Sari, Suryo)
- husband (Joko, Susi)
- wife (Maria, Arman)
- children (Doni, Arman)
- father (Arman, Haris)
- children (Ari, Susi) children (Susi, Suryo)
Based on the facts above, answer the questions below:
- Draw a semantic networks diagram of the family tree!
- What is the content of the variable X for the expression: paman(X,Ari)?
- What is the content of the variable X for the expression: ibu(X,Susi)?
- What is the content of the Y variable for the expression: siblings(Lusi,Y)?
- What is the content of the Y variable for the expression: kakak(Arman,Y)?
- What is the content of the variable X for the expression: nenek(X,Doni)?
- What is the content of the Y variable for the expression: anak-anak(Sari,Y)?