Research Results

Let us take the example of stated project.

We will start the analysis with the table that illustrates categorial weights of characteristics (all tables and graphs are built by the program on user request). We have chosen the four-category variant as the most convenient both from the point of view of information compression and prediction accuracy achieved.

The table demonstrates the presence of four basic independent categories of women's evaluation, common for men.

1. The first criterion (see Column G1 in Table 1) describes a family partner (categorial weight to G1 0.92 or 92%), as a physically fit person (0.69) who a man is ready to obey (acknowledge as leader) (0.62), and who is not expected to "mother" a man and does not demand mothering (-0.56). It is clear that the mentioned set of features in a category reflects mens attitudes to a probable spouse. It is surprising that young men of Russia dont associate a family partner with a sex one (0.03). As they say it in the city of Odessa, "a wife and a lover are two big differences".

2. The second criterion (G2) defines an organizing leader (probably, a popular person, a group star), having the following weights of primary characteristics: leader (0.71), and communicator (0.92), and also partially a caring protector (0.44).

3. The third criterion (G 3) actually defines a girl-friend (sex-friend): reliable friend (0.95) and sex-partner (0.78).

4. The fourth criterion  (G4) makes sexual partnership improbable (-0.55) and characterizes a doer, performer (0.95) and a caretaker (0.55). This can  be used in characterizing a close relative or a dependent reliable companion (a colleague under supervision and others). The opposite pole of a factor (with the opposite sign in Column G4) characterizes an independent (-0.95), physically fit (0.48) sexual partner (0.55), who has no responsibility whatsoever (-0.52). it actually reflects a purely sexual interest.

The categories above reflect basic subconscious attitudes of young men (Russians) who evaluated womens appearance.

It should be noted that a number of features converge in each class of objects (up to the point of synonimity, as in Russian green = unripe in the class of berries) to form independent generalizing categories.

For this set we have limited ourselves to the selection of three independent states or mens opinions.

Let us take for example the table of states related to Photo 3 subject in terms of primary features.

State 1 is the most important one, since it is realized in 44 % of perception cases (see last column of Table 2). It is characterized by the following estimates: family partner (0.70), sex-partner (0.49) and reliable friend (0.32).

The second state has an unstable realization in 20 % of cases. In this state an object is perceived as a sex-partner (0.74) and excluded as a family partner (-0.60).

An object in the third state is perceived as a reliable friend (0.50) and experienced communicator (0.61), not as a sex-partner (-0.45).

Generally speaking, an attitude aimed toward family relationship lowers sexual attraction of a woman, which is clearly seen from the growth of sexual interest in the second state compared to the first one (from 0.49 to 0.74), family attitudes decreasing (from 0.70 to 0.60).

Categorial representation of the most probable first state of all research subjects can be viewed in Table 3.

The program enables  a researcher to analyze categorial representation of any states selected for any object (See Table 4).

Table 3 shows that there are following ways of perception possible for Subject 3:

First type of perception: a possible family partner (051) and a sex-friend (0.48).

The second type of perception: sex-friend (0.53), family attitudes excluded (-0.56).

The third one: organizing leader (0.44).

One can make sure that this representation is well consistent with what we have stated above, yet it is described through young peoples perceptional stereotypes.

Let us turn to motivational vector analysis. Apparently, any requirement has relevant means for its optimal realization. Means of satisfying requirements for a studied mentality are, in our case, research subjects. Clearly enough, subject rating (desirability) will under given conditions run the more, the closer is its evaluation vector to the best (ideal) means of need-satisfaction, and the longer this vector is. Angular coordinates that determine the estimation vector direction in a semantic space are in one-to-one correlation with the subjects features, while its length corresponds to rigidity (stability of these features, subject image and its quality). The ideal means of satisfying a requirement is determined by a certain direction, assigned by vector M in a semantic space with the specified conditions: projections of all estimation vectors onto this direction have to be proportional to these subjects ratings (desirability) for the given ranging conditions.

In a simplest case, if we are restricted to one view on the subject and two categories G2 and G3, then the studied situation can be illustrated for three subjects (1,3,4) with the following graph.

Fig. 1 Defining the Motivational Vector

and here are the second and third categories (organizing leader and sex-friend);   are estimation vectors corresponding to subjects 1.3.4; is a motivational vector (for the first ranging condition possibility of an intimate relationship); are the corresponding ratings of research subjects. As seen from Fig. 1, in spite of a greater rigidity (stability) degree of the first subject, the third subjects rating is greater due to better orientation in direction.

Had we studied the situation with goods marketing promotion, it would have been obvious that resources wasted on the third products advertisement would have been much smaller than those wasted on the first one. Assuming that is an angle between vector and vector .

Therefore: . So, the ratio of efficient to inefficient advertisement costs would be calculated from the correlation: . It is natural that launching of any commercial without preliminary testing will inevitably effect in waste of investment related to re-orientation of the product image.

In general, to calculate the motivational vector, it is necessary to take into account all significant states with the corresponding probabilities of the opinions (states) selected.

Proceeding from the first condition. The motivational vector has coordinates as indicated in Table 5.

 

Table 5. Motivational Vector

Категории	Semantic Coordinates	Feature Explicitness
G1	0,09	1,00
G2	0,27	1,00
G3	0,92	1,00
G4	-0,19	-1,00

Interestingly enough, to experience psychological intimacy with a woman, men prefer all the features to be maximally expressed (100% - see last column in Table 5). However, Semantic Analysis shows that only one criterion actually works it is a sex-friend (0.92).

To get a better understanding of this lets take a simple example. Let there be two balls, a cast-iron one and a cork one, flying at the same speed toward the wall. These objects are described in terms of one property - speed - that has various degrees of rigidity (stability) determined by their inertness or masses of the balls. Evidently enough, the most rigid object would have a greater contribution to the barrier destruction, while speed rigidity is greater in the first object.

Thus, a regular sociological survey that considers only expression of characteristics of research subjects is not capable of solving the problem as such, since an adequate object description is possible only in a semantic space.

Significance of each object state for intimacy-realization is represented in Table 6.

 

Table 6. Significance of subject states for realization of motive- intimacy for the third object

State 1	State 2	State 3
0,28	0,44	0,16

It is evident that the second state is possibly giving the basic weight here, however, due to its low rigidity and probability, real contributions to the intimacy rating for this object are the following (See Table 7).

 

Table 7. Weights of three subject states in realization of - intimacy motive for the third research object

Research Subjects	State 1	State 2	State 3	Total Rating
Photo 3	104,99	66,98	12,85	184,82

That means that the greatest weight to the subject rating for M1 is given by the first state (104.99).

Rating weights of the categories for the intimacy condition are demonstrated in the following table (8)

It is clear that men experience the greatest attraction to Subject 3 through category G3 sex-friend (264.17). Rating weights in all subject categories are represented in Graph 2.

Fig. 2

Motivation vector is analyzed likewise.  In this variant ranging of the same subject was performed on condition Possibility of a happy life with this woman.

 

Table 9. Motivational vector

Categories	Semantic Coordinates	Feature Explicitness
G1	0,59	0,95
G2	0,53	0,94
G3	0,07	0,32
G4	-0,57	-0,95

As evident from Table 9, men relate their happiness, first of all, to the first and the second categories: family partner (0.59) and organizing leader (0.53), and negatively connect it to the fourth category, which means, sex without responsibility, the sign changed to opposite. These are the three sources and three components of mens concept of happiness: a good wife, a good friend, sex with no consequence. It is interesting that efficiency, and considerateness negatively affect happiness expectation (they count as responsibilities). It probably creates an image of an older relative and produces a different attitude to a woman.

Rating weights on the second condition are shown in Graph 3.

Graph. 3

As seen from the above graph, Object 3 has a high estimate on this motive, primarily due to the first category of Family Partner.

Prediction accuracy for the subject rating on the second selection motive, four categories and three states is represented on Graph 4.

Graph. 4. Accuracy of Rating Prediction on the Second Condition

The line on Graph 4 indicates the meanings calculated from the semantic model, while the points correspond to empirical data.

Results of this project can be applied to selecting a womans image for commercials associated with family values, for example, apartment, interior etc.

Using these examples, we have demonstrated some direct results of Semantic Analysis. The methods potential is definitely much wider.

Back