The first stage in the development of our project was the design of the psycholinguistic experiment.
In this phase, we first outlined the various parameters we wanted to test in our experiment. Before this phase, our plan for the project was a bit too ambitious. It was only after the experiment goals were outlined that we were able to really plan our work. We decided to study the effects of the following parameters on the positioning of the LO ( the location of the sweet spot ) :

1. the shape of the LO
2. the shape of the RO
3. the relative orientation of the LO
   
4. the size of the LO
5. the size of the RO
6. the size of the boundary/enclosure
7. using combinations of prepositions
8. using multiple ROs

Once the experiment outline was ready, we decided upon the structure of the experiment.
The experiment consisted of 56 tests. Each subject was to take the entire suite of tests, and the order of the tests would not vary between subjects. Also, the test had to be made user-friendly and as less boring as possible. We tried to meet these goals in our tests.

47 subjects volunteered to take the test.
(Click here to see a few sample tests from the entire suite.)

The data obtained from each subject was stored in a separate file in a prescribed format:
we stored the id of the test, the instruction to be followed, the input data file which was used to draw the test, and the final position in which the subject placed the LO. This data was stored for each of the 56 tests in the experiment.
(Click here to see a sample data file.)


ANALYSIS OF THE DATA and INTERPRETATION OF THE RESULTS

Once the data from all these 47 subjects had been obtained, we began our analysis of the data we had obtained.
The statistical tools we used for our calculations are :
ANOVA( analysis of variance ) and REGRESSION ANALYSIS.

After analysis of the data, we were able to distinguish the pesence of certain patterns in the input.
For example, we saw that the positioning of the LO is independent of the shape of the LO.
(Click here for an explanation of ANOVA for this example.)

We also concluded, among other things, that when asked to place a LO between 2 ROs humans position the LO almost exactly at the center of the space available between the 2 ROs. This conclusion was arrived at on the basis of linear regression.

INTEGRATION OF THE DATA OBTAINED WITH THE SPATIAL POSITIONING PROGRAM

The relations and independencies we had obtained from the above steps were now incorporated into the spatial positioning program.

The program first finds the location of the sweet spot depending on the results provided by the analysis of the data. ( this stage has not yet been completely implemented. At present, we have coded only a few of the inferences into the program ).

After the sweet spot has been calculated, the program generates potential fields for the preposition. This is done for each preposition for which positioning is done.

Now, the entire screen is scanned, and the point with the minimum potential is where the LO is placed for the choice of prepositions entered.
(Click here to see the output of the program).