//-----------------------
samples-cleaned.csv
//-----------------------
contact: gilles.bailly@telecom-paristech.fr

you can find the original paper here:
http://www.gillesbailly.fr/publis/BAILLY_MenuModel.pdf



//----------------------------
The useful columns are:
- independent variables:		user	block	trial	length	organization	item	 for the conditions
- dependent variables:		mouse_x		mouse_y		gazepointx		gazepointy		fixationx		fixationy

as well as the column time


//-----------------------------
The experiment was conducted on a windows PC and a 20 LCD display at 1280x960 resolution. 
A traditional optical mouse was used. 
All Participants used the same sensitivity and acceleration of the mouse.
We used an 1750 Tobii eye-tracker. 
The sampling data rate is 50Hz (raw eye movement data points every 20 ms); 
the latency was 20ms and the spatial resolution is 0.25 degrees (0.19cm on the screen) while the height of an item is 0.75 cm. 
The distance of the users’ eyes from the screen was 65cm. 
It means that participants can simultaneously focus on 3 items if we consider a fovea of 2 degrees.

useful additional information
- item width in pixels: 
- item height in pixels: 19
- separator item height: 9


-----------------------------

details for each column

//trial_id
a unique id for each trial

//date_ms
the date of the event in ms -> I suggest to use the column time instead.

//date_start_trial
the date when the trial starts

//date_stop_trial
the date when the trial stops (when the user selected an item)

//user
the user id

//block
the block id

//trial
trial id within block

//length
the length of the menu: 8 - 12 or 16

//organization
the organization of the menu: unordered - alphabetic - semantic

//item
the id of the target item (item that the user has to select)

//item_pos
the position of the item in the menu (first, second, etc.)
item_pos == item for the unordered and alphabetical organizations.
However, they are not always equal for the semantic organization.
Indeed, semantic organization also have "separator" items. In our data set, 
each logical groups contain 4 items. So, for a menu with 12 items:
item	item_pos
1		1
2		2
3		3
4		4
5		6
6		7
7		8
8		9
9		11
10		12
11		13
12		14

//item_name
the name of the target item

//mouse_x
the position of the mouse on the x axis
0 is the position of the top left corner of the menu

//mouse_y
the position of the mouse on the y axis
0 is the position of the top left corner of the menu

//gazepointx
the position of the gaze on the x axis
0 is the position of the top left corner of the menu

//gazepointy
the position of the gaze on the y axis
0 is the position of the top left corner of the menu

//fixationx
the position of the fixation on the x axis
0 is the position of the top left corner of the menu

//fixationy
the position of the fixation on the y axis
0 is the position of the top left corner of the menu

//validity_left
value given by tobii: validity for the left eye.

//validity_right
value given by tobii: validity for the right eye.

//validity_lr
value given by tobii: validity for the left and right eye.

//validity_accuracy
1 means that the corresponding trial has all gaze points.
I think that I removed all trials with missing gazes

//fixation_item
I do not know.

//date_loc_fix	date_loc_gaze
I implemented an algorithm to determine when the target item has been localized.
the algorithm does not work all the time but it seems that it works quite well

//time
timestamp; 0 is the beginning of the trial

//selection_time
it is the selection_time