kotsoc/lt2212-v19-a2

LT2212 V19 Assignment 2

LT2212 V19 Assignment 2

From Asad Sayeed's statistical NLP course at the University of Gothenburg.

My name: Konstantinos Peratinos

Additional instructions

Only meaningful additional information is that result.txt files contain two extra lines of information before every matrix. A line with the class of the folder and a line with the ammount of lines that the matrix takes. This was done in order to parse it succesfully with simdoc

File naming convention

Files were names resultX.txt where X is the question number

Results and discussion

Simdoc need a bit of finetuning to function for uneven matrixes but i ran out of time

Vocabulary restriction.

Well i did not think of it excessively just chose a shortened vocabulary of -B = 40

Result table

File names	Grain-Grain	Crude-Crude	Crude-Grain	Grain-Crude
result1.txt	-	-	-	- ## Could not calculate due to different dimensions of the classes
result2.txt	0.297	0.396	0.162	0.162
result3.txt	-	-	-	- ## Could not calculate due to different dimensions of the classese
result4.txt	0.226	0.248	0.171	0.171
result5.txt	0.214	0.270	0.056	0.052
result6.txt	-	-	-	- ## Could not calculate due to different dimensions of the classes
result7.txt	0.094	0.097	0.092	0.092
result8.txt	0.11	0.10	0.07	0.07 ## Could not calculate due to different dimensions of the classes truncating to 1000 did not make them uniform

The hypothesis in your own words

Discussion of trends in results in light of the hypothesis

Bonus answers

(Delete if you're not answering the bonus.)