I have set of users and the topics they discuss as follows.
User, Topic, Number_of_times_discussed
Jone, movies, 3
Emi, food, 6
Jenne, movies, 1
Ej, tv series, 2
Micheal, tech, 8
Jone, comedy, 1
I also got another list that describes how interesting the topic was to the listeners. This interestingness score
is in the range of 0-1
. Note that some topics do not have interestingness score provided in the dataset (e.g., comedy
).
Topic, Interestingness_score
movies, 0.9
food, 0.1
tvseries, 0.6
tech, 0.8
Now, I want to have a weighted score for each connection that the users have with topics based on the two values I have. The first method I tried was as follows (see weighted_score
column).
User, Topic, Number_of_times_discussed, weighted_score
Jone, movies, 3, 3+(3*0.9)
Emi, food, 6, 6+(6*0.1)
Jenne, movies, 1, 1+(1*0.9)
Ej, tv series, 2 2+(2*0.6)
Micheal, tech, 8 8+(8*0.8)
Jone, comedy, 1 1+(1*0) <- since I do not have a interestingness_score for comedy
However, I would like to know if there are more statistical ways to induce the intrestingness_score
for each connection to weight them. I am happy to provide more details if needed.