Final Poster

Movie Recommendation System based on Script Analysis,Cosine Similarity. NIRANJAN ,KARTHIK,RAHUL | RAJKUMAR R | SCOPE Introduction

Results

With the User based Collaborative filtering approach in which we will be looking for the most similar users for the current user in Item based collaborative filtering approach we will be using the items that are most similar to the current item for which we are going to predict the rating by using the item similarity weights and using the K most similar items and predicting the unknown rating. Then we will recommend the top N items having highest predicted rating as recommendations to the user.

Motivation Generally we hear quotes from people like: “like food in that restaurant ,you can try it”, “I saw this movie, you’ll like it” ,“Don’t go see that movie!” .So instead of asking other people for recommendation .we want to build a movie recommendation system System based on Content Filtering & Script Analysis.

SCOPE of the Project This system provides a personalized services to assist users in finding favourite items along with huge number of available online movies in the world wide web. We identify temporal preferences of an individual based on their interest like romance, horror,etc and provide personalization for users

Methodology The dataset that I’m working with is MovieLens, one of the most common datasets that is available on the internet for building a Recommender System. The version of the dataset that we aworking with (1M) contains 1,000,209 anonymous ratings of approximately 3,900 movies made by 6,040 MovieLens users who joined MovieLens in 2000. After processing the data and doing some exploratory analysis, here are the most interesting features of this dataset: Here’s a word-cloud visualization of the movie titles:

##

TP

FP

FN

TN precision

recall

TPR

## 10 2.438095 7.180952 66.63810 365.7429 0.2534653 0.03684398 0.03684398 ## 20 4.857143 14.380952 64.21905 358.5429 0.2524752 0.08070848 0.08070848 ## 30 6.952381 21.904762 62.12381 351.0190 0.2409241 0.11743856 0.11743856 ## 40 9.104762 29.371429 59.97143 343.5524 0.2366337 0.15409146 0.15409146 ## 50 11.152381 36.942857 57.92381 335.9810 0.2318812 0.19071582 0.19071582 ## 60 13.695238 43.923810 55.38095 329.0000 0.2374257 0.23003707 0.23003707 ##

FPR

## 10 0.01909268 ## 20 0.03845362 ## 30 0.05865530 ## 40 0.07869669 ## 50 0.09897217

Beautiful, isn’t it? I can recognize that there are a lot of movie franchises in this dataset, as evidenced by words like II and III… In addition to that, Day, Love, Life, Time, Night, Man, Dead, American are among the most commonly occuring words. Here’s a distribution of the user ratings:

## 60 0.11752674 In order to have a look at all the splits at the same time, I sum up the indices of columns TP, FP, FN and TN: ##

TP

FP

FN

TN

## 10 10.98095 28.16190 291.2571 1437.600 ## 20 22.20000 56.08571 280.0381 1409.676 ## 30 32.47619 84.94286 269.7619 1380.819 ## 40 41.93333 114.53333 260.3048 1351.229 ## 50 51.44762 144.00952 250.7905 1321.752 ## 60 61.05714 173.01905 241.1810 1292.743

Conclusion It appears that users are quite generous in their ratings. The mean rating is 3.58 on a scale of 5. Half the movies have a rating of 4 and 5. I personally think that a 5-level rating skill wasn’t a good indicator as people could have different rating styles (i.e. person A could always use 4 for an average movie, whereas person B only gives 4 out for their favorites). Each user rated at least 20 movies, so I doubt the distribution could be caused just by chance variance in the quality of movies.

In this project, we have developed and evaluated a collaborative filtering recommender (CFR) system for recommending movies. The online app was created to demonstrate the User-based Collaborative Filtering approach for recommendation model. Strengths: User-based Collaborative Filtering gives recommendations that can be complements to the item the user was interacting with. Weaknesses: User-based Collaborative Filtering is a type of Memory-based Collaborative Filtering that uses all user data in the database to create recommendations.

References 1. P. Resnick, H. R. Varian, "Recommender Systems", Communications of the ACM, vol. 40, pp. 5658, 1997. Show Context Access at ACM Google Scholar 2. H. Lieberman, "Autonomous Interface Agents", Proceedings of CHI'97 (Atlanta GA March 1997), pp. 67-74. Show Context Access at ACM Google Scholar