Παιδιά μια μικρή βοήθεια

Συζητήσεις για την Java

Συντονιστές: WebDev Moderators, Super-Moderators

Απάντηση
Psycho287
Δημοσιεύσεις: 1
Εγγραφή: 14 Νοέμ 2007 13:19
Τοποθεσία: Kallithea

Παιδιά μια μικρή βοήθεια

Δημοσίευση από Psycho287 » 14 Νοέμ 2007 13:28

Γεια σας παιδιά!
Είμαι πολύ πιο νέος από ότι ένας αρχάριος!!!
Έχουμε έναν καθηγητή ο οποίος είναι λίγο τρελός!! Μας έβαλε μια εργασία η οποία είναι πολύ δύσκολη τουλάχιστον για μένα!!!
Μπορείτε να με βοηθήσετε?
Η εργασία λέει:
Recommender System (Content Based)
A recommender system is a system behind an e-shop that can suggest items that might be of interest to a customer; suggestion is based on the past purchasing behaviour of the customer. For instance, some one who purchases movies of a certain kind his future purchase is likely to be similar. Thus a recommender system could predict (within reason) the movies a user would like to see.
For example, if I favour movies who fall into the ‘musical’ and ‘comedy’ categories, chances are that future movies will fall into the same categories. We assume that there is a list of users (actually their ID), a list of movies (movie ID, movie title) and a list of movies seen by each customer along with a rating 1-5 (5 meaning absolutely loved it, 1 meaning I hated it).

Specification: Algorithmic
You are given the task of designing a movie recommender system for a user. That is given a movie that belongs in a category it will try to predict the rating. For instance,
consider the following movie representation (35 is the movie ID, "Mad Love" is the movie title and the rest of the 19 values (0/1) denote presence or absence of a movie characteristic
36|Mad Love |0|0|0|0|0|0|0|0|1|0|0|0|0|0|1|0|0|0
Thus, movie 36 has value 1 at positions 9 and 15 which correspond to science ‘fiction’ and ‘fantasy’ (see table below). The system will try to infer the rating of this film. In order to do this it will compare the vector of the 19 film characteristics with N closest films which the user has seen and rated. The majority vote of ratings will suggest the rating of the unknown film.

Specification: Graphical User Interface
In the graphical user interface, you should be able to enter the userID in one field, and in another field you should obtain the movie IDs. In addion the user would specify the maximum number of films for which he needs a recommendation.

Extra points (3%)
You can also print the movie Title along with the movie ID.

Extra points (2%)
The system could print the condidence of the recommendation. The confidence is based on how close the N closest films are to the user.

Extra points (1%)
The "userRatings" file "movies" file and "unseenMovies" will be read from the GUI.

UserRatings File
All ratings are contained in the file "userRatings.data" and are in the following format:
UserID MovieID Rating Timestamp
UserID is the identity number of the user, movieID is the ID of the movie, Rating is a number 1-5 (the higher the better), TimeStamp refers to the time the user made the recommendation.


Movies File
Movie information is in the file "movies.data" and is in the following
format:
MovieID Title ReleaseDate Site Genres
Titles are identical to titles provided by the IMDB (including year of release). Genres are pipe-separated and are selected from the following genres:
0. Unknown
1. Action
2. Adventure
3. Animation
4. Children's
5. Comedy
6. Crime
7. Documentary
8. Drama
9. Fantasy 10. Film-Noir
11. Horror
12. Musical
13. Mystery
14. Romance
15. Sci-Fi
16. Thriller
17. War
18. Western

Unseen Movies
The unseen Movies are the movies not seen by any user (and thus not rated), it has the same format as the movies file.

Απάντηση

Επιστροφή στο “Java”

Μέλη σε σύνδεση

Μέλη σε αυτήν τη Δ. Συζήτηση: Δεν υπάρχουν εγγεγραμμένα μέλη και 0 επισκέπτες