Research Article
Hybrid Recommendation Scheme Based on Deep Learning
Algorithm 1
Hybrid collaborative filtering, pretreatment.
| Input: User feature mount feature. product feature matrix feature, user rating matrix , user list and product list | | Output: Residual data set | (1) | Init Empty similarity matrix sim and sim //REviewed users and products similarity matrix; | (2) | Search the user list and the product List in ; | (3) | Construct. List and ListI into pairwise tuples set and | (4) | for in and i in set do | (5) | Compute sim and sim ; | (6) | sim add in sim , sim ; | (7) | Calculate all users similarity matrix Sim and all items similarity matrix sim | (8) | for in do | (9) | Define polarity score: | (10) | Search the K-clustering U and T-clustering-I in SimRU SimRI | (11) | Compute the rating of all users of uS on the product It//user collaboration; | (12) | User Uk’s rating on the products in the iS collection of all fields//item collaboration; | (13) | Query , construct and respectively; | (14) | Select relevant users and products are selected to construct a real score matrix | (15) | repeact//after multiple iterations, find the Wrong samplesby updating the learning r ate and weight coefficients | (16) | Package ; | (17) | Add data in Data; | (18) | until Get all data | (19) | final; | (20) | return Data; |
|