چكيده به لاتين
Abstract:
In recent years, Wikipedia has grown substantially and users are involved with a large amount of
data, so it is better to use recommender systems to avoid user confusion for editing articles. The
Content Based approaches utilize a series of discrete characteristics of an item in order to recommend additional items with similar properties, while the Collaborative Filtering approaches predict
the interests of users by collaboratively learning from interests of related users. For the problem
of article recommendation in Wikipedia, we can use both approaches. For collaborative filtering,
we use matrix factorization method, and then with Content Based methods and finding similarity
between articles we improve item’s latent vector. for finding similar items we use Latent Dirichlet
Allocation to find the distribution of topics over articles and other features such as categories, then
compute similarity with shared characteristics between articles.
For evaluation, RMSE is used to compare the predicted rates with real rates and NDCG is used for
top-k recommedation evaluation. in both cases results have improved, comparing to the system base
on matrix factorization.
