文件列表:

__init__.py (0, 2018-07-07)
__pycache__/ (0, 2018-07-07)
__pycache__/__init__.cpython-35.pyc (200, 2018-07-07)
__pycache__/__init__.cpython-36.pyc (173, 2018-07-07)
__pycache__/collaborativeFilteringModel.cpython-35.pyc (1081, 2018-07-07)
__pycache__/collaborativeFilteringModel.cpython-36.pyc (970, 2018-07-07)
__pycache__/collaborativeRecommender.cpython-35.pyc (1805, 2018-07-07)
__pycache__/collaborativeRecommender.cpython-36.pyc (1669, 2018-07-07)
__pycache__/contentBasedFilteringModel.cpython-35.pyc (2689, 2018-07-07)
__pycache__/contentBasedFilteringModel.cpython-36.pyc (2467, 2018-07-07)
__pycache__/contentBasedRecommender.cpython-35.pyc (2839, 2018-07-07)
__pycache__/contentBasedRecommender.cpython-36.pyc (2657, 2018-07-07)
__pycache__/hybridModel.cpython-35.pyc (3173, 2018-07-07)
__pycache__/hybridModel.cpython-36.pyc (2885, 2018-07-07)
__pycache__/init.cpython-35.pyc (196, 2018-07-07)
__pycache__/init.cpython-36.pyc (5416, 2018-07-07)
__pycache__/popularityModel.cpython-35.pyc (513, 2018-07-07)
__pycache__/popularityModel.cpython-36.pyc (466, 2018-07-07)
__pycache__/popularityRecommender.cpython-35.pyc (1685, 2018-07-07)
__pycache__/popularityRecommender.cpython-36.pyc (1560, 2018-07-07)
collaborativeFilteringModel.py (1400, 2018-07-07)
collaborativeRecommender.py (1848, 2018-07-07)
contentBasedFilteringModel.py (2686, 2018-07-07)
contentBasedRecommender.py (2287, 2018-07-07)
hybridModel.py (4025, 2018-07-07)
init.py (8704, 2018-07-07)
popularityModel.py (305, 2018-07-07)
popularityRecommender.py (1558, 2018-07-07)

# Personalize-e-news-recommendation-system Personalize news recommendation system consists with main four components. News extraction and classification, news aggregation, summarization and recommendation. Out of these four components, my responsibility is to implement a proper method for the news recommendation component. I proposed hybrid news recommendation system consist of main four models. Popularity model, content-based filtering model, collaborative filtering model and location aware personalization model. Popularity model ranks the news articles according to their event type score. If user doesn’t have an account, this model helps to recommend news for the users. Content-based filtering model calculates the similarity between news article using BoW (bag of words) and order the articles according to their TF-IDF score. Collaborative filtering model consists of main two parts. User-based collaborative filtering and item-based collaborative filtering. User-based collaborative filtering means, finding the similarity between users and then do the recommendation. Item-based collaborative filtering means, finding the similarity between news articles and then do the recommendation. SVD (single value decomposition) is used as a Metrix factorization technique for finding the similarities between users’ and as well as similarities between news articles. When generating user profiles and tracking user interactions location information also stores in the database. Location aware personalization model gather all the news articles which are related to the user’s location details. Location information’s are calculated from user’s device IP address. For the evaluation, I used one of cross validation technique called holdout to split the data set as train data and test data. All the data before the current date take as train data set and all the current date data take as test data set. To calculate the accuracy of above models with our hybrid approach I used one of Top-N accuracy metric technique called Recall@N. according to the results, it proves my approach produces high accuracy than other techniques.

近期下载者：

相关文件：

评论：[我要评论] [举报此文件]

收藏者：