Improving Recommender Systems by Situation Analysis using News Mining Techniques

Recommender systems are a subclass of information filtering system that seek to predict the rating or preference that a user would give to an item. One of the most widespread approaches to the design of Recommender Systems is Collaborative Filtering which is based on collecting and analyzing a large amount of information on user s behaviors, activities or preferences and predicting what users will like based on their similarity to other users or similarity of items. However, considering the influence of time, location and situation on user s preferences and noteworthiness are disregarded in this approach. Although multidimensional Recommender Systems have recently designed to compensate this defect in previous models, but it is the first attempt of using News Mining in general to consider environmental factors in estimation of user s ratings. News Mining refers to the process of deriving high quality information from news by crawling news web pages. The overarching goal in News Mining is, essentially, to turn news components (e.g. title, body, tags, images, date, etc.) into data for analysis, via application of natural language processing (NLP) and analytical methods. The increasing usage of online news emphasizes the importance of considering news effects in purchasing behavior of customers. Alongside historical and similarity patterns, different political, economic, health, etc. situations or certain events such as Olympic games or natural disasters will lead to different forms in customer reactions and demands. In this work, we proposed a novel method to predict user ratings of items by integrating previous collaborative model that focuses on item-similarity and News Mining techniques to take the influence of different situations —which are reflected in news, on purchasing manner of individuals into account. We have implied commonly used metrics- Root Mean Squared Error (RMSE), to evaluate performance of our method. The evaluations show this method will increase performance of Recommender systems

کلیدواژه ها:

Recommender Systems ، Collaborative Filtering ، News Mining