A Study On Information Recommendation System For Exploration And Production Portal Based On Big Data

With the rapid growth of PetroChina Exploration and Production Portal information release,it is difficult to find useful information quickly by traditional information search engines for users because of information overloaded.Accordingly,the study on information recommendation system based on big data is a good solution.According to the characteristics of the PetroChina Exploration and Production Portal information,the thesis proposes a portal information recommendation algorithm based on big data.In this algorithm,we firstly analyze the portal information by web crawler and Chinese segmentation,and then calculate feature words of each portal page by the TF-IDF model.Secondly,filter out the clustering label words through fuzzy clustering of feature words.Next,get user feature words by three MapReduce jobs on the user behavior log.The next job is to match user feature words with the portal page label words to generate a recommendation list.Simultaneously,another recommendation list has been generated by Item-based collaborative filtering algorithm.Finally,the final recommendations list would be output after filtering sorting and mixing the two groups of recommended listIn this thesis,the process and storage of large data sets is completed by Hadoop and other related technologies.And the information recommendation is realized by means of Mahout recommended engine.The recommendation system uses C # as the programming language on development design based on the.NET platform to complete the docking with the exploration and production portal.It presents recommendations through the well-designed user interface.Users can take appropriate action for these recommendations and the operation of user's behavior will be implicit feedback to the recommendation engine module to improve the efficiency of the algorithm.In addition,the user interface shows user's browsing records and the characteristic words as well,which can greatly improve the reliability of the system.Overall,this research contributes to the improvement of the intelligence level and the further application of the PetroChina Exploration and Production Portal.